[-- Attachment #1.1: Type: text/plain, Size: 579 bytes --] Hi, From the HoTT book, the truncation of any type A has two constructors: 1) for any a : A, there is |a| : ||A|| 2) for any x,y : ||A||, x = y. I get that if A is inhabited, then ||A|| is inhabited by (1). But is it true that, if ||A|| is inhabited, then A is inhabited? -- You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. For more options, visit https://groups.google.com/d/optout. [-- Attachment #1.2: Type: text/html, Size: 863 bytes --] <div dir="ltr"><div>Hi,</div><div><br></div><div>From the HoTT book, the truncation of any type A has two constructors:</div><div><br></div><div>1) for any a : A, there is |a| : ||A||</div><div>2) for any x,y : ||A||, x = y. </div><div><br></div><div>I get that if A is inhabited, then ||A|| is inhabited by (1). But is it true that, if ||A|| is inhabited, then A is inhabited? </div></div> <p></p> -- <br /> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group.<br /> To unsubscribe from this group and stop receiving emails from it, send an email to <a href="mailto:HomotopyTypeTheory+unsubscribe@googlegroups.com">HomotopyTypeTheory+unsubscribe@googlegroups.com</a>.<br /> For more options, visit <a href="https://groups.google.com/d/optout">https://groups.google.com/d/optout</a>.<br />

[-- Attachment #1: Type: text/plain, Size: 1116 bytes --] You can't have a function which, for all A, gives you ||A|| -> A. See the exercises 3.11 and 3.12! -- Nicolai On 05/03/19 22:31, Jean Joseph wrote: > Hi, > > From the HoTT book, the truncation of any type A has two constructors: > > 1) for any a : A, there is |a| : ||A|| > 2) for any x,y : ||A||, x = y. > > I get that if A is inhabited, then ||A|| is inhabited by (1). But is > it true that, if ||A|| is inhabited, then A is inhabited? > -- > You received this message because you are subscribed to the Google > Groups "Homotopy Type Theory" group. > To unsubscribe from this group and stop receiving emails from it, send > an email to HomotopyTypeTheory+unsubscribe@googlegroups.com > <mailto:HomotopyTypeTheory+unsubscribe@googlegroups.com>. > For more options, visit https://groups.google.com/d/optout. -- You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. For more options, visit https://groups.google.com/d/optout. [-- Attachment #2: Type: text/html, Size: 2099 bytes --] <html> <head> <meta http-equiv="Content-Type" content="text/html; charset=utf-8"> </head> <body text="#000000" bgcolor="#FFFFFF"> You can't have a function which, for all A, gives you ||A|| -> A. See the exercises 3.11 and 3.12!<br> -- Nicolai<br> <br> <div class="moz-cite-prefix">On 05/03/19 22:31, Jean Joseph wrote:<br> </div> <blockquote type="cite" cite="mid:0f5b8d0e-9f1d-47a7-9d39-a9112afb77ea@googlegroups.com"> <div dir="ltr"> <div>Hi,</div> <div><br> </div> <div>From the HoTT book, the truncation of any type A has two constructors:</div> <div><br> </div> <div>1) for any a : A, there is |a| : ||A||</div> <div>2) for any x,y : ||A||, x = y. </div> <div><br> </div> <div>I get that if A is inhabited, then ||A|| is inhabited by (1). But is it true that, if ||A|| is inhabited, then A is inhabited? </div> </div> -- <br> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group.<br> To unsubscribe from this group and stop receiving emails from it, send an email to <a href="mailto:HomotopyTypeTheory+unsubscribe@googlegroups.com" moz-do-not-send="true">HomotopyTypeTheory+unsubscribe@googlegroups.com</a>.<br> For more options, visit <a href="https://groups.google.com/d/optout" moz-do-not-send="true">https://groups.google.com/d/optout</a>.<br> </blockquote> <br> </body> </html> <p></p> -- <br /> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group.<br /> To unsubscribe from this group and stop receiving emails from it, send an email to <a href="mailto:HomotopyTypeTheory+unsubscribe@googlegroups.com">HomotopyTypeTheory+unsubscribe@googlegroups.com</a>.<br /> For more options, visit <a href="https://groups.google.com/d/optout">https://groups.google.com/d/optout</a>.<br />

[-- Attachment #1.1: Type: text/plain, Size: 1868 bytes --] Or you can read the paper https://lmcs.episciences.org/3217/ regarding what Nicolai said. Moreover, in the HoTT book, it is shown that if || X||->X holds for all X, then univalence can't hold. (It is global choice, which can't be invariant under equivalence.) The above paper shows that unrestricted ||X||->X it gives excluded middle. However, for a lot of kinds of types one can show that ||X||->X does hold. For example, if they have a constant endo-function. Moreover, for any type X, the availability of ||X||->X is logically equivalent to the availability of a constant map X->X (before we know whether X has a point or not, in which case the availability of a constant endo-map is trivial). Martin On Tuesday, 5 March 2019 22:47:55 UTC, Nicolai Kraus wrote: > > You can't have a function which, for all A, gives you ||A|| -> A. See the > exercises 3.11 and 3.12! > -- Nicolai > > On 05/03/19 22:31, Jean Joseph wrote: > > Hi, > > From the HoTT book, the truncation of any type A has two constructors: > > 1) for any a : A, there is |a| : ||A|| > 2) for any x,y : ||A||, x = y. > > I get that if A is inhabited, then ||A|| is inhabited by (1). But is it > true that, if ||A|| is inhabited, then A is inhabited? > -- > You received this message because you are subscribed to the Google Groups > "Homotopy Type Theory" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to HomotopyTypeTheory+unsubscribe@googlegroups.com <javascript:>. > For more options, visit https://groups.google.com/d/optout. > > > -- You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. For more options, visit https://groups.google.com/d/optout. [-- Attachment #1.2: Type: text/html, Size: 3238 bytes --] <div dir="ltr">Or you can read the paper https://lmcs.episciences.org/3217/ regarding what Nicolai said.<div><br></div><div>Moreover, in the HoTT book, it is shown that if || X||->X holds for all X, then univalence can't hold. (It is global choice, which can't be invariant under equivalence.)</div><div><br></div><div>The above paper shows that unrestricted ||X||->X it gives excluded middle. </div><div><br></div><div>However, for a lot of kinds of types one can show that ||X||->X does hold. For example, if they have a constant endo-function. Moreover, for any type X, the availability of ||X||->X is logically equivalent to the availability of a constant map X->X (before we know whether X has a point or not, in which case the availability of a constant endo-map is trivial).</div><div><br></div><div>Martin<br><br>On Tuesday, 5 March 2019 22:47:55 UTC, Nicolai Kraus wrote:<blockquote class="gmail_quote" style="margin: 0;margin-left: 0.8ex;border-left: 1px #ccc solid;padding-left: 1ex;"> <div text="#000000" bgcolor="#FFFFFF"> You can't have a function which, for all A, gives you ||A|| -> A. See the exercises 3.11 and 3.12!<br> -- Nicolai<br> <br> <div>On 05/03/19 22:31, Jean Joseph wrote:<br> </div> <blockquote type="cite"> <div dir="ltr"> <div>Hi,</div> <div><br> </div> <div>From the HoTT book, the truncation of any type A has two constructors:</div> <div><br> </div> <div>1) for any a : A, there is |a| : ||A||</div> <div>2) for any x,y : ||A||, x = y. </div> <div><br> </div> <div>I get that if A is inhabited, then ||A|| is inhabited by (1). But is it true that, if ||A|| is inhabited, then A is inhabited? </div> </div> -- <br> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group.<br> To unsubscribe from this group and stop receiving emails from it, send an email to <a href="javascript:" target="_blank" gdf-obfuscated-mailto="gFXtP5RWAQAJ" rel="nofollow" onmousedown="this.href='javascript:';return true;" onclick="this.href='javascript:';return true;">HomotopyTypeTheory+<wbr>unsubscribe@googlegroups.com</a>.<br> For more options, visit <a href="https://groups.google.com/d/optout" target="_blank" rel="nofollow" onmousedown="this.href='https://groups.google.com/d/optout';return true;" onclick="this.href='https://groups.google.com/d/optout';return true;">https://groups.google.com/d/<wbr>optout</a>.<br> </blockquote> <br> </div> </blockquote></div></div> <p></p> -- <br /> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group.<br /> To unsubscribe from this group and stop receiving emails from it, send an email to <a href="mailto:HomotopyTypeTheory+unsubscribe@googlegroups.com">HomotopyTypeTheory+unsubscribe@googlegroups.com</a>.<br /> For more options, visit <a href="https://groups.google.com/d/optout">https://groups.google.com/d/optout</a>.<br />

Just in case anyone reading this thread later is confused about a more beginner point than the ones Nicolai and Martin made, one possible stumbling block here is that, if someone means “is inhabited” in an external sense (there is a closed term of that type), then the answer is yes (at least in some models): if ||A|| is inhabited then A is inhabited. For example, in cubical models with canonicity, it is true that a closed term of type ||A|| evaluates to a value that has as a subterm a closed term of type A (the other values of ||A|| are some “formal compositions” of values of ||A||, but there has to be an |a| in there at the base case). This is consistent with what Martin and Nicolai said because “if A is inhabited then B is inhabited” (in this external sense) doesn’t necessarily mean there is a map A -> B internally. -Dan > On Mar 5, 2019, at 6:07 PM, Martín Hötzel Escardó <escardo.martin@gmail.com> wrote: > > Or you can read the paper https://lmcs.episciences.org/3217/ regarding what Nicolai said. > > Moreover, in the HoTT book, it is shown that if || X||->X holds for all X, then univalence can't hold. (It is global choice, which can't be invariant under equivalence.) > > The above paper shows that unrestricted ||X||->X it gives excluded middle. > > However, for a lot of kinds of types one can show that ||X||->X does hold. For example, if they have a constant endo-function. Moreover, for any type X, the availability of ||X||->X is logically equivalent to the availability of a constant map X->X (before we know whether X has a point or not, in which case the availability of a constant endo-map is trivial). > > Martin > > On Tuesday, 5 March 2019 22:47:55 UTC, Nicolai Kraus wrote: > You can't have a function which, for all A, gives you ||A|| -> A. See the exercises 3.11 and 3.12! > -- Nicolai > > On 05/03/19 22:31, Jean Joseph wrote: >> Hi, >> >> From the HoTT book, the truncation of any type A has two constructors: >> >> 1) for any a : A, there is |a| : ||A|| >> 2) for any x,y : ||A||, x = y. >> >> I get that if A is inhabited, then ||A|| is inhabited by (1). But is it true that, if ||A|| is inhabited, then A is inhabited? >> -- >> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. >> To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. >> For more options, visit https://groups.google.com/d/optout. > > > -- > You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. > To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. > For more options, visit https://groups.google.com/d/optout. -- You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. For more options, visit https://groups.google.com/d/optout.

[-- Attachment #1.1: Type: text/plain, Size: 4140 bytes --] I got confused now. :-) Seriously now, what you say seems related to the fact that from a proof |- t : || X || in the empty context, you get |- x : X in cubical type theory. This follows from Simon's canonicity result (at least for X=natural numbers), and is like the so-called "existence property" in the internal language of the free elementary topos. This says that from a proof |- exists (x:X), A x in the empty context, you get |- x : X and |- A x. This says that exists in the empty context behaves like Sigma. But only in the empty context, because otherwise it behaves like "local existence" as in Kripke-Joyal semantics. Martin On Thursday, 7 March 2019 14:10:56 UTC, dlicata wrote: > > Just in case anyone reading this thread later is confused about a more > beginner point than the ones Nicolai and Martin made, one possible > stumbling block here is that, if someone means “is inhabited” in an > external sense (there is a closed term of that type), then the answer is > yes (at least in some models): if ||A|| is inhabited then A is inhabited. > For example, in cubical models with canonicity, it is true that a closed > term of type ||A|| evaluates to a value that has as a subterm a closed term > of type A (the other values of ||A|| are some “formal compositions” of > values of ||A||, but there has to be an |a| in there at the base case). > This is consistent with what Martin and Nicolai said because “if A is > inhabited then B is inhabited” (in this external sense) doesn’t necessarily > mean there is a map A -> B internally. > > -Dan > > > On Mar 5, 2019, at 6:07 PM, Martín Hötzel Escardó <escardo...@gmail.com > <javascript:>> wrote: > > > > Or you can read the paper https://lmcs.episciences.org/3217/ regarding > what Nicolai said. > > > > Moreover, in the HoTT book, it is shown that if || X||->X holds for all > X, then univalence can't hold. (It is global choice, which can't be > invariant under equivalence.) > > > > The above paper shows that unrestricted ||X||->X it gives excluded > middle. > > > > However, for a lot of kinds of types one can show that ||X||->X does > hold. For example, if they have a constant endo-function. Moreover, for any > type X, the availability of ||X||->X is logically equivalent to the > availability of a constant map X->X (before we know whether X has a point > or not, in which case the availability of a constant endo-map is trivial). > > > > Martin > > > > On Tuesday, 5 March 2019 22:47:55 UTC, Nicolai Kraus wrote: > > You can't have a function which, for all A, gives you ||A|| -> A. See > the exercises 3.11 and 3.12! > > -- Nicolai > > > > On 05/03/19 22:31, Jean Joseph wrote: > >> Hi, > >> > >> From the HoTT book, the truncation of any type A has two constructors: > >> > >> 1) for any a : A, there is |a| : ||A|| > >> 2) for any x,y : ||A||, x = y. > >> > >> I get that if A is inhabited, then ||A|| is inhabited by (1). But is it > true that, if ||A|| is inhabited, then A is inhabited? > >> -- > >> You received this message because you are subscribed to the Google > Groups "Homotopy Type Theory" group. > >> To unsubscribe from this group and stop receiving emails from it, send > an email to HomotopyTypeTheory+unsubscribe@googlegroups.com <javascript:>. > > >> For more options, visit https://groups.google.com/d/optout. > > > > > > -- > > You received this message because you are subscribed to the Google > Groups "Homotopy Type Theory" group. > > To unsubscribe from this group and stop receiving emails from it, send > an email to HomotopyTypeTheory+unsubscribe@googlegroups.com <javascript:>. > > > For more options, visit https://groups.google.com/d/optout. > > -- You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. For more options, visit https://groups.google.com/d/optout. [-- Attachment #1.2: Type: text/html, Size: 6217 bytes --] <div dir="ltr">I got confused now. :-)<div><br></div><div>Seriously now, what you say seems related to the fact that from a proof |- t : || X || in the empty context, you get |- x : X in cubical type theory. This follows from Simon's canonicity result (at least for X=natural numbers), and is like the so-called "existence property" in the internal language of the free elementary topos. This says that from a proof |- exists (x:X), A x in the empty context, you get |- x : X and |- A x. This says that exists in the empty context behaves like Sigma. But only in the empty context, because otherwise it behaves like "local existence" as in Kripke-Joyal semantics. </div><div><br></div><div>Martin</div><div><br>On Thursday, 7 March 2019 14:10:56 UTC, dlicata wrote:<blockquote class="gmail_quote" style="margin: 0;margin-left: 0.8ex;border-left: 1px #ccc solid;padding-left: 1ex;">Just in case anyone reading this thread later is confused about a more beginner point than the ones Nicolai and Martin made, one possible stumbling block here is that, if someone means “is inhabited” in an external sense (there is a closed term of that type), then the answer is yes (at least in some models): if ||A|| is inhabited then A is inhabited. For example, in cubical models with canonicity, it is true that a closed term of type ||A|| evaluates to a value that has as a subterm a closed term of type A (the other values of ||A|| are some “formal compositions” of values of ||A||, but there has to be an |a| in there at the base case). This is consistent with what Martin and Nicolai said because “if A is inhabited then B is inhabited” (in this external sense) doesn’t necessarily mean there is a map A -> B internally. <br> <br>-Dan <br> <br>> On Mar 5, 2019, at 6:07 PM, Martín Hötzel Escardó <<a href="javascript:" target="_blank" gdf-obfuscated-mailto="9zSP2TkZAwAJ" rel="nofollow" onmousedown="this.href='javascript:';return true;" onclick="this.href='javascript:';return true;">escardo...@gmail.com</a>> wrote: <br>> <br>> Or you can read the paper <a href="https://lmcs.episciences.org/3217/" target="_blank" rel="nofollow" onmousedown="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Flmcs.episciences.org%2F3217%2F\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNF2sTOfc7ovV50gWLoaFL_7S1LECg';return true;" onclick="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Flmcs.episciences.org%2F3217%2F\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNF2sTOfc7ovV50gWLoaFL_7S1LECg';return true;">https://lmcs.episciences.org/<wbr>3217/</a> regarding what Nicolai said. <br>> <br>> Moreover, in the HoTT book, it is shown that if || X||->X holds for all X, then univalence can't hold. (It is global choice, which can't be invariant under equivalence.) <br>> <br>> The above paper shows that unrestricted ||X||->X it gives excluded middle. <br>> <br>> However, for a lot of kinds of types one can show that ||X||->X does hold. For example, if they have a constant endo-function. Moreover, for any type X, the availability of ||X||->X is logically equivalent to the availability of a constant map X->X (before we know whether X has a point or not, in which case the availability of a constant endo-map is trivial). <br>> <br>> Martin <br>> <br>> On Tuesday, 5 March 2019 22:47:55 UTC, Nicolai Kraus wrote: <br>> You can't have a function which, for all A, gives you ||A|| -> A. See the exercises 3.11 and 3.12! <br>> -- Nicolai <br>> <br>> On 05/03/19 22:31, Jean Joseph wrote: <br>>> Hi, <br>>> <br>>> From the HoTT book, the truncation of any type A has two constructors: <br>>> <br>>> 1) for any a : A, there is |a| : ||A|| <br>>> 2) for any x,y : ||A||, x = y. <br>>> <br>>> I get that if A is inhabited, then ||A|| is inhabited by (1). But is it true that, if ||A|| is inhabited, then A is inhabited? <br>>> -- <br>>> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. <br>>> To unsubscribe from this group and stop receiving emails from it, send an email to <a href="javascript:" target="_blank" gdf-obfuscated-mailto="9zSP2TkZAwAJ" rel="nofollow" onmousedown="this.href='javascript:';return true;" onclick="this.href='javascript:';return true;">HomotopyTypeTheory+<wbr>unsubscribe@googlegroups.com</a>. <br>>> For more options, visit <a href="https://groups.google.com/d/optout" target="_blank" rel="nofollow" onmousedown="this.href='https://groups.google.com/d/optout';return true;" onclick="this.href='https://groups.google.com/d/optout';return true;">https://groups.google.com/d/<wbr>optout</a>. <br>> <br>> <br>> -- <br>> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. <br>> To unsubscribe from this group and stop receiving emails from it, send an email to <a href="javascript:" target="_blank" gdf-obfuscated-mailto="9zSP2TkZAwAJ" rel="nofollow" onmousedown="this.href='javascript:';return true;" onclick="this.href='javascript:';return true;">HomotopyTypeTheory+<wbr>unsubscribe@googlegroups.com</a>. <br>> For more options, visit <a href="https://groups.google.com/d/optout" target="_blank" rel="nofollow" onmousedown="this.href='https://groups.google.com/d/optout';return true;" onclick="this.href='https://groups.google.com/d/optout';return true;">https://groups.google.com/d/<wbr>optout</a>. <br> <br></blockquote></div></div> <p></p> -- <br /> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group.<br /> To unsubscribe from this group and stop receiving emails from it, send an email to <a href="mailto:HomotopyTypeTheory+unsubscribe@googlegroups.com">HomotopyTypeTheory+unsubscribe@googlegroups.com</a>.<br /> For more options, visit <a href="https://groups.google.com/d/optout">https://groups.google.com/d/optout</a>.<br />

[-- Attachment #1: Type: text/plain, Size: 5249 bytes --] On a similar note, I had originally (and mistakenly, at least according to the terminology in the HoTT book, which I looked up) thought that declaring “A is inhabited” rather than just “A” was meant to refer to the type || A || rather than A, in which case, I read the question “if || A || is inhabited, then A is inhabited” as || || A || || -> || A || The above statement does hold, and in some way I think of this (though I’m not sure how to connect it formally) as an internalization of Dan’s and Martin’s external statements. (According to the HoTT book, we say “A is merely inhabited” to say || A ||.) > On Mar 7, 2019, at 11:16 AM, Martín Hötzel Escardó <escardo.martin@gmail.com> wrote: > > I got confused now. :-) > > Seriously now, what you say seems related to the fact that from a proof |- t : || X || in the empty context, you get |- x : X in cubical type theory. This follows from Simon's canonicity result (at least for X=natural numbers), and is like the so-called "existence property" in the internal language of the free elementary topos. This says that from a proof |- exists (x:X), A x in the empty context, you get |- x : X and |- A x. This says that exists in the empty context behaves like Sigma. But only in the empty context, because otherwise it behaves like "local existence" as in Kripke-Joyal semantics. > > Martin > > On Thursday, 7 March 2019 14:10:56 UTC, dlicata wrote: > Just in case anyone reading this thread later is confused about a more beginner point than the ones Nicolai and Martin made, one possible stumbling block here is that, if someone means “is inhabited” in an external sense (there is a closed term of that type), then the answer is yes (at least in some models): if ||A|| is inhabited then A is inhabited. For example, in cubical models with canonicity, it is true that a closed term of type ||A|| evaluates to a value that has as a subterm a closed term of type A (the other values of ||A|| are some “formal compositions” of values of ||A||, but there has to be an |a| in there at the base case). This is consistent with what Martin and Nicolai said because “if A is inhabited then B is inhabited” (in this external sense) doesn’t necessarily mean there is a map A -> B internally. > > -Dan > > > On Mar 5, 2019, at 6:07 PM, Martín Hötzel Escardó <escardo...@gmail.com <>> wrote: > > > > Or you can read the paper https://lmcs.episciences.org/3217/ <https://lmcs.episciences.org/3217/> regarding what Nicolai said. > > > > Moreover, in the HoTT book, it is shown that if || X||->X holds for all X, then univalence can't hold. (It is global choice, which can't be invariant under equivalence.) > > > > The above paper shows that unrestricted ||X||->X it gives excluded middle. > > > > However, for a lot of kinds of types one can show that ||X||->X does hold. For example, if they have a constant endo-function. Moreover, for any type X, the availability of ||X||->X is logically equivalent to the availability of a constant map X->X (before we know whether X has a point or not, in which case the availability of a constant endo-map is trivial). > > > > Martin > > > > On Tuesday, 5 March 2019 22:47:55 UTC, Nicolai Kraus wrote: > > You can't have a function which, for all A, gives you ||A|| -> A. See the exercises 3.11 and 3.12! > > -- Nicolai > > > > On 05/03/19 22:31, Jean Joseph wrote: > >> Hi, > >> > >> From the HoTT book, the truncation of any type A has two constructors: > >> > >> 1) for any a : A, there is |a| : ||A|| > >> 2) for any x,y : ||A||, x = y. > >> > >> I get that if A is inhabited, then ||A|| is inhabited by (1). But is it true that, if ||A|| is inhabited, then A is inhabited? > >> -- > >> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. > >> To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com <>. > >> For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>. > > > > > > -- > > You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. > > To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com <>. > > For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>. > > > -- > You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. > To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com <mailto:HomotopyTypeTheory+unsubscribe@googlegroups.com>. > For more options, visit https://groups.google.com/d/optout <https://groups.google.com/d/optout>. -- You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. For more options, visit https://groups.google.com/d/optout. [-- Attachment #2: Type: text/html, Size: 7441 bytes --] <html><head><meta http-equiv="Content-Type" content="text/html; charset=utf-8"></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; line-break: after-white-space;" class=""><div>On a similar note, I had originally (and mistakenly, at least according to the terminology in the HoTT book, which I looked up) thought that declaring “A is inhabited” rather than just “A” was meant to refer to the type || A || rather than A, in which case, I read the question “if || A || is inhabited, then A is inhabited” as</div><div><br class=""></div><div>|| || A || || -> || A ||</div><div><br class=""></div><div>The above statement does hold, and in some way I think of this (though I’m not sure how to connect it formally) as an internalization of Dan’s and Martin’s external statements.</div><div><br class=""></div><div>(According to the HoTT book, we say “A is <i class="">merely</i> inhabited” to say || A ||.)</div><div><br class=""><blockquote type="cite" class=""><div class="">On Mar 7, 2019, at 11:16 AM, Martín Hötzel Escardó <<a href="mailto:escardo.martin@gmail.com" class="">escardo.martin@gmail.com</a>> wrote:</div><br class="Apple-interchange-newline"><div class=""><div dir="ltr" class="">I got confused now. :-)<div class=""><br class=""></div><div class="">Seriously now, what you say seems related to the fact that from a proof |- t : || X || in the empty context, you get |- x : X in cubical type theory. This follows from Simon's canonicity result (at least for X=natural numbers), and is like the so-called "existence property" in the internal language of the free elementary topos. This says that from a proof |- exists (x:X), A x in the empty context, you get |- x : X and |- A x. This says that exists in the empty context behaves like Sigma. But only in the empty context, because otherwise it behaves like "local existence" as in Kripke-Joyal semantics. </div><div class=""><br class=""></div><div class="">Martin</div><div class=""><br class="">On Thursday, 7 March 2019 14:10:56 UTC, dlicata wrote:<blockquote class="gmail_quote" style="margin: 0;margin-left: 0.8ex;border-left: 1px #ccc solid;padding-left: 1ex;">Just in case anyone reading this thread later is confused about a more beginner point than the ones Nicolai and Martin made, one possible stumbling block here is that, if someone means “is inhabited” in an external sense (there is a closed term of that type), then the answer is yes (at least in some models): if ||A|| is inhabited then A is inhabited. For example, in cubical models with canonicity, it is true that a closed term of type ||A|| evaluates to a value that has as a subterm a closed term of type A (the other values of ||A|| are some “formal compositions” of values of ||A||, but there has to be an |a| in there at the base case). This is consistent with what Martin and Nicolai said because “if A is inhabited then B is inhabited” (in this external sense) doesn’t necessarily mean there is a map A -> B internally. <br class=""> <br class="">-Dan <br class=""> <br class="">> On Mar 5, 2019, at 6:07 PM, Martín Hötzel Escardó <<a target="_blank" gdf-obfuscated-mailto="9zSP2TkZAwAJ" rel="nofollow" class="">escardo...@gmail.com</a>> wrote: <br class="">> <br class="">> Or you can read the paper <a href="https://lmcs.episciences.org/3217/" target="_blank" rel="nofollow" class="">https://lmcs.episciences.org/<wbr class="">3217/</a> regarding what Nicolai said. <br class="">> <br class="">> Moreover, in the HoTT book, it is shown that if || X||->X holds for all X, then univalence can't hold. (It is global choice, which can't be invariant under equivalence.) <br class="">> <br class="">> The above paper shows that unrestricted ||X||->X it gives excluded middle. <br class="">> <br class="">> However, for a lot of kinds of types one can show that ||X||->X does hold. For example, if they have a constant endo-function. Moreover, for any type X, the availability of ||X||->X is logically equivalent to the availability of a constant map X->X (before we know whether X has a point or not, in which case the availability of a constant endo-map is trivial). <br class="">> <br class="">> Martin <br class="">> <br class="">> On Tuesday, 5 March 2019 22:47:55 UTC, Nicolai Kraus wrote: <br class="">> You can't have a function which, for all A, gives you ||A|| -> A. See the exercises 3.11 and 3.12! <br class="">> -- Nicolai <br class="">> <br class="">> On 05/03/19 22:31, Jean Joseph wrote: <br class="">>> Hi, <br class="">>> <br class="">>> From the HoTT book, the truncation of any type A has two constructors: <br class="">>> <br class="">>> 1) for any a : A, there is |a| : ||A|| <br class="">>> 2) for any x,y : ||A||, x = y. <br class="">>> <br class="">>> I get that if A is inhabited, then ||A|| is inhabited by (1). But is it true that, if ||A|| is inhabited, then A is inhabited? <br class="">>> -- <br class="">>> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. <br class="">>> To unsubscribe from this group and stop receiving emails from it, send an email to <a target="_blank" gdf-obfuscated-mailto="9zSP2TkZAwAJ" rel="nofollow" class="">HomotopyTypeTheory+<wbr class="">unsubscribe@googlegroups.com</a>. <br class="">>> For more options, visit <a href="https://groups.google.com/d/optout" target="_blank" rel="nofollow" class="">https://groups.google.com/d/<wbr class="">optout</a>. <br class="">> <br class="">> <br class="">> -- <br class="">> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. <br class="">> To unsubscribe from this group and stop receiving emails from it, send an email to <a target="_blank" gdf-obfuscated-mailto="9zSP2TkZAwAJ" rel="nofollow" class="">HomotopyTypeTheory+<wbr class="">unsubscribe@googlegroups.com</a>. <br class="">> For more options, visit <a href="https://groups.google.com/d/optout" target="_blank" rel="nofollow" class="">https://groups.google.com/d/<wbr class="">optout</a>. <br class=""> <br class=""></blockquote></div></div><div class=""><br class="webkit-block-placeholder"></div> -- <br class=""> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group.<br class=""> To unsubscribe from this group and stop receiving emails from it, send an email to <a href="mailto:HomotopyTypeTheory+unsubscribe@googlegroups.com" class="">HomotopyTypeTheory+unsubscribe@googlegroups.com</a>.<br class=""> For more options, visit <a href="https://groups.google.com/d/optout" class="">https://groups.google.com/d/optout</a>.<br class=""> </div></blockquote></div><br class=""></body></html> <p></p> -- <br /> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group.<br /> To unsubscribe from this group and stop receiving emails from it, send an email to <a href="mailto:HomotopyTypeTheory+unsubscribe@googlegroups.com">HomotopyTypeTheory+unsubscribe@googlegroups.com</a>.<br /> For more options, visit <a href="https://groups.google.com/d/optout">https://groups.google.com/d/optout</a>.<br />

[-- Attachment #1.1: Type: text/plain, Size: 4736 bytes --] The existence property is proved for CCHM cubicaltt by Simon in: https://arxiv.org/abs/1607.04156 See corollary 5.2. This works a bit more generally than what Martín said, in particular in any context with only dimension variables we can compute a witness to an existence. So if in context G = i_1 : II, ..., i_n : II (possibly empty) we have: G |- t : exists (x : X), A(x) then we can compute G |- u : X so that G |- B(u). -- Anders On Thursday, March 7, 2019 at 11:16:48 AM UTC-5, Martín Hötzel Escardó wrote: > > I got confused now. :-) > > Seriously now, what you say seems related to the fact that from a proof |- > t : || X || in the empty context, you get |- x : X in cubical type theory. > This follows from Simon's canonicity result (at least for X=natural > numbers), and is like the so-called "existence property" in the internal > language of the free elementary topos. This says that from a proof |- > exists (x:X), A x in the empty context, you get |- x : X and |- A x. This > says that exists in the empty context behaves like Sigma. But only in the > empty context, because otherwise it behaves like "local existence" as in > Kripke-Joyal semantics. > > Martin > > On Thursday, 7 March 2019 14:10:56 UTC, dlicata wrote: >> >> Just in case anyone reading this thread later is confused about a more >> beginner point than the ones Nicolai and Martin made, one possible >> stumbling block here is that, if someone means “is inhabited” in an >> external sense (there is a closed term of that type), then the answer is >> yes (at least in some models): if ||A|| is inhabited then A is inhabited. >> For example, in cubical models with canonicity, it is true that a closed >> term of type ||A|| evaluates to a value that has as a subterm a closed term >> of type A (the other values of ||A|| are some “formal compositions” of >> values of ||A||, but there has to be an |a| in there at the base case). >> This is consistent with what Martin and Nicolai said because “if A is >> inhabited then B is inhabited” (in this external sense) doesn’t necessarily >> mean there is a map A -> B internally. >> >> -Dan >> >> > On Mar 5, 2019, at 6:07 PM, Martín Hötzel Escardó <escardo...@gmail.com> >> wrote: >> > >> > Or you can read the paper https://lmcs.episciences.org/3217/ regarding >> what Nicolai said. >> > >> > Moreover, in the HoTT book, it is shown that if || X||->X holds for all >> X, then univalence can't hold. (It is global choice, which can't be >> invariant under equivalence.) >> > >> > The above paper shows that unrestricted ||X||->X it gives excluded >> middle. >> > >> > However, for a lot of kinds of types one can show that ||X||->X does >> hold. For example, if they have a constant endo-function. Moreover, for any >> type X, the availability of ||X||->X is logically equivalent to the >> availability of a constant map X->X (before we know whether X has a point >> or not, in which case the availability of a constant endo-map is trivial). >> > >> > Martin >> > >> > On Tuesday, 5 March 2019 22:47:55 UTC, Nicolai Kraus wrote: >> > You can't have a function which, for all A, gives you ||A|| -> A. See >> the exercises 3.11 and 3.12! >> > -- Nicolai >> > >> > On 05/03/19 22:31, Jean Joseph wrote: >> >> Hi, >> >> >> >> From the HoTT book, the truncation of any type A has two constructors: >> >> >> >> 1) for any a : A, there is |a| : ||A|| >> >> 2) for any x,y : ||A||, x = y. >> >> >> >> I get that if A is inhabited, then ||A|| is inhabited by (1). But is >> it true that, if ||A|| is inhabited, then A is inhabited? >> >> -- >> >> You received this message because you are subscribed to the Google >> Groups "Homotopy Type Theory" group. >> >> To unsubscribe from this group and stop receiving emails from it, send >> an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. >> >> For more options, visit https://groups.google.com/d/optout. >> > >> > >> > -- >> > You received this message because you are subscribed to the Google >> Groups "Homotopy Type Theory" group. >> > To unsubscribe from this group and stop receiving emails from it, send >> an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. >> > For more options, visit https://groups.google.com/d/optout. >> >> -- You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. For more options, visit https://groups.google.com/d/optout. [-- Attachment #1.2: Type: text/html, Size: 6480 bytes --] <div dir="ltr"><div>The existence property is proved for CCHM cubicaltt by Simon in:</div><div><br></div><div>https://arxiv.org/abs/1607.04156</div><div><br></div><div>See corollary 5.2. This works a bit more generally than what Martín said, in particular in any context with only dimension variables we can compute a witness to an existence. So if in context G = i_1 : II, ..., i_n : II (possibly empty) we have:<br></div><div><br></div><div>G |- t : exists (x : X), A(x)</div><div><br></div><div>then we can compute G |- u : X so that G |- B(u).</div><div><br></div><div>--</div><div>Anders<br></div><br>On Thursday, March 7, 2019 at 11:16:48 AM UTC-5, Martín Hötzel Escardó wrote:<blockquote class="gmail_quote" style="margin: 0;margin-left: 0.8ex;border-left: 1px #ccc solid;padding-left: 1ex;"><div dir="ltr">I got confused now. :-)<div><br></div><div>Seriously now, what you say seems related to the fact that from a proof |- t : || X || in the empty context, you get |- x : X in cubical type theory. This follows from Simon's canonicity result (at least for X=natural numbers), and is like the so-called "existence property" in the internal language of the free elementary topos. This says that from a proof |- exists (x:X), A x in the empty context, you get |- x : X and |- A x. This says that exists in the empty context behaves like Sigma. But only in the empty context, because otherwise it behaves like "local existence" as in Kripke-Joyal semantics. </div><div><br></div><div>Martin</div><div><br>On Thursday, 7 March 2019 14:10:56 UTC, dlicata wrote:<blockquote class="gmail_quote" style="margin:0;margin-left:0.8ex;border-left:1px #ccc solid;padding-left:1ex">Just in case anyone reading this thread later is confused about a more beginner point than the ones Nicolai and Martin made, one possible stumbling block here is that, if someone means “is inhabited” in an external sense (there is a closed term of that type), then the answer is yes (at least in some models): if ||A|| is inhabited then A is inhabited. For example, in cubical models with canonicity, it is true that a closed term of type ||A|| evaluates to a value that has as a subterm a closed term of type A (the other values of ||A|| are some “formal compositions” of values of ||A||, but there has to be an |a| in there at the base case). This is consistent with what Martin and Nicolai said because “if A is inhabited then B is inhabited” (in this external sense) doesn’t necessarily mean there is a map A -> B internally. <br> <br>-Dan <br> <br>> On Mar 5, 2019, at 6:07 PM, Martín Hötzel Escardó <<a rel="nofollow">escardo...@gmail.com</a>> wrote: <br>> <br>> Or you can read the paper <a href="https://lmcs.episciences.org/3217/" rel="nofollow" target="_blank" onmousedown="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Flmcs.episciences.org%2F3217%2F\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNF2sTOfc7ovV50gWLoaFL_7S1LECg';return true;" onclick="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Flmcs.episciences.org%2F3217%2F\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNF2sTOfc7ovV50gWLoaFL_7S1LECg';return true;">https://lmcs.episciences.org/<wbr>3217/</a> regarding what Nicolai said. <br>> <br>> Moreover, in the HoTT book, it is shown that if || X||->X holds for all X, then univalence can't hold. (It is global choice, which can't be invariant under equivalence.) <br>> <br>> The above paper shows that unrestricted ||X||->X it gives excluded middle. <br>> <br>> However, for a lot of kinds of types one can show that ||X||->X does hold. For example, if they have a constant endo-function. Moreover, for any type X, the availability of ||X||->X is logically equivalent to the availability of a constant map X->X (before we know whether X has a point or not, in which case the availability of a constant endo-map is trivial). <br>> <br>> Martin <br>> <br>> On Tuesday, 5 March 2019 22:47:55 UTC, Nicolai Kraus wrote: <br>> You can't have a function which, for all A, gives you ||A|| -> A. See the exercises 3.11 and 3.12! <br>> -- Nicolai <br>> <br>> On 05/03/19 22:31, Jean Joseph wrote: <br>>> Hi, <br>>> <br>>> From the HoTT book, the truncation of any type A has two constructors: <br>>> <br>>> 1) for any a : A, there is |a| : ||A|| <br>>> 2) for any x,y : ||A||, x = y. <br>>> <br>>> I get that if A is inhabited, then ||A|| is inhabited by (1). But is it true that, if ||A|| is inhabited, then A is inhabited? <br>>> -- <br>>> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. <br>>> To unsubscribe from this group and stop receiving emails from it, send an email to <a rel="nofollow">HomotopyTypeTheory+<wbr>unsubscribe@googlegroups.com</a>. <br>>> For more options, visit <a href="https://groups.google.com/d/optout" rel="nofollow" target="_blank" onmousedown="this.href='https://groups.google.com/d/optout';return true;" onclick="this.href='https://groups.google.com/d/optout';return true;">https://groups.google.com/d/<wbr>optout</a>. <br>> <br>> <br>> -- <br>> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. <br>> To unsubscribe from this group and stop receiving emails from it, send an email to <a rel="nofollow">HomotopyTypeTheory+<wbr>unsubscribe@googlegroups.com</a>. <br>> For more options, visit <a href="https://groups.google.com/d/optout" rel="nofollow" target="_blank" onmousedown="this.href='https://groups.google.com/d/optout';return true;" onclick="this.href='https://groups.google.com/d/optout';return true;">https://groups.google.com/d/<wbr>optout</a>. <br> <br></blockquote></div></div></blockquote></div> <p></p> -- <br /> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group.<br /> To unsubscribe from this group and stop receiving emails from it, send an email to <a href="mailto:HomotopyTypeTheory+unsubscribe@googlegroups.com">HomotopyTypeTheory+unsubscribe@googlegroups.com</a>.<br /> For more options, visit <a href="https://groups.google.com/d/optout">https://groups.google.com/d/optout</a>.<br />

[-- Attachment #1.1: Type: text/plain, Size: 5162 bytes --] So I presume that when we ask cubical Agda to normalize a term of type || Sigma (x:X), A x || we will get a term of the form |x,a| and so we will see the x in normal form, where |-| is the map into the truncation, right? Martin. On Thursday, 7 March 2019 21:52:12 UTC, Anders Mörtberg wrote: > > The existence property is proved for CCHM cubicaltt by Simon in: > > https://arxiv.org/abs/1607.04156 > > See corollary 5.2. This works a bit more generally than what Martín said, > in particular in any context with only dimension variables we can compute a > witness to an existence. So if in context G = i_1 : II, ..., i_n : II > (possibly empty) we have: > > G |- t : exists (x : X), A(x) > > then we can compute G |- u : X so that G |- B(u). > > -- > Anders > > On Thursday, March 7, 2019 at 11:16:48 AM UTC-5, Martín Hötzel Escardó > wrote: >> >> I got confused now. :-) >> >> Seriously now, what you say seems related to the fact that from a proof >> |- t : || X || in the empty context, you get |- x : X in cubical type >> theory. This follows from Simon's canonicity result (at least for X=natural >> numbers), and is like the so-called "existence property" in the internal >> language of the free elementary topos. This says that from a proof |- >> exists (x:X), A x in the empty context, you get |- x : X and |- A x. This >> says that exists in the empty context behaves like Sigma. But only in the >> empty context, because otherwise it behaves like "local existence" as in >> Kripke-Joyal semantics. >> >> Martin >> >> On Thursday, 7 March 2019 14:10:56 UTC, dlicata wrote: >>> >>> Just in case anyone reading this thread later is confused about a more >>> beginner point than the ones Nicolai and Martin made, one possible >>> stumbling block here is that, if someone means “is inhabited” in an >>> external sense (there is a closed term of that type), then the answer is >>> yes (at least in some models): if ||A|| is inhabited then A is inhabited. >>> For example, in cubical models with canonicity, it is true that a closed >>> term of type ||A|| evaluates to a value that has as a subterm a closed term >>> of type A (the other values of ||A|| are some “formal compositions” of >>> values of ||A||, but there has to be an |a| in there at the base case). >>> This is consistent with what Martin and Nicolai said because “if A is >>> inhabited then B is inhabited” (in this external sense) doesn’t necessarily >>> mean there is a map A -> B internally. >>> >>> -Dan >>> >>> > On Mar 5, 2019, at 6:07 PM, Martín Hötzel Escardó < >>> escardo...@gmail.com> wrote: >>> > >>> > Or you can read the paper https://lmcs.episciences.org/3217/ >>> regarding what Nicolai said. >>> > >>> > Moreover, in the HoTT book, it is shown that if || X||->X holds for >>> all X, then univalence can't hold. (It is global choice, which can't be >>> invariant under equivalence.) >>> > >>> > The above paper shows that unrestricted ||X||->X it gives excluded >>> middle. >>> > >>> > However, for a lot of kinds of types one can show that ||X||->X does >>> hold. For example, if they have a constant endo-function. Moreover, for any >>> type X, the availability of ||X||->X is logically equivalent to the >>> availability of a constant map X->X (before we know whether X has a point >>> or not, in which case the availability of a constant endo-map is trivial). >>> > >>> > Martin >>> > >>> > On Tuesday, 5 March 2019 22:47:55 UTC, Nicolai Kraus wrote: >>> > You can't have a function which, for all A, gives you ||A|| -> A. See >>> the exercises 3.11 and 3.12! >>> > -- Nicolai >>> > >>> > On 05/03/19 22:31, Jean Joseph wrote: >>> >> Hi, >>> >> >>> >> From the HoTT book, the truncation of any type A has two >>> constructors: >>> >> >>> >> 1) for any a : A, there is |a| : ||A|| >>> >> 2) for any x,y : ||A||, x = y. >>> >> >>> >> I get that if A is inhabited, then ||A|| is inhabited by (1). But is >>> it true that, if ||A|| is inhabited, then A is inhabited? >>> >> -- >>> >> You received this message because you are subscribed to the Google >>> Groups "Homotopy Type Theory" group. >>> >> To unsubscribe from this group and stop receiving emails from it, >>> send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. >>> >> For more options, visit https://groups.google.com/d/optout. >>> > >>> > >>> > -- >>> > You received this message because you are subscribed to the Google >>> Groups "Homotopy Type Theory" group. >>> > To unsubscribe from this group and stop receiving emails from it, send >>> an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. >>> > For more options, visit https://groups.google.com/d/optout. >>> >>> -- You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. For more options, visit https://groups.google.com/d/optout. [-- Attachment #1.2: Type: text/html, Size: 7386 bytes --] <div dir="ltr">So I presume that when we ask cubical Agda to normalize a term of type || Sigma (x:X), A x || we will get a term of the form |x,a| and so we will see the x in normal form, where |-| is the map into the truncation, right? Martin.<br><br>On Thursday, 7 March 2019 21:52:12 UTC, Anders Mörtberg wrote:<blockquote class="gmail_quote" style="margin: 0;margin-left: 0.8ex;border-left: 1px #ccc solid;padding-left: 1ex;"><div dir="ltr"><div>The existence property is proved for CCHM cubicaltt by Simon in:</div><div><br></div><div><a href="https://arxiv.org/abs/1607.04156" target="_blank" rel="nofollow" onmousedown="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Farxiv.org%2Fabs%2F1607.04156\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNHaiqP0--DZhiZa0817yu3ILyH9TQ';return true;" onclick="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Farxiv.org%2Fabs%2F1607.04156\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNHaiqP0--DZhiZa0817yu3ILyH9TQ';return true;">https://arxiv.org/abs/1607.<wbr>04156</a></div><div><br></div><div>See corollary 5.2. This works a bit more generally than what Martín said, in particular in any context with only dimension variables we can compute a witness to an existence. So if in context G = i_1 : II, ..., i_n : II (possibly empty) we have:<br></div><div><br></div><div>G |- t : exists (x : X), A(x)</div><div><br></div><div>then we can compute G |- u : X so that G |- B(u).</div><div><br></div><div>--</div><div>Anders<br></div><br>On Thursday, March 7, 2019 at 11:16:48 AM UTC-5, Martín Hötzel Escardó wrote:<blockquote class="gmail_quote" style="margin:0;margin-left:0.8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr">I got confused now. :-)<div><br></div><div>Seriously now, what you say seems related to the fact that from a proof |- t : || X || in the empty context, you get |- x : X in cubical type theory. This follows from Simon's canonicity result (at least for X=natural numbers), and is like the so-called "existence property" in the internal language of the free elementary topos. This says that from a proof |- exists (x:X), A x in the empty context, you get |- x : X and |- A x. This says that exists in the empty context behaves like Sigma. But only in the empty context, because otherwise it behaves like "local existence" as in Kripke-Joyal semantics. </div><div><br></div><div>Martin</div><div><br>On Thursday, 7 March 2019 14:10:56 UTC, dlicata wrote:<blockquote class="gmail_quote" style="margin:0;margin-left:0.8ex;border-left:1px #ccc solid;padding-left:1ex">Just in case anyone reading this thread later is confused about a more beginner point than the ones Nicolai and Martin made, one possible stumbling block here is that, if someone means “is inhabited” in an external sense (there is a closed term of that type), then the answer is yes (at least in some models): if ||A|| is inhabited then A is inhabited. For example, in cubical models with canonicity, it is true that a closed term of type ||A|| evaluates to a value that has as a subterm a closed term of type A (the other values of ||A|| are some “formal compositions” of values of ||A||, but there has to be an |a| in there at the base case). This is consistent with what Martin and Nicolai said because “if A is inhabited then B is inhabited” (in this external sense) doesn’t necessarily mean there is a map A -> B internally. <br> <br>-Dan <br> <br>> On Mar 5, 2019, at 6:07 PM, Martín Hötzel Escardó <<a rel="nofollow">escardo...@gmail.com</a>> wrote: <br>> <br>> Or you can read the paper <a href="https://lmcs.episciences.org/3217/" rel="nofollow" target="_blank" onmousedown="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Flmcs.episciences.org%2F3217%2F\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNF2sTOfc7ovV50gWLoaFL_7S1LECg';return true;" onclick="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Flmcs.episciences.org%2F3217%2F\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNF2sTOfc7ovV50gWLoaFL_7S1LECg';return true;">https://lmcs.episciences.org/<wbr>3217/</a> regarding what Nicolai said. <br>> <br>> Moreover, in the HoTT book, it is shown that if || X||->X holds for all X, then univalence can't hold. (It is global choice, which can't be invariant under equivalence.) <br>> <br>> The above paper shows that unrestricted ||X||->X it gives excluded middle. <br>> <br>> However, for a lot of kinds of types one can show that ||X||->X does hold. For example, if they have a constant endo-function. Moreover, for any type X, the availability of ||X||->X is logically equivalent to the availability of a constant map X->X (before we know whether X has a point or not, in which case the availability of a constant endo-map is trivial). <br>> <br>> Martin <br>> <br>> On Tuesday, 5 March 2019 22:47:55 UTC, Nicolai Kraus wrote: <br>> You can't have a function which, for all A, gives you ||A|| -> A. See the exercises 3.11 and 3.12! <br>> -- Nicolai <br>> <br>> On 05/03/19 22:31, Jean Joseph wrote: <br>>> Hi, <br>>> <br>>> From the HoTT book, the truncation of any type A has two constructors: <br>>> <br>>> 1) for any a : A, there is |a| : ||A|| <br>>> 2) for any x,y : ||A||, x = y. <br>>> <br>>> I get that if A is inhabited, then ||A|| is inhabited by (1). But is it true that, if ||A|| is inhabited, then A is inhabited? <br>>> -- <br>>> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. <br>>> To unsubscribe from this group and stop receiving emails from it, send an email to <a rel="nofollow">HomotopyTypeTheory+<wbr>unsubscribe@googlegroups.com</a>. <br>>> For more options, visit <a href="https://groups.google.com/d/optout" rel="nofollow" target="_blank" onmousedown="this.href='https://groups.google.com/d/optout';return true;" onclick="this.href='https://groups.google.com/d/optout';return true;">https://groups.google.com/d/<wbr>optout</a>. <br>> <br>> <br>> -- <br>> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. <br>> To unsubscribe from this group and stop receiving emails from it, send an email to <a rel="nofollow">HomotopyTypeTheory+<wbr>unsubscribe@googlegroups.com</a>. <br>> For more options, visit <a href="https://groups.google.com/d/optout" rel="nofollow" target="_blank" onmousedown="this.href='https://groups.google.com/d/optout';return true;" onclick="this.href='https://groups.google.com/d/optout';return true;">https://groups.google.com/d/<wbr>optout</a>. <br> <br></blockquote></div></div></blockquote></div></blockquote></div> <p></p> -- <br /> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group.<br /> To unsubscribe from this group and stop receiving emails from it, send an email to <a href="mailto:HomotopyTypeTheory+unsubscribe@googlegroups.com">HomotopyTypeTheory+unsubscribe@googlegroups.com</a>.<br /> For more options, visit <a href="https://groups.google.com/d/optout">https://groups.google.com/d/optout</a>.<br />

That would be true if the term you are normalizing is in the empty interval context, and the cubical type theory has “empty system regularity” (like https://www.cs.cmu.edu/~cangiuli/papers/ccctt.pdf). Otherwise, if you evaluate something in the empty interval context, you might see something like hcom [] (hcom [] (hcom [] (hcom [] (… |x,a| … )))) with |x,a| in there somewhere. In HITs, Kan composition is treated as a constructor of the type, and though there are no interesting lines to compose in the empty interval context, the uninteresting compositions don’t vanish in all flavors of cubical type theory. > On Mar 7, 2019, at 5:41 PM, Martín Hötzel Escardó <escardo.martin@gmail.com> wrote: > > So I presume that when we ask cubical Agda to normalize a term of type || Sigma (x:X), A x || we will get a term of the form |x,a| and so we will see the x in normal form, where |-| is the map into the truncation, right? Martin. > > On Thursday, 7 March 2019 21:52:12 UTC, Anders Mörtberg wrote: > The existence property is proved for CCHM cubicaltt by Simon in: > > https://arxiv.org/abs/1607.04156 > > See corollary 5.2. This works a bit more generally than what Martín said, in particular in any context with only dimension variables we can compute a witness to an existence. So if in context G = i_1 : II, ..., i_n : II (possibly empty) we have: > > G |- t : exists (x : X), A(x) > > then we can compute G |- u : X so that G |- B(u). > > -- > Anders > > On Thursday, March 7, 2019 at 11:16:48 AM UTC-5, Martín Hötzel Escardó wrote: > I got confused now. :-) > > Seriously now, what you say seems related to the fact that from a proof |- t : || X || in the empty context, you get |- x : X in cubical type theory. This follows from Simon's canonicity result (at least for X=natural numbers), and is like the so-called "existence property" in the internal language of the free elementary topos. This says that from a proof |- exists (x:X), A x in the empty context, you get |- x : X and |- A x. This says that exists in the empty context behaves like Sigma. But only in the empty context, because otherwise it behaves like "local existence" as in Kripke-Joyal semantics. > > Martin > > On Thursday, 7 March 2019 14:10:56 UTC, dlicata wrote: > Just in case anyone reading this thread later is confused about a more beginner point than the ones Nicolai and Martin made, one possible stumbling block here is that, if someone means “is inhabited” in an external sense (there is a closed term of that type), then the answer is yes (at least in some models): if ||A|| is inhabited then A is inhabited. For example, in cubical models with canonicity, it is true that a closed term of type ||A|| evaluates to a value that has as a subterm a closed term of type A (the other values of ||A|| are some “formal compositions” of values of ||A||, but there has to be an |a| in there at the base case). This is consistent with what Martin and Nicolai said because “if A is inhabited then B is inhabited” (in this external sense) doesn’t necessarily mean there is a map A -> B internally. > > -Dan > > > On Mar 5, 2019, at 6:07 PM, Martín Hötzel Escardó <escardo...@gmail.com> wrote: > > > > Or you can read the paper https://lmcs.episciences.org/3217/ regarding what Nicolai said. > > > > Moreover, in the HoTT book, it is shown that if || X||->X holds for all X, then univalence can't hold. (It is global choice, which can't be invariant under equivalence.) > > > > The above paper shows that unrestricted ||X||->X it gives excluded middle. > > > > However, for a lot of kinds of types one can show that ||X||->X does hold. For example, if they have a constant endo-function. Moreover, for any type X, the availability of ||X||->X is logically equivalent to the availability of a constant map X->X (before we know whether X has a point or not, in which case the availability of a constant endo-map is trivial). > > > > Martin > > > > On Tuesday, 5 March 2019 22:47:55 UTC, Nicolai Kraus wrote: > > You can't have a function which, for all A, gives you ||A|| -> A. See the exercises 3.11 and 3.12! > > -- Nicolai > > > > On 05/03/19 22:31, Jean Joseph wrote: > >> Hi, > >> > >> From the HoTT book, the truncation of any type A has two constructors: > >> > >> 1) for any a : A, there is |a| : ||A|| > >> 2) for any x,y : ||A||, x = y. > >> > >> I get that if A is inhabited, then ||A|| is inhabited by (1). But is it true that, if ||A|| is inhabited, then A is inhabited? > >> -- > >> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. > >> To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. > >> For more options, visit https://groups.google.com/d/optout. > > > > > > -- > > You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. > > To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. > > For more options, visit https://groups.google.com/d/optout. > > > -- > You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. > To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. > For more options, visit https://groups.google.com/d/optout. -- You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. For more options, visit https://groups.google.com/d/optout.

[-- Attachment #1.1: Type: text/plain, Size: 6618 bytes --] Oh, this is annoying, because it seems to mean that we would need unbounded search (to drop all "hcom []"'s) until we can read the |x,a|, which is against the spirit of, say, Martin-Loef type theories. Martin On Thursday, 7 March 2019 22:51:20 UTC, dlicata wrote: > > That would be true if the term you are normalizing is in the empty > interval context, and the cubical type theory has “empty system regularity” > (like https://www.cs.cmu.edu/~cangiuli/papers/ccctt.pdf). > > Otherwise, if you evaluate something in the empty interval context, you > might see something like > hcom [] (hcom [] (hcom [] (hcom [] (… |x,a| … )))) > with |x,a| in there somewhere. In HITs, Kan composition is treated as a > constructor of the type, and though there are no interesting lines to > compose in the empty interval context, the uninteresting compositions don’t > vanish in all flavors of cubical type theory. > > > On Mar 7, 2019, at 5:41 PM, Martín Hötzel Escardó <escardo...@gmail.com > <javascript:>> wrote: > > > > So I presume that when we ask cubical Agda to normalize a term of type > || Sigma (x:X), A x || we will get a term of the form |x,a| and so we will > see the x in normal form, where |-| is the map into the truncation, right? > Martin. > > > > On Thursday, 7 March 2019 21:52:12 UTC, Anders Mörtberg wrote: > > The existence property is proved for CCHM cubicaltt by Simon in: > > > > https://arxiv.org/abs/1607.04156 > > > > See corollary 5.2. This works a bit more generally than what Martín > said, in particular in any context with only dimension variables we can > compute a witness to an existence. So if in context G = i_1 : II, ..., i_n > : II (possibly empty) we have: > > > > G |- t : exists (x : X), A(x) > > > > then we can compute G |- u : X so that G |- B(u). > > > > -- > > Anders > > > > On Thursday, March 7, 2019 at 11:16:48 AM UTC-5, Martín Hötzel Escardó > wrote: > > I got confused now. :-) > > > > Seriously now, what you say seems related to the fact that from a proof > |- t : || X || in the empty context, you get |- x : X in cubical type > theory. This follows from Simon's canonicity result (at least for X=natural > numbers), and is like the so-called "existence property" in the internal > language of the free elementary topos. This says that from a proof |- > exists (x:X), A x in the empty context, you get |- x : X and |- A x. This > says that exists in the empty context behaves like Sigma. But only in the > empty context, because otherwise it behaves like "local existence" as in > Kripke-Joyal semantics. > > > > Martin > > > > On Thursday, 7 March 2019 14:10:56 UTC, dlicata wrote: > > Just in case anyone reading this thread later is confused about a more > beginner point than the ones Nicolai and Martin made, one possible > stumbling block here is that, if someone means “is inhabited” in an > external sense (there is a closed term of that type), then the answer is > yes (at least in some models): if ||A|| is inhabited then A is inhabited. > For example, in cubical models with canonicity, it is true that a closed > term of type ||A|| evaluates to a value that has as a subterm a closed term > of type A (the other values of ||A|| are some “formal compositions” of > values of ||A||, but there has to be an |a| in there at the base case). > This is consistent with what Martin and Nicolai said because “if A is > inhabited then B is inhabited” (in this external sense) doesn’t necessarily > mean there is a map A -> B internally. > > > > -Dan > > > > > On Mar 5, 2019, at 6:07 PM, Martín Hötzel Escardó < > escardo...@gmail.com> wrote: > > > > > > Or you can read the paper https://lmcs.episciences.org/3217/ > regarding what Nicolai said. > > > > > > Moreover, in the HoTT book, it is shown that if || X||->X holds for > all X, then univalence can't hold. (It is global choice, which can't be > invariant under equivalence.) > > > > > > The above paper shows that unrestricted ||X||->X it gives excluded > middle. > > > > > > However, for a lot of kinds of types one can show that ||X||->X does > hold. For example, if they have a constant endo-function. Moreover, for any > type X, the availability of ||X||->X is logically equivalent to the > availability of a constant map X->X (before we know whether X has a point > or not, in which case the availability of a constant endo-map is trivial). > > > > > > Martin > > > > > > On Tuesday, 5 March 2019 22:47:55 UTC, Nicolai Kraus wrote: > > > You can't have a function which, for all A, gives you ||A|| -> A. See > the exercises 3.11 and 3.12! > > > -- Nicolai > > > > > > On 05/03/19 22:31, Jean Joseph wrote: > > >> Hi, > > >> > > >> From the HoTT book, the truncation of any type A has two > constructors: > > >> > > >> 1) for any a : A, there is |a| : ||A|| > > >> 2) for any x,y : ||A||, x = y. > > >> > > >> I get that if A is inhabited, then ||A|| is inhabited by (1). But is > it true that, if ||A|| is inhabited, then A is inhabited? > > >> -- > > >> You received this message because you are subscribed to the Google > Groups "Homotopy Type Theory" group. > > >> To unsubscribe from this group and stop receiving emails from it, > send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com > <javascript:>. > > >> For more options, visit https://groups.google.com/d/optout. > > > > > > > > > -- > > > You received this message because you are subscribed to the Google > Groups "Homotopy Type Theory" group. > > > To unsubscribe from this group and stop receiving emails from it, send > an email to HomotopyTypeTheory+unsubscribe@googlegroups.com <javascript:>. > > > > For more options, visit https://groups.google.com/d/optout. > > > > > > -- > > You received this message because you are subscribed to the Google > Groups "Homotopy Type Theory" group. > > To unsubscribe from this group and stop receiving emails from it, send > an email to HomotopyTypeTheory+unsubscribe@googlegroups.com <javascript:>. > > > For more options, visit https://groups.google.com/d/optout. > > -- You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. For more options, visit https://groups.google.com/d/optout. [-- Attachment #1.2: Type: text/html, Size: 10303 bytes --] <div dir="ltr">Oh, this is annoying, because it seems to mean that we would need unbounded search (to drop all "hcom []"'s) until we can read the |x,a|, which is against the spirit of, say, Martin-Loef type theories. Martin<br><br>On Thursday, 7 March 2019 22:51:20 UTC, dlicata wrote:<blockquote class="gmail_quote" style="margin: 0;margin-left: 0.8ex;border-left: 1px #ccc solid;padding-left: 1ex;">That would be true if the term you are normalizing is in the empty interval context, and the cubical type theory has “empty system regularity” (like <a href="https://www.cs.cmu.edu/~cangiuli/papers/ccctt.pdf" target="_blank" rel="nofollow" onmousedown="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Fwww.cs.cmu.edu%2F~cangiuli%2Fpapers%2Fccctt.pdf\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNESV2Pele9DWGVlPQp1lf9qMnycyw';return true;" onclick="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Fwww.cs.cmu.edu%2F~cangiuli%2Fpapers%2Fccctt.pdf\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNESV2Pele9DWGVlPQp1lf9qMnycyw';return true;">https://www.cs.cmu.edu/~<wbr>cangiuli/papers/ccctt.pdf</a>). <br> <br>Otherwise, if you evaluate something in the empty interval context, you might see something like <br>hcom [] (hcom [] (hcom [] (hcom [] (… |x,a| … )))) <br>with |x,a| in there somewhere. In HITs, Kan composition is treated as a constructor of the type, and though there are no interesting lines to compose in the empty interval context, the uninteresting compositions don’t vanish in all flavors of cubical type theory. <br> <br>> On Mar 7, 2019, at 5:41 PM, Martín Hötzel Escardó <<a href="javascript:" target="_blank" gdf-obfuscated-mailto="jtq7x581AwAJ" rel="nofollow" onmousedown="this.href='javascript:';return true;" onclick="this.href='javascript:';return true;">escardo...@gmail.com</a>> wrote: <br>> <br>> So I presume that when we ask cubical Agda to normalize a term of type || Sigma (x:X), A x || we will get a term of the form |x,a| and so we will see the x in normal form, where |-| is the map into the truncation, right? Martin. <br>> <br>> On Thursday, 7 March 2019 21:52:12 UTC, Anders Mörtberg wrote: <br>> The existence property is proved for CCHM cubicaltt by Simon in: <br>> <br>> <a href="https://arxiv.org/abs/1607.04156" target="_blank" rel="nofollow" onmousedown="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Farxiv.org%2Fabs%2F1607.04156\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNHaiqP0--DZhiZa0817yu3ILyH9TQ';return true;" onclick="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Farxiv.org%2Fabs%2F1607.04156\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNHaiqP0--DZhiZa0817yu3ILyH9TQ';return true;">https://arxiv.org/abs/1607.<wbr>04156</a> <br>> <br>> See corollary 5.2. This works a bit more generally than what Martín said, in particular in any context with only dimension variables we can compute a witness to an existence. So if in context G = i_1 : II, ..., i_n : II (possibly empty) we have: <br>> <br>> G |- t : exists (x : X), A(x) <br>> <br>> then we can compute G |- u : X so that G |- B(u). <br>> <br>> -- <br>> Anders <br>> <br>> On Thursday, March 7, 2019 at 11:16:48 AM UTC-5, Martín Hötzel Escardó wrote: <br>> I got confused now. :-) <br>> <br>> Seriously now, what you say seems related to the fact that from a proof |- t : || X || in the empty context, you get |- x : X in cubical type theory. This follows from Simon's canonicity result (at least for X=natural numbers), and is like the so-called "existence property" in the internal language of the free elementary topos. This says that from a proof |- exists (x:X), A x in the empty context, you get |- x : X and |- A x. This says that exists in the empty context behaves like Sigma. But only in the empty context, because otherwise it behaves like "local existence" as in Kripke-Joyal semantics. <br>> <br>> Martin <br>> <br>> On Thursday, 7 March 2019 14:10:56 UTC, dlicata wrote: <br>> Just in case anyone reading this thread later is confused about a more beginner point than the ones Nicolai and Martin made, one possible stumbling block here is that, if someone means “is inhabited” in an external sense (there is a closed term of that type), then the answer is yes (at least in some models): if ||A|| is inhabited then A is inhabited. For example, in cubical models with canonicity, it is true that a closed term of type ||A|| evaluates to a value that has as a subterm a closed term of type A (the other values of ||A|| are some “formal compositions” of values of ||A||, but there has to be an |a| in there at the base case). This is consistent with what Martin and Nicolai said because “if A is inhabited then B is inhabited” (in this external sense) doesn’t necessarily mean there is a map A -> B internally. <br>> <br>> -Dan <br>> <br>> > On Mar 5, 2019, at 6:07 PM, Martín Hötzel Escardó <<a>escardo...@gmail.com</a>> wrote: <br>> > <br>> > Or you can read the paper <a href="https://lmcs.episciences.org/3217/" target="_blank" rel="nofollow" onmousedown="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Flmcs.episciences.org%2F3217%2F\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNF2sTOfc7ovV50gWLoaFL_7S1LECg';return true;" onclick="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Flmcs.episciences.org%2F3217%2F\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNF2sTOfc7ovV50gWLoaFL_7S1LECg';return true;">https://lmcs.episciences.org/<wbr>3217/</a> regarding what Nicolai said. <br>> > <br>> > Moreover, in the HoTT book, it is shown that if || X||->X holds for all X, then univalence can't hold. (It is global choice, which can't be invariant under equivalence.) <br>> > <br>> > The above paper shows that unrestricted ||X||->X it gives excluded middle. <br>> > <br>> > However, for a lot of kinds of types one can show that ||X||->X does hold. For example, if they have a constant endo-function. Moreover, for any type X, the availability of ||X||->X is logically equivalent to the availability of a constant map X->X (before we know whether X has a point or not, in which case the availability of a constant endo-map is trivial). <br>> > <br>> > Martin <br>> > <br>> > On Tuesday, 5 March 2019 22:47:55 UTC, Nicolai Kraus wrote: <br>> > You can't have a function which, for all A, gives you ||A|| -> A. See the exercises 3.11 and 3.12! <br>> > -- Nicolai <br>> > <br>> > On 05/03/19 22:31, Jean Joseph wrote: <br>> >> Hi, <br>> >> <br>> >> From the HoTT book, the truncation of any type A has two constructors: <br>> >> <br>> >> 1) for any a : A, there is |a| : ||A|| <br>> >> 2) for any x,y : ||A||, x = y. <br>> >> <br>> >> I get that if A is inhabited, then ||A|| is inhabited by (1). But is it true that, if ||A|| is inhabited, then A is inhabited? <br>> >> -- <br>> >> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. <br>> >> To unsubscribe from this group and stop receiving emails from it, send an email to <a href="javascript:" target="_blank" gdf-obfuscated-mailto="jtq7x581AwAJ" rel="nofollow" onmousedown="this.href='javascript:';return true;" onclick="this.href='javascript:';return true;">HomotopyTypeTheory+<wbr>unsubscribe@googlegroups.com</a>. <br>> >> For more options, visit <a href="https://groups.google.com/d/optout" target="_blank" rel="nofollow" onmousedown="this.href='https://groups.google.com/d/optout';return true;" onclick="this.href='https://groups.google.com/d/optout';return true;">https://groups.google.com/d/<wbr>optout</a>. <br>> > <br>> > <br>> > -- <br>> > You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. <br>> > To unsubscribe from this group and stop receiving emails from it, send an email to <a href="javascript:" target="_blank" gdf-obfuscated-mailto="jtq7x581AwAJ" rel="nofollow" onmousedown="this.href='javascript:';return true;" onclick="this.href='javascript:';return true;">HomotopyTypeTheory+<wbr>unsubscribe@googlegroups.com</a>. <br>> > For more options, visit <a href="https://groups.google.com/d/optout" target="_blank" rel="nofollow" onmousedown="this.href='https://groups.google.com/d/optout';return true;" onclick="this.href='https://groups.google.com/d/optout';return true;">https://groups.google.com/d/<wbr>optout</a>. <br>> <br>> <br>> -- <br>> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. <br>> To unsubscribe from this group and stop receiving emails from it, send an email to <a href="javascript:" target="_blank" gdf-obfuscated-mailto="jtq7x581AwAJ" rel="nofollow" onmousedown="this.href='javascript:';return true;" onclick="this.href='javascript:';return true;">HomotopyTypeTheory+<wbr>unsubscribe@googlegroups.com</a>. <br>> For more options, visit <a href="https://groups.google.com/d/optout" target="_blank" rel="nofollow" onmousedown="this.href='https://groups.google.com/d/optout';return true;" onclick="this.href='https://groups.google.com/d/optout';return true;">https://groups.google.com/d/<wbr>optout</a>. <br> <br></blockquote></div> <p></p> -- <br /> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group.<br /> To unsubscribe from this group and stop receiving emails from it, send an email to <a href="mailto:HomotopyTypeTheory+unsubscribe@googlegroups.com">HomotopyTypeTheory+unsubscribe@googlegroups.com</a>.<br /> For more options, visit <a href="https://groups.google.com/d/optout">https://groups.google.com/d/optout</a>.<br />

[-- Attachment #1.1: Type: text/plain, Size: 7219 bytes --] And this is a wildly speculative question. If we used Andrew Swan's identity type derived from the cubical path type only, as in the abstract library file https://github.com/agda/cubical/blob/master/Cubical/Core/HoTT-UF.agda) would we still get this phenomenon? Maybe not? What I mean is that we use normal Agda, together with what is offered in that file and nothing else (so that we are using HoTT book axiomatic type theory). Martin On Thursday, 7 March 2019 23:01:33 UTC, Martín Hötzel Escardó wrote: > > Oh, this is annoying, because it seems to mean that we would need > unbounded search (to drop all "hcom []"'s) until we can read the |x,a|, > which is against the spirit of, say, Martin-Loef type theories. Martin > > On Thursday, 7 March 2019 22:51:20 UTC, dlicata wrote: >> >> That would be true if the term you are normalizing is in the empty >> interval context, and the cubical type theory has “empty system regularity” >> (like https://www.cs.cmu.edu/~cangiuli/papers/ccctt.pdf). >> >> Otherwise, if you evaluate something in the empty interval context, you >> might see something like >> hcom [] (hcom [] (hcom [] (hcom [] (… |x,a| … )))) >> with |x,a| in there somewhere. In HITs, Kan composition is treated as a >> constructor of the type, and though there are no interesting lines to >> compose in the empty interval context, the uninteresting compositions don’t >> vanish in all flavors of cubical type theory. >> >> > On Mar 7, 2019, at 5:41 PM, Martín Hötzel Escardó <escardo...@gmail.com> >> wrote: >> > >> > So I presume that when we ask cubical Agda to normalize a term of type >> || Sigma (x:X), A x || we will get a term of the form |x,a| and so we will >> see the x in normal form, where |-| is the map into the truncation, right? >> Martin. >> > >> > On Thursday, 7 March 2019 21:52:12 UTC, Anders Mörtberg wrote: >> > The existence property is proved for CCHM cubicaltt by Simon in: >> > >> > https://arxiv.org/abs/1607.04156 >> > >> > See corollary 5.2. This works a bit more generally than what Martín >> said, in particular in any context with only dimension variables we can >> compute a witness to an existence. So if in context G = i_1 : II, ..., i_n >> : II (possibly empty) we have: >> > >> > G |- t : exists (x : X), A(x) >> > >> > then we can compute G |- u : X so that G |- B(u). >> > >> > -- >> > Anders >> > >> > On Thursday, March 7, 2019 at 11:16:48 AM UTC-5, Martín Hötzel Escardó >> wrote: >> > I got confused now. :-) >> > >> > Seriously now, what you say seems related to the fact that from a proof >> |- t : || X || in the empty context, you get |- x : X in cubical type >> theory. This follows from Simon's canonicity result (at least for X=natural >> numbers), and is like the so-called "existence property" in the internal >> language of the free elementary topos. This says that from a proof |- >> exists (x:X), A x in the empty context, you get |- x : X and |- A x. This >> says that exists in the empty context behaves like Sigma. But only in the >> empty context, because otherwise it behaves like "local existence" as in >> Kripke-Joyal semantics. >> > >> > Martin >> > >> > On Thursday, 7 March 2019 14:10:56 UTC, dlicata wrote: >> > Just in case anyone reading this thread later is confused about a more >> beginner point than the ones Nicolai and Martin made, one possible >> stumbling block here is that, if someone means “is inhabited” in an >> external sense (there is a closed term of that type), then the answer is >> yes (at least in some models): if ||A|| is inhabited then A is inhabited. >> For example, in cubical models with canonicity, it is true that a closed >> term of type ||A|| evaluates to a value that has as a subterm a closed term >> of type A (the other values of ||A|| are some “formal compositions” of >> values of ||A||, but there has to be an |a| in there at the base case). >> This is consistent with what Martin and Nicolai said because “if A is >> inhabited then B is inhabited” (in this external sense) doesn’t necessarily >> mean there is a map A -> B internally. >> > >> > -Dan >> > >> > > On Mar 5, 2019, at 6:07 PM, Martín Hötzel Escardó < >> escardo...@gmail.com> wrote: >> > > >> > > Or you can read the paper https://lmcs.episciences.org/3217/ >> regarding what Nicolai said. >> > > >> > > Moreover, in the HoTT book, it is shown that if || X||->X holds for >> all X, then univalence can't hold. (It is global choice, which can't be >> invariant under equivalence.) >> > > >> > > The above paper shows that unrestricted ||X||->X it gives excluded >> middle. >> > > >> > > However, for a lot of kinds of types one can show that ||X||->X does >> hold. For example, if they have a constant endo-function. Moreover, for any >> type X, the availability of ||X||->X is logically equivalent to the >> availability of a constant map X->X (before we know whether X has a point >> or not, in which case the availability of a constant endo-map is trivial). >> > > >> > > Martin >> > > >> > > On Tuesday, 5 March 2019 22:47:55 UTC, Nicolai Kraus wrote: >> > > You can't have a function which, for all A, gives you ||A|| -> A. See >> the exercises 3.11 and 3.12! >> > > -- Nicolai >> > > >> > > On 05/03/19 22:31, Jean Joseph wrote: >> > >> Hi, >> > >> >> > >> From the HoTT book, the truncation of any type A has two >> constructors: >> > >> >> > >> 1) for any a : A, there is |a| : ||A|| >> > >> 2) for any x,y : ||A||, x = y. >> > >> >> > >> I get that if A is inhabited, then ||A|| is inhabited by (1). But is >> it true that, if ||A|| is inhabited, then A is inhabited? >> > >> -- >> > >> You received this message because you are subscribed to the Google >> Groups "Homotopy Type Theory" group. >> > >> To unsubscribe from this group and stop receiving emails from it, >> send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. >> > >> For more options, visit https://groups.google.com/d/optout. >> > > >> > > >> > > -- >> > > You received this message because you are subscribed to the Google >> Groups "Homotopy Type Theory" group. >> > > To unsubscribe from this group and stop receiving emails from it, >> send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. >> > > For more options, visit https://groups.google.com/d/optout. >> > >> > >> > -- >> > You received this message because you are subscribed to the Google >> Groups "Homotopy Type Theory" group. >> > To unsubscribe from this group and stop receiving emails from it, send >> an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. >> > For more options, visit https://groups.google.com/d/optout. >> >> -- You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. For more options, visit https://groups.google.com/d/optout. [-- Attachment #1.2: Type: text/html, Size: 10223 bytes --] <div dir="ltr">And this is a wildly speculative question. If we used Andrew Swan's identity type derived from the cubical path type only, as in the abstract library file https://github.com/agda/cubical/blob/master/Cubical/Core/HoTT-UF.agda) would we still get this phenomenon? Maybe not? What I mean is that we use normal Agda, together with what is offered in that file and nothing else (so that we are using HoTT book axiomatic type theory). Martin<br><br>On Thursday, 7 March 2019 23:01:33 UTC, Martín Hötzel Escardó wrote:<blockquote class="gmail_quote" style="margin: 0;margin-left: 0.8ex;border-left: 1px #ccc solid;padding-left: 1ex;"><div dir="ltr">Oh, this is annoying, because it seems to mean that we would need unbounded search (to drop all "hcom []"'s) until we can read the |x,a|, which is against the spirit of, say, Martin-Loef type theories. Martin<br><br>On Thursday, 7 March 2019 22:51:20 UTC, dlicata wrote:<blockquote class="gmail_quote" style="margin:0;margin-left:0.8ex;border-left:1px #ccc solid;padding-left:1ex">That would be true if the term you are normalizing is in the empty interval context, and the cubical type theory has “empty system regularity” (like <a href="https://www.cs.cmu.edu/~cangiuli/papers/ccctt.pdf" rel="nofollow" target="_blank" onmousedown="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Fwww.cs.cmu.edu%2F~cangiuli%2Fpapers%2Fccctt.pdf\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNESV2Pele9DWGVlPQp1lf9qMnycyw';return true;" onclick="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Fwww.cs.cmu.edu%2F~cangiuli%2Fpapers%2Fccctt.pdf\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNESV2Pele9DWGVlPQp1lf9qMnycyw';return true;">https://www.cs.cmu.edu/~<wbr>cangiuli/papers/ccctt.pdf</a>). <br> <br>Otherwise, if you evaluate something in the empty interval context, you might see something like <br>hcom [] (hcom [] (hcom [] (hcom [] (… |x,a| … )))) <br>with |x,a| in there somewhere. In HITs, Kan composition is treated as a constructor of the type, and though there are no interesting lines to compose in the empty interval context, the uninteresting compositions don’t vanish in all flavors of cubical type theory. <br> <br>> On Mar 7, 2019, at 5:41 PM, Martín Hötzel Escardó <<a rel="nofollow">escardo...@gmail.com</a>> wrote: <br>> <br>> So I presume that when we ask cubical Agda to normalize a term of type || Sigma (x:X), A x || we will get a term of the form |x,a| and so we will see the x in normal form, where |-| is the map into the truncation, right? Martin. <br>> <br>> On Thursday, 7 March 2019 21:52:12 UTC, Anders Mörtberg wrote: <br>> The existence property is proved for CCHM cubicaltt by Simon in: <br>> <br>> <a href="https://arxiv.org/abs/1607.04156" rel="nofollow" target="_blank" onmousedown="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Farxiv.org%2Fabs%2F1607.04156\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNHaiqP0--DZhiZa0817yu3ILyH9TQ';return true;" onclick="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Farxiv.org%2Fabs%2F1607.04156\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNHaiqP0--DZhiZa0817yu3ILyH9TQ';return true;">https://arxiv.org/abs/1607.<wbr>04156</a> <br>> <br>> See corollary 5.2. This works a bit more generally than what Martín said, in particular in any context with only dimension variables we can compute a witness to an existence. So if in context G = i_1 : II, ..., i_n : II (possibly empty) we have: <br>> <br>> G |- t : exists (x : X), A(x) <br>> <br>> then we can compute G |- u : X so that G |- B(u). <br>> <br>> -- <br>> Anders <br>> <br>> On Thursday, March 7, 2019 at 11:16:48 AM UTC-5, Martín Hötzel Escardó wrote: <br>> I got confused now. :-) <br>> <br>> Seriously now, what you say seems related to the fact that from a proof |- t : || X || in the empty context, you get |- x : X in cubical type theory. This follows from Simon's canonicity result (at least for X=natural numbers), and is like the so-called "existence property" in the internal language of the free elementary topos. This says that from a proof |- exists (x:X), A x in the empty context, you get |- x : X and |- A x. This says that exists in the empty context behaves like Sigma. But only in the empty context, because otherwise it behaves like "local existence" as in Kripke-Joyal semantics. <br>> <br>> Martin <br>> <br>> On Thursday, 7 March 2019 14:10:56 UTC, dlicata wrote: <br>> Just in case anyone reading this thread later is confused about a more beginner point than the ones Nicolai and Martin made, one possible stumbling block here is that, if someone means “is inhabited” in an external sense (there is a closed term of that type), then the answer is yes (at least in some models): if ||A|| is inhabited then A is inhabited. For example, in cubical models with canonicity, it is true that a closed term of type ||A|| evaluates to a value that has as a subterm a closed term of type A (the other values of ||A|| are some “formal compositions” of values of ||A||, but there has to be an |a| in there at the base case). This is consistent with what Martin and Nicolai said because “if A is inhabited then B is inhabited” (in this external sense) doesn’t necessarily mean there is a map A -> B internally. <br>> <br>> -Dan <br>> <br>> > On Mar 5, 2019, at 6:07 PM, Martín Hötzel Escardó <<a>escardo...@gmail.com</a>> wrote: <br>> > <br>> > Or you can read the paper <a href="https://lmcs.episciences.org/3217/" rel="nofollow" target="_blank" onmousedown="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Flmcs.episciences.org%2F3217%2F\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNF2sTOfc7ovV50gWLoaFL_7S1LECg';return true;" onclick="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Flmcs.episciences.org%2F3217%2F\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNF2sTOfc7ovV50gWLoaFL_7S1LECg';return true;">https://lmcs.episciences.org/<wbr>3217/</a> regarding what Nicolai said. <br>> > <br>> > Moreover, in the HoTT book, it is shown that if || X||->X holds for all X, then univalence can't hold. (It is global choice, which can't be invariant under equivalence.) <br>> > <br>> > The above paper shows that unrestricted ||X||->X it gives excluded middle. <br>> > <br>> > However, for a lot of kinds of types one can show that ||X||->X does hold. For example, if they have a constant endo-function. Moreover, for any type X, the availability of ||X||->X is logically equivalent to the availability of a constant map X->X (before we know whether X has a point or not, in which case the availability of a constant endo-map is trivial). <br>> > <br>> > Martin <br>> > <br>> > On Tuesday, 5 March 2019 22:47:55 UTC, Nicolai Kraus wrote: <br>> > You can't have a function which, for all A, gives you ||A|| -> A. See the exercises 3.11 and 3.12! <br>> > -- Nicolai <br>> > <br>> > On 05/03/19 22:31, Jean Joseph wrote: <br>> >> Hi, <br>> >> <br>> >> From the HoTT book, the truncation of any type A has two constructors: <br>> >> <br>> >> 1) for any a : A, there is |a| : ||A|| <br>> >> 2) for any x,y : ||A||, x = y. <br>> >> <br>> >> I get that if A is inhabited, then ||A|| is inhabited by (1). But is it true that, if ||A|| is inhabited, then A is inhabited? <br>> >> -- <br>> >> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. <br>> >> To unsubscribe from this group and stop receiving emails from it, send an email to <a rel="nofollow">HomotopyTypeTheory+<wbr>unsubscribe@googlegroups.com</a>. <br>> >> For more options, visit <a href="https://groups.google.com/d/optout" rel="nofollow" target="_blank" onmousedown="this.href='https://groups.google.com/d/optout';return true;" onclick="this.href='https://groups.google.com/d/optout';return true;">https://groups.google.com/d/<wbr>optout</a>. <br>> > <br>> > <br>> > -- <br>> > You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. <br>> > To unsubscribe from this group and stop receiving emails from it, send an email to <a rel="nofollow">HomotopyTypeTheory+<wbr>unsubscribe@googlegroups.com</a>. <br>> > For more options, visit <a href="https://groups.google.com/d/optout" rel="nofollow" target="_blank" onmousedown="this.href='https://groups.google.com/d/optout';return true;" onclick="this.href='https://groups.google.com/d/optout';return true;">https://groups.google.com/d/<wbr>optout</a>. <br>> <br>> <br>> -- <br>> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. <br>> To unsubscribe from this group and stop receiving emails from it, send an email to <a rel="nofollow">HomotopyTypeTheory+<wbr>unsubscribe@googlegroups.com</a>. <br>> For more options, visit <a href="https://groups.google.com/d/optout" rel="nofollow" target="_blank" onmousedown="this.href='https://groups.google.com/d/optout';return true;" onclick="this.href='https://groups.google.com/d/optout';return true;">https://groups.google.com/d/<wbr>optout</a>. <br> <br></blockquote></div></blockquote></div> <p></p> -- <br /> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group.<br /> To unsubscribe from this group and stop receiving emails from it, send an email to <a href="mailto:HomotopyTypeTheory+unsubscribe@googlegroups.com">HomotopyTypeTheory+unsubscribe@googlegroups.com</a>.<br /> For more options, visit <a href="https://groups.google.com/d/optout">https://groups.google.com/d/optout</a>.<br />

In fact, in Cubical Agda you will not get these hcomp's with empty systems. The reason is that because of the way hcomp works in Agda there is a very nice trick to implement the "generalized hcomp" operation of the paper that Dan linked to. I summarized the trick in: https://github.com/agda/agda/issues/3415 I added this to Agda some month ago and it should be possible to update Simon's canonicity proof to get a stronger result saying that the only elements of HITs in the empty context are point constructors (like in the AFH paper). For this to work you also have to impose a "validity" constraint (Def 12 in the ccctt paper Dan linked to) so that empty systems cannot result from substitutions. This is currently not done in Cubical Agda, but if you start with a term with only valid systems then you should never get an empty system. So the extraction of witnesses from existence statements should work as Martín said in Cubical Agda. -- Anders On Thu, Mar 7, 2019 at 6:23 PM Martín Hötzel Escardó <escardo.martin@gmail.com> wrote: > > And this is a wildly speculative question. If we used Andrew Swan's identity type derived from the cubical path type only, as in the abstract library file https://github.com/agda/cubical/blob/master/Cubical/Core/HoTT-UF.agda) would we still get this phenomenon? Maybe not? What I mean is that we use normal Agda, together with what is offered in that file and nothing else (so that we are using HoTT book axiomatic type theory). Martin > > On Thursday, 7 March 2019 23:01:33 UTC, Martín Hötzel Escardó wrote: >> >> Oh, this is annoying, because it seems to mean that we would need unbounded search (to drop all "hcom []"'s) until we can read the |x,a|, which is against the spirit of, say, Martin-Loef type theories. Martin >> >> On Thursday, 7 March 2019 22:51:20 UTC, dlicata wrote: >>> >>> That would be true if the term you are normalizing is in the empty interval context, and the cubical type theory has “empty system regularity” (like https://www.cs.cmu.edu/~cangiuli/papers/ccctt.pdf). >>> >>> Otherwise, if you evaluate something in the empty interval context, you might see something like >>> hcom [] (hcom [] (hcom [] (hcom [] (… |x,a| … )))) >>> with |x,a| in there somewhere. In HITs, Kan composition is treated as a constructor of the type, and though there are no interesting lines to compose in the empty interval context, the uninteresting compositions don’t vanish in all flavors of cubical type theory. >>> >>> > On Mar 7, 2019, at 5:41 PM, Martín Hötzel Escardó <escardo...@gmail.com> wrote: >>> > >>> > So I presume that when we ask cubical Agda to normalize a term of type || Sigma (x:X), A x || we will get a term of the form |x,a| and so we will see the x in normal form, where |-| is the map into the truncation, right? Martin. >>> > >>> > On Thursday, 7 March 2019 21:52:12 UTC, Anders Mörtberg wrote: >>> > The existence property is proved for CCHM cubicaltt by Simon in: >>> > >>> > https://arxiv.org/abs/1607.04156 >>> > >>> > See corollary 5.2. This works a bit more generally than what Martín said, in particular in any context with only dimension variables we can compute a witness to an existence. So if in context G = i_1 : II, ..., i_n : II (possibly empty) we have: >>> > >>> > G |- t : exists (x : X), A(x) >>> > >>> > then we can compute G |- u : X so that G |- B(u). >>> > >>> > -- >>> > Anders >>> > >>> > On Thursday, March 7, 2019 at 11:16:48 AM UTC-5, Martín Hötzel Escardó wrote: >>> > I got confused now. :-) >>> > >>> > Seriously now, what you say seems related to the fact that from a proof |- t : || X || in the empty context, you get |- x : X in cubical type theory. This follows from Simon's canonicity result (at least for X=natural numbers), and is like the so-called "existence property" in the internal language of the free elementary topos. This says that from a proof |- exists (x:X), A x in the empty context, you get |- x : X and |- A x. This says that exists in the empty context behaves like Sigma. But only in the empty context, because otherwise it behaves like "local existence" as in Kripke-Joyal semantics. >>> > >>> > Martin >>> > >>> > On Thursday, 7 March 2019 14:10:56 UTC, dlicata wrote: >>> > Just in case anyone reading this thread later is confused about a more beginner point than the ones Nicolai and Martin made, one possible stumbling block here is that, if someone means “is inhabited” in an external sense (there is a closed term of that type), then the answer is yes (at least in some models): if ||A|| is inhabited then A is inhabited. For example, in cubical models with canonicity, it is true that a closed term of type ||A|| evaluates to a value that has as a subterm a closed term of type A (the other values of ||A|| are some “formal compositions” of values of ||A||, but there has to be an |a| in there at the base case). This is consistent with what Martin and Nicolai said because “if A is inhabited then B is inhabited” (in this external sense) doesn’t necessarily mean there is a map A -> B internally. >>> > >>> > -Dan >>> > >>> > > On Mar 5, 2019, at 6:07 PM, Martín Hötzel Escardó <escardo...@gmail.com> wrote: >>> > > >>> > > Or you can read the paper https://lmcs.episciences.org/3217/ regarding what Nicolai said. >>> > > >>> > > Moreover, in the HoTT book, it is shown that if || X||->X holds for all X, then univalence can't hold. (It is global choice, which can't be invariant under equivalence.) >>> > > >>> > > The above paper shows that unrestricted ||X||->X it gives excluded middle. >>> > > >>> > > However, for a lot of kinds of types one can show that ||X||->X does hold. For example, if they have a constant endo-function. Moreover, for any type X, the availability of ||X||->X is logically equivalent to the availability of a constant map X->X (before we know whether X has a point or not, in which case the availability of a constant endo-map is trivial). >>> > > >>> > > Martin >>> > > >>> > > On Tuesday, 5 March 2019 22:47:55 UTC, Nicolai Kraus wrote: >>> > > You can't have a function which, for all A, gives you ||A|| -> A. See the exercises 3.11 and 3.12! >>> > > -- Nicolai >>> > > >>> > > On 05/03/19 22:31, Jean Joseph wrote: >>> > >> Hi, >>> > >> >>> > >> From the HoTT book, the truncation of any type A has two constructors: >>> > >> >>> > >> 1) for any a : A, there is |a| : ||A|| >>> > >> 2) for any x,y : ||A||, x = y. >>> > >> >>> > >> I get that if A is inhabited, then ||A|| is inhabited by (1). But is it true that, if ||A|| is inhabited, then A is inhabited? >>> > >> -- >>> > >> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. >>> > >> To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. >>> > >> For more options, visit https://groups.google.com/d/optout. >>> > > >>> > > >>> > > -- >>> > > You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. >>> > > To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. >>> > > For more options, visit https://groups.google.com/d/optout. >>> > >>> > >>> > -- >>> > You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. >>> > To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. >>> > For more options, visit https://groups.google.com/d/optout. >>> > -- > You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. > To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. > For more options, visit https://groups.google.com/d/optout. -- You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. For more options, visit https://groups.google.com/d/optout.

nice! the negation trick is clever. Martin, I don’t understand why this situation is any different than natural numbers, though. If I have a closed term of type nat, I can normalize it, but then I (externally) need to read through all of the successors to figure out what number it is. (Or maybe I can only weak head normalize it, in which case I need to interleave further weak head normalization after every successor.) Is this an unbounded search? The parallel is to read “hcom []” as successor and “|x,a|” as zero. -Dan > On Mar 8, 2019, at 9:59 AM, Anders Mortberg <andersmortberg@gmail.com> wrote: > > In fact, in Cubical Agda you will not get these hcomp's with empty > systems. The reason is that because of the way hcomp works in Agda > there is a very nice trick to implement the "generalized hcomp" > operation of the paper that Dan linked to. I summarized the trick in: > > https://github.com/agda/agda/issues/3415 > > I added this to Agda some month ago and it should be possible to > update Simon's canonicity proof to get a stronger result saying that > the only elements of HITs in the empty context are point constructors > (like in the AFH paper). For this to work you also have to impose a > "validity" constraint (Def 12 in the ccctt paper Dan linked to) so > that empty systems cannot result from substitutions. This is currently > not done in Cubical Agda, but if you start with a term with only valid > systems then you should never get an empty system. > > So the extraction of witnesses from existence statements should work > as Martín said in Cubical Agda. > > -- > Anders > > On Thu, Mar 7, 2019 at 6:23 PM Martín Hötzel Escardó > <escardo.martin@gmail.com> wrote: >> >> And this is a wildly speculative question. If we used Andrew Swan's identity type derived from the cubical path type only, as in the abstract library file https://github.com/agda/cubical/blob/master/Cubical/Core/HoTT-UF.agda) would we still get this phenomenon? Maybe not? What I mean is that we use normal Agda, together with what is offered in that file and nothing else (so that we are using HoTT book axiomatic type theory). Martin >> >> On Thursday, 7 March 2019 23:01:33 UTC, Martín Hötzel Escardó wrote: >>> >>> Oh, this is annoying, because it seems to mean that we would need unbounded search (to drop all "hcom []"'s) until we can read the |x,a|, which is against the spirit of, say, Martin-Loef type theories. Martin >>> >>> On Thursday, 7 March 2019 22:51:20 UTC, dlicata wrote: >>>> >>>> That would be true if the term you are normalizing is in the empty interval context, and the cubical type theory has “empty system regularity” (like https://www.cs.cmu.edu/~cangiuli/papers/ccctt.pdf). >>>> >>>> Otherwise, if you evaluate something in the empty interval context, you might see something like >>>> hcom [] (hcom [] (hcom [] (hcom [] (… |x,a| … )))) >>>> with |x,a| in there somewhere. In HITs, Kan composition is treated as a constructor of the type, and though there are no interesting lines to compose in the empty interval context, the uninteresting compositions don’t vanish in all flavors of cubical type theory. >>>> >>>>> On Mar 7, 2019, at 5:41 PM, Martín Hötzel Escardó <escardo...@gmail.com> wrote: >>>>> >>>>> So I presume that when we ask cubical Agda to normalize a term of type || Sigma (x:X), A x || we will get a term of the form |x,a| and so we will see the x in normal form, where |-| is the map into the truncation, right? Martin. >>>>> >>>>> On Thursday, 7 March 2019 21:52:12 UTC, Anders Mörtberg wrote: >>>>> The existence property is proved for CCHM cubicaltt by Simon in: >>>>> >>>>> https://arxiv.org/abs/1607.04156 >>>>> >>>>> See corollary 5.2. This works a bit more generally than what Martín said, in particular in any context with only dimension variables we can compute a witness to an existence. So if in context G = i_1 : II, ..., i_n : II (possibly empty) we have: >>>>> >>>>> G |- t : exists (x : X), A(x) >>>>> >>>>> then we can compute G |- u : X so that G |- B(u). >>>>> >>>>> -- >>>>> Anders >>>>> >>>>> On Thursday, March 7, 2019 at 11:16:48 AM UTC-5, Martín Hötzel Escardó wrote: >>>>> I got confused now. :-) >>>>> >>>>> Seriously now, what you say seems related to the fact that from a proof |- t : || X || in the empty context, you get |- x : X in cubical type theory. This follows from Simon's canonicity result (at least for X=natural numbers), and is like the so-called "existence property" in the internal language of the free elementary topos. This says that from a proof |- exists (x:X), A x in the empty context, you get |- x : X and |- A x. This says that exists in the empty context behaves like Sigma. But only in the empty context, because otherwise it behaves like "local existence" as in Kripke-Joyal semantics. >>>>> >>>>> Martin >>>>> >>>>> On Thursday, 7 March 2019 14:10:56 UTC, dlicata wrote: >>>>> Just in case anyone reading this thread later is confused about a more beginner point than the ones Nicolai and Martin made, one possible stumbling block here is that, if someone means “is inhabited” in an external sense (there is a closed term of that type), then the answer is yes (at least in some models): if ||A|| is inhabited then A is inhabited. For example, in cubical models with canonicity, it is true that a closed term of type ||A|| evaluates to a value that has as a subterm a closed term of type A (the other values of ||A|| are some “formal compositions” of values of ||A||, but there has to be an |a| in there at the base case). This is consistent with what Martin and Nicolai said because “if A is inhabited then B is inhabited” (in this external sense) doesn’t necessarily mean there is a map A -> B internally. >>>>> >>>>> -Dan >>>>> >>>>>> On Mar 5, 2019, at 6:07 PM, Martín Hötzel Escardó <escardo...@gmail.com> wrote: >>>>>> >>>>>> Or you can read the paper https://lmcs.episciences.org/3217/ regarding what Nicolai said. >>>>>> >>>>>> Moreover, in the HoTT book, it is shown that if || X||->X holds for all X, then univalence can't hold. (It is global choice, which can't be invariant under equivalence.) >>>>>> >>>>>> The above paper shows that unrestricted ||X||->X it gives excluded middle. >>>>>> >>>>>> However, for a lot of kinds of types one can show that ||X||->X does hold. For example, if they have a constant endo-function. Moreover, for any type X, the availability of ||X||->X is logically equivalent to the availability of a constant map X->X (before we know whether X has a point or not, in which case the availability of a constant endo-map is trivial). >>>>>> >>>>>> Martin >>>>>> >>>>>> On Tuesday, 5 March 2019 22:47:55 UTC, Nicolai Kraus wrote: >>>>>> You can't have a function which, for all A, gives you ||A|| -> A. See the exercises 3.11 and 3.12! >>>>>> -- Nicolai >>>>>> >>>>>> On 05/03/19 22:31, Jean Joseph wrote: >>>>>>> Hi, >>>>>>> >>>>>>> From the HoTT book, the truncation of any type A has two constructors: >>>>>>> >>>>>>> 1) for any a : A, there is |a| : ||A|| >>>>>>> 2) for any x,y : ||A||, x = y. >>>>>>> >>>>>>> I get that if A is inhabited, then ||A|| is inhabited by (1). But is it true that, if ||A|| is inhabited, then A is inhabited? >>>>>>> -- >>>>>>> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. >>>>>>> To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. >>>>>>> For more options, visit https://groups.google.com/d/optout. >>>>>> >>>>>> >>>>>> -- >>>>>> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. >>>>>> To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. >>>>>> For more options, visit https://groups.google.com/d/optout. >>>>> >>>>> >>>>> -- >>>>> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. >>>>> To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. >>>>> For more options, visit https://groups.google.com/d/optout. >>>> >> -- >> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. >> To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. >> For more options, visit https://groups.google.com/d/optout. -- You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. For more options, visit https://groups.google.com/d/optout.

[-- Attachment #1.1: Type: text/plain, Size: 10441 bytes --] On Friday, 8 March 2019 15:13:14 UTC, dlicata wrote: > > nice! the negation trick is clever. > Yes, this is really nice. Martin, I don’t understand why this situation is any different than natural > numbers, though. If I have a closed term of type nat, I can normalize it, > but then I (externally) need to read through all of the successors to > figure out what number it is. (Or maybe I can only weak head normalize it, > in which case I need to interleave further weak head normalization after > every successor.) Is this an unbounded search? The parallel is to read > “hcom []” as successor and “|x,a|” as zero. > > Except that you would have different normal forms for the same thing, and this thing would be prefixed by an unbounded number of *printed* clock ticks. It is not right to say that unbounded search is needed, I agree. But it wouldn't be pleasant to have to see 10^6 ticks. I would rather wait for 10^6 units of time (which should be fast) and then see the answer than see a million hcomp's followed by the answer on my computer screen . :-) Fortunately, Anders says that the clock ticks are not printed in the version of cubical type theory implemented by Agda --cubical, which is nice! Martin > -Dan > > > On Mar 8, 2019, at 9:59 AM, Anders Mortberg <andersm...@gmail.com > <javascript:>> wrote: > > > > In fact, in Cubical Agda you will not get these hcomp's with empty > > systems. The reason is that because of the way hcomp works in Agda > > there is a very nice trick to implement the "generalized hcomp" > > operation of the paper that Dan linked to. I summarized the trick in: > > > > https://github.com/agda/agda/issues/3415 > > > > I added this to Agda some month ago and it should be possible to > > update Simon's canonicity proof to get a stronger result saying that > > the only elements of HITs in the empty context are point constructors > > (like in the AFH paper). For this to work you also have to impose a > > "validity" constraint (Def 12 in the ccctt paper Dan linked to) so > > that empty systems cannot result from substitutions. This is currently > > not done in Cubical Agda, but if you start with a term with only valid > > systems then you should never get an empty system. > > > > So the extraction of witnesses from existence statements should work > > as Martín said in Cubical Agda. > > > > -- > > Anders > > > > On Thu, Mar 7, 2019 at 6:23 PM Martín Hötzel Escardó > > <escardo...@gmail.com <javascript:>> wrote: > >> > >> And this is a wildly speculative question. If we used Andrew Swan's > identity type derived from the cubical path type only, as in the abstract > library file > https://github.com/agda/cubical/blob/master/Cubical/Core/HoTT-UF.agda) > would we still get this phenomenon? Maybe not? What I mean is that we use > normal Agda, together with what is offered in that file and nothing else > (so that we are using HoTT book axiomatic type theory). Martin > >> > >> On Thursday, 7 March 2019 23:01:33 UTC, Martín Hötzel Escardó wrote: > >>> > >>> Oh, this is annoying, because it seems to mean that we would need > unbounded search (to drop all "hcom []"'s) until we can read the |x,a|, > which is against the spirit of, say, Martin-Loef type theories. Martin > >>> > >>> On Thursday, 7 March 2019 22:51:20 UTC, dlicata wrote: > >>>> > >>>> That would be true if the term you are normalizing is in the empty > interval context, and the cubical type theory has “empty system regularity” > (like https://www.cs.cmu.edu/~cangiuli/papers/ccctt.pdf). > >>>> > >>>> Otherwise, if you evaluate something in the empty interval context, > you might see something like > >>>> hcom [] (hcom [] (hcom [] (hcom [] (… |x,a| … )))) > >>>> with |x,a| in there somewhere. In HITs, Kan composition is treated > as a constructor of the type, and though there are no interesting lines to > compose in the empty interval context, the uninteresting compositions don’t > vanish in all flavors of cubical type theory. > >>>> > >>>>> On Mar 7, 2019, at 5:41 PM, Martín Hötzel Escardó < > escardo...@gmail.com> wrote: > >>>>> > >>>>> So I presume that when we ask cubical Agda to normalize a term of > type || Sigma (x:X), A x || we will get a term of the form |x,a| and so we > will see the x in normal form, where |-| is the map into the truncation, > right? Martin. > >>>>> > >>>>> On Thursday, 7 March 2019 21:52:12 UTC, Anders Mörtberg wrote: > >>>>> The existence property is proved for CCHM cubicaltt by Simon in: > >>>>> > >>>>> https://arxiv.org/abs/1607.04156 > >>>>> > >>>>> See corollary 5.2. This works a bit more generally than what Martín > said, in particular in any context with only dimension variables we can > compute a witness to an existence. So if in context G = i_1 : II, ..., i_n > : II (possibly empty) we have: > >>>>> > >>>>> G |- t : exists (x : X), A(x) > >>>>> > >>>>> then we can compute G |- u : X so that G |- B(u). > >>>>> > >>>>> -- > >>>>> Anders > >>>>> > >>>>> On Thursday, March 7, 2019 at 11:16:48 AM UTC-5, Martín Hötzel > Escardó wrote: > >>>>> I got confused now. :-) > >>>>> > >>>>> Seriously now, what you say seems related to the fact that from a > proof |- t : || X || in the empty context, you get |- x : X in cubical type > theory. This follows from Simon's canonicity result (at least for X=natural > numbers), and is like the so-called "existence property" in the internal > language of the free elementary topos. This says that from a proof |- > exists (x:X), A x in the empty context, you get |- x : X and |- A x. This > says that exists in the empty context behaves like Sigma. But only in the > empty context, because otherwise it behaves like "local existence" as in > Kripke-Joyal semantics. > >>>>> > >>>>> Martin > >>>>> > >>>>> On Thursday, 7 March 2019 14:10:56 UTC, dlicata wrote: > >>>>> Just in case anyone reading this thread later is confused about a > more beginner point than the ones Nicolai and Martin made, one possible > stumbling block here is that, if someone means “is inhabited” in an > external sense (there is a closed term of that type), then the answer is > yes (at least in some models): if ||A|| is inhabited then A is inhabited. > For example, in cubical models with canonicity, it is true that a closed > term of type ||A|| evaluates to a value that has as a subterm a closed term > of type A (the other values of ||A|| are some “formal compositions” of > values of ||A||, but there has to be an |a| in there at the base case). > This is consistent with what Martin and Nicolai said because “if A is > inhabited then B is inhabited” (in this external sense) doesn’t necessarily > mean there is a map A -> B internally. > >>>>> > >>>>> -Dan > >>>>> > >>>>>> On Mar 5, 2019, at 6:07 PM, Martín Hötzel Escardó < > escardo...@gmail.com> wrote: > >>>>>> > >>>>>> Or you can read the paper https://lmcs.episciences.org/3217/ > regarding what Nicolai said. > >>>>>> > >>>>>> Moreover, in the HoTT book, it is shown that if || X||->X holds for > all X, then univalence can't hold. (It is global choice, which can't be > invariant under equivalence.) > >>>>>> > >>>>>> The above paper shows that unrestricted ||X||->X it gives excluded > middle. > >>>>>> > >>>>>> However, for a lot of kinds of types one can show that ||X||->X > does hold. For example, if they have a constant endo-function. Moreover, > for any type X, the availability of ||X||->X is logically equivalent to the > availability of a constant map X->X (before we know whether X has a point > or not, in which case the availability of a constant endo-map is trivial). > >>>>>> > >>>>>> Martin > >>>>>> > >>>>>> On Tuesday, 5 March 2019 22:47:55 UTC, Nicolai Kraus wrote: > >>>>>> You can't have a function which, for all A, gives you ||A|| -> A. > See the exercises 3.11 and 3.12! > >>>>>> -- Nicolai > >>>>>> > >>>>>> On 05/03/19 22:31, Jean Joseph wrote: > >>>>>>> Hi, > >>>>>>> > >>>>>>> From the HoTT book, the truncation of any type A has two > constructors: > >>>>>>> > >>>>>>> 1) for any a : A, there is |a| : ||A|| > >>>>>>> 2) for any x,y : ||A||, x = y. > >>>>>>> > >>>>>>> I get that if A is inhabited, then ||A|| is inhabited by (1). But > is it true that, if ||A|| is inhabited, then A is inhabited? > >>>>>>> -- > >>>>>>> You received this message because you are subscribed to the Google > Groups "Homotopy Type Theory" group. > >>>>>>> To unsubscribe from this group and stop receiving emails from it, > send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com > <javascript:>. > >>>>>>> For more options, visit https://groups.google.com/d/optout. > >>>>>> > >>>>>> > >>>>>> -- > >>>>>> You received this message because you are subscribed to the Google > Groups "Homotopy Type Theory" group. > >>>>>> To unsubscribe from this group and stop receiving emails from it, > send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com > <javascript:>. > >>>>>> For more options, visit https://groups.google.com/d/optout. > >>>>> > >>>>> > >>>>> -- > >>>>> You received this message because you are subscribed to the Google > Groups "Homotopy Type Theory" group. > >>>>> To unsubscribe from this group and stop receiving emails from it, > send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com > <javascript:>. > >>>>> For more options, visit https://groups.google.com/d/optout. > >>>> > >> -- > >> You received this message because you are subscribed to the Google > Groups "Homotopy Type Theory" group. > >> To unsubscribe from this group and stop receiving emails from it, send > an email to HomotopyTypeTheory+unsubscribe@googlegroups.com <javascript:>. > > >> For more options, visit https://groups.google.com/d/optout. > > -- You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. To unsubscribe from this group and stop receiving emails from it, send an email to HomotopyTypeTheory+unsubscribe@googlegroups.com. For more options, visit https://groups.google.com/d/optout. [-- Attachment #1.2: Type: text/html, Size: 17374 bytes --] <div dir="ltr"><br><br>On Friday, 8 March 2019 15:13:14 UTC, dlicata wrote:<blockquote class="gmail_quote" style="margin: 0;margin-left: 0.8ex;border-left: 1px #ccc solid;padding-left: 1ex;">nice! the negation trick is clever. <br></blockquote><div><br></div><div>Yes, this is really nice.</div><div><br></div><blockquote class="gmail_quote" style="margin: 0;margin-left: 0.8ex;border-left: 1px #ccc solid;padding-left: 1ex;">Martin, I don’t understand why this situation is any different than natural numbers, though. If I have a closed term of type nat, I can normalize it, but then I (externally) need to read through all of the successors to figure out what number it is. (Or maybe I can only weak head normalize it, in which case I need to interleave further weak head normalization after every successor.) Is this an unbounded search? The parallel is to read “hcom []” as successor and “|x,a|” as zero. <br> <br></blockquote><div><br></div><div>Except that you would have different normal forms for the same thing, and this thing would be prefixed by an unbounded number of *printed* clock ticks. </div><div><br></div><div>It is not right to say that unbounded search is needed, I agree. But it wouldn't be pleasant to have to see 10^6 ticks. I would rather wait for 10^6 units of time (which should be fast) and then see the answer than see a million hcomp's followed by the answer on my computer screen . :-)</div><div><br></div><div>Fortunately, Anders says that the clock ticks are not printed in the version of cubical type theory implemented by Agda --cubical, which is nice!</div><div><br></div><div>Martin</div><div> </div><blockquote class="gmail_quote" style="margin: 0;margin-left: 0.8ex;border-left: 1px #ccc solid;padding-left: 1ex;">-Dan <br> <br>> On Mar 8, 2019, at 9:59 AM, Anders Mortberg <<a href="javascript:" target="_blank" gdf-obfuscated-mailto="AOed4TRrAwAJ" rel="nofollow" onmousedown="this.href='javascript:';return true;" onclick="this.href='javascript:';return true;">andersm...@gmail.com</a>> wrote: <br>> <br>> In fact, in Cubical Agda you will not get these hcomp's with empty <br>> systems. The reason is that because of the way hcomp works in Agda <br>> there is a very nice trick to implement the "generalized hcomp" <br>> operation of the paper that Dan linked to. I summarized the trick in: <br>> <br>> <a href="https://github.com/agda/agda/issues/3415" target="_blank" rel="nofollow" onmousedown="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Fgithub.com%2Fagda%2Fagda%2Fissues%2F3415\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNE6Vz_70NQLJ7oGpThpsHf8BChIeA';return true;" onclick="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Fgithub.com%2Fagda%2Fagda%2Fissues%2F3415\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNE6Vz_70NQLJ7oGpThpsHf8BChIeA';return true;">https://github.com/agda/agda/<wbr>issues/3415</a> <br>> <br>> I added this to Agda some month ago and it should be possible to <br>> update Simon's canonicity proof to get a stronger result saying that <br>> the only elements of HITs in the empty context are point constructors <br>> (like in the AFH paper). For this to work you also have to impose a <br>> "validity" constraint (Def 12 in the ccctt paper Dan linked to) so <br>> that empty systems cannot result from substitutions. This is currently <br>> not done in Cubical Agda, but if you start with a term with only valid <br>> systems then you should never get an empty system. <br>> <br>> So the extraction of witnesses from existence statements should work <br>> as Martín said in Cubical Agda. <br>> <br>> -- <br>> Anders <br>> <br>> On Thu, Mar 7, 2019 at 6:23 PM Martín Hötzel Escardó <br>> <<a href="javascript:" target="_blank" gdf-obfuscated-mailto="AOed4TRrAwAJ" rel="nofollow" onmousedown="this.href='javascript:';return true;" onclick="this.href='javascript:';return true;">escardo...@gmail.com</a>> wrote: <br>>> <br>>> And this is a wildly speculative question. If we used Andrew Swan's identity type derived from the cubical path type only, as in the abstract library file <a href="https://github.com/agda/cubical/blob/master/Cubical/Core/HoTT-UF.agda" target="_blank" rel="nofollow" onmousedown="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Fgithub.com%2Fagda%2Fcubical%2Fblob%2Fmaster%2FCubical%2FCore%2FHoTT-UF.agda\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNGNwBJEl32G4Bh4OW2St9OLKMIyUA';return true;" onclick="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Fgithub.com%2Fagda%2Fcubical%2Fblob%2Fmaster%2FCubical%2FCore%2FHoTT-UF.agda\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNGNwBJEl32G4Bh4OW2St9OLKMIyUA';return true;">https://github.com/agda/<wbr>cubical/blob/master/Cubical/<wbr>Core/HoTT-UF.agda</a>) would we still get this phenomenon? Maybe not? What I mean is that we use normal Agda, together with what is offered in that file and nothing else (so that we are using HoTT book axiomatic type theory). Martin <br>>> <br>>> On Thursday, 7 March 2019 23:01:33 UTC, Martín Hötzel Escardó wrote: <br>>>> <br>>>> Oh, this is annoying, because it seems to mean that we would need unbounded search (to drop all "hcom []"'s) until we can read the |x,a|, which is against the spirit of, say, Martin-Loef type theories. Martin <br>>>> <br>>>> On Thursday, 7 March 2019 22:51:20 UTC, dlicata wrote: <br>>>>> <br>>>>> That would be true if the term you are normalizing is in the empty interval context, and the cubical type theory has “empty system regularity” (like <a href="https://www.cs.cmu.edu/~cangiuli/papers/ccctt.pdf" target="_blank" rel="nofollow" onmousedown="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Fwww.cs.cmu.edu%2F~cangiuli%2Fpapers%2Fccctt.pdf\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNESV2Pele9DWGVlPQp1lf9qMnycyw';return true;" onclick="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Fwww.cs.cmu.edu%2F~cangiuli%2Fpapers%2Fccctt.pdf\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNESV2Pele9DWGVlPQp1lf9qMnycyw';return true;">https://www.cs.cmu.edu/~<wbr>cangiuli/papers/ccctt.pdf</a>). <br>>>>> <br>>>>> Otherwise, if you evaluate something in the empty interval context, you might see something like <br>>>>> hcom [] (hcom [] (hcom [] (hcom [] (… |x,a| … )))) <br>>>>> with |x,a| in there somewhere. In HITs, Kan composition is treated as a constructor of the type, and though there are no interesting lines to compose in the empty interval context, the uninteresting compositions don’t vanish in all flavors of cubical type theory. <br>>>>> <br>>>>>> On Mar 7, 2019, at 5:41 PM, Martín Hötzel Escardó <<a>escardo...@gmail.com</a>> wrote: <br>>>>>> <br>>>>>> So I presume that when we ask cubical Agda to normalize a term of type || Sigma (x:X), A x || we will get a term of the form |x,a| and so we will see the x in normal form, where |-| is the map into the truncation, right? Martin. <br>>>>>> <br>>>>>> On Thursday, 7 March 2019 21:52:12 UTC, Anders Mörtberg wrote: <br>>>>>> The existence property is proved for CCHM cubicaltt by Simon in: <br>>>>>> <br>>>>>> <a href="https://arxiv.org/abs/1607.04156" target="_blank" rel="nofollow" onmousedown="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Farxiv.org%2Fabs%2F1607.04156\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNHaiqP0--DZhiZa0817yu3ILyH9TQ';return true;" onclick="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Farxiv.org%2Fabs%2F1607.04156\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNHaiqP0--DZhiZa0817yu3ILyH9TQ';return true;">https://arxiv.org/abs/1607.<wbr>04156</a> <br>>>>>> <br>>>>>> See corollary 5.2. This works a bit more generally than what Martín said, in particular in any context with only dimension variables we can compute a witness to an existence. So if in context G = i_1 : II, ..., i_n : II (possibly empty) we have: <br>>>>>> <br>>>>>> G |- t : exists (x : X), A(x) <br>>>>>> <br>>>>>> then we can compute G |- u : X so that G |- B(u). <br>>>>>> <br>>>>>> -- <br>>>>>> Anders <br>>>>>> <br>>>>>> On Thursday, March 7, 2019 at 11:16:48 AM UTC-5, Martín Hötzel Escardó wrote: <br>>>>>> I got confused now. :-) <br>>>>>> <br>>>>>> Seriously now, what you say seems related to the fact that from a proof |- t : || X || in the empty context, you get |- x : X in cubical type theory. This follows from Simon's canonicity result (at least for X=natural numbers), and is like the so-called "existence property" in the internal language of the free elementary topos. This says that from a proof |- exists (x:X), A x in the empty context, you get |- x : X and |- A x. This says that exists in the empty context behaves like Sigma. But only in the empty context, because otherwise it behaves like "local existence" as in Kripke-Joyal semantics. <br>>>>>> <br>>>>>> Martin <br>>>>>> <br>>>>>> On Thursday, 7 March 2019 14:10:56 UTC, dlicata wrote: <br>>>>>> Just in case anyone reading this thread later is confused about a more beginner point than the ones Nicolai and Martin made, one possible stumbling block here is that, if someone means “is inhabited” in an external sense (there is a closed term of that type), then the answer is yes (at least in some models): if ||A|| is inhabited then A is inhabited. For example, in cubical models with canonicity, it is true that a closed term of type ||A|| evaluates to a value that has as a subterm a closed term of type A (the other values of ||A|| are some “formal compositions” of values of ||A||, but there has to be an |a| in there at the base case). This is consistent with what Martin and Nicolai said because “if A is inhabited then B is inhabited” (in this external sense) doesn’t necessarily mean there is a map A -> B internally. <br>>>>>> <br>>>>>> -Dan <br>>>>>> <br>>>>>>> On Mar 5, 2019, at 6:07 PM, Martín Hötzel Escardó <<a>escardo...@gmail.com</a>> wrote: <br>>>>>>> <br>>>>>>> Or you can read the paper <a href="https://lmcs.episciences.org/3217/" target="_blank" rel="nofollow" onmousedown="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Flmcs.episciences.org%2F3217%2F\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNF2sTOfc7ovV50gWLoaFL_7S1LECg';return true;" onclick="this.href='https://www.google.com/url?q\x3dhttps%3A%2F%2Flmcs.episciences.org%2F3217%2F\x26sa\x3dD\x26sntz\x3d1\x26usg\x3dAFQjCNF2sTOfc7ovV50gWLoaFL_7S1LECg';return true;">https://lmcs.episciences.org/<wbr>3217/</a> regarding what Nicolai said. <br>>>>>>> <br>>>>>>> Moreover, in the HoTT book, it is shown that if || X||->X holds for all X, then univalence can't hold. (It is global choice, which can't be invariant under equivalence.) <br>>>>>>> <br>>>>>>> The above paper shows that unrestricted ||X||->X it gives excluded middle. <br>>>>>>> <br>>>>>>> However, for a lot of kinds of types one can show that ||X||->X does hold. For example, if they have a constant endo-function. Moreover, for any type X, the availability of ||X||->X is logically equivalent to the availability of a constant map X->X (before we know whether X has a point or not, in which case the availability of a constant endo-map is trivial). <br>>>>>>> <br>>>>>>> Martin <br>>>>>>> <br>>>>>>> On Tuesday, 5 March 2019 22:47:55 UTC, Nicolai Kraus wrote: <br>>>>>>> You can't have a function which, for all A, gives you ||A|| -> A. See the exercises 3.11 and 3.12! <br>>>>>>> -- Nicolai <br>>>>>>> <br>>>>>>> On 05/03/19 22:31, Jean Joseph wrote: <br>>>>>>>> Hi, <br>>>>>>>> <br>>>>>>>> From the HoTT book, the truncation of any type A has two constructors: <br>>>>>>>> <br>>>>>>>> 1) for any a : A, there is |a| : ||A|| <br>>>>>>>> 2) for any x,y : ||A||, x = y. <br>>>>>>>> <br>>>>>>>> I get that if A is inhabited, then ||A|| is inhabited by (1). But is it true that, if ||A|| is inhabited, then A is inhabited? <br>>>>>>>> -- <br>>>>>>>> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. <br>>>>>>>> To unsubscribe from this group and stop receiving emails from it, send an email to <a href="javascript:" target="_blank" gdf-obfuscated-mailto="AOed4TRrAwAJ" rel="nofollow" onmousedown="this.href='javascript:';return true;" onclick="this.href='javascript:';return true;">HomotopyTypeTheory+<wbr>unsubscribe@googlegroups.com</a>. <br>>>>>>>> For more options, visit <a href="https://groups.google.com/d/optout" target="_blank" rel="nofollow" onmousedown="this.href='https://groups.google.com/d/optout';return true;" onclick="this.href='https://groups.google.com/d/optout';return true;">https://groups.google.com/d/<wbr>optout</a>. <br>>>>>>> <br>>>>>>> <br>>>>>>> -- <br>>>>>>> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. <br>>>>>>> To unsubscribe from this group and stop receiving emails from it, send an email to <a href="javascript:" target="_blank" gdf-obfuscated-mailto="AOed4TRrAwAJ" rel="nofollow" onmousedown="this.href='javascript:';return true;" onclick="this.href='javascript:';return true;">HomotopyTypeTheory+<wbr>unsubscribe@googlegroups.com</a>. <br>>>>>>> For more options, visit <a href="https://groups.google.com/d/optout" target="_blank" rel="nofollow" onmousedown="this.href='https://groups.google.com/d/optout';return true;" onclick="this.href='https://groups.google.com/d/optout';return true;">https://groups.google.com/d/<wbr>optout</a>. <br>>>>>> <br>>>>>> <br>>>>>> -- <br>>>>>> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. <br>>>>>> To unsubscribe from this group and stop receiving emails from it, send an email to <a href="javascript:" target="_blank" gdf-obfuscated-mailto="AOed4TRrAwAJ" rel="nofollow" onmousedown="this.href='javascript:';return true;" onclick="this.href='javascript:';return true;">HomotopyTypeTheory+<wbr>unsubscribe@googlegroups.com</a>. <br>>>>>> For more options, visit <a href="https://groups.google.com/d/optout" target="_blank" rel="nofollow" onmousedown="this.href='https://groups.google.com/d/optout';return true;" onclick="this.href='https://groups.google.com/d/optout';return true;">https://groups.google.com/d/<wbr>optout</a>. <br>>>>> <br>>> -- <br>>> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group. <br>>> To unsubscribe from this group and stop receiving emails from it, send an email to <a href="javascript:" target="_blank" gdf-obfuscated-mailto="AOed4TRrAwAJ" rel="nofollow" onmousedown="this.href='javascript:';return true;" onclick="this.href='javascript:';return true;">HomotopyTypeTheory+<wbr>unsubscribe@googlegroups.com</a>. <br>>> For more options, visit <a href="https://groups.google.com/d/optout" target="_blank" rel="nofollow" onmousedown="this.href='https://groups.google.com/d/optout';return true;" onclick="this.href='https://groups.google.com/d/optout';return true;">https://groups.google.com/d/<wbr>optout</a>. <br> <br></blockquote></div> <p></p> -- <br /> You received this message because you are subscribed to the Google Groups "Homotopy Type Theory" group.<br /> To unsubscribe from this group and stop receiving emails from it, send an email to <a href="mailto:HomotopyTypeTheory+unsubscribe@googlegroups.com">HomotopyTypeTheory+unsubscribe@googlegroups.com</a>.<br /> For more options, visit <a href="https://groups.google.com/d/optout">https://groups.google.com/d/optout</a>.<br />