Vertical alignment of formulas

Vertical alignment of formulas

[Emil Hedevang Lohse]

Hello,

Suppose I have the following code:

  blah blah blah blah blah blah blah blah blah

  \startmpformula
    formula
  \stopmpformula

and that it yields the following output

  blah blah blah blah blah blah blah 
  blah blah 
  
               formula

that is, the final "blah" is placed so far to the left of "formula"
that it would be desireable to reduce the vertical space between the
text and the formula such that we get the following output instead:

  blah blah blah blah blah blah blah 
  blah blah 
               formula

The question is now: How do I do that?

Regards,

-- 
Emil Hedevang Lohse <http://home.imf.au.dk/emil/> 

Alle spørgsmål er lige dumme. 
Og spørgsmålet "Kan ænder flyve?" er ikke dumt.

Re: Vertical alignment of formulas

[Giuseppe Bilotta]

Tuesday, September 16, 2003 Emil Hedevang Lohse wrote:


> Hello,

> Suppose I have the following code:

>   blah blah blah blah blah blah blah blah blah

>   \startmpformula
>     formula
>   \stopmpformula

> and that it yields the following output

>   blah blah blah blah blah blah blah 
>   blah blah 
  
>                formula

> that is, the final "blah" is placed so far to the left of "formula"
> that it would be desireable to reduce the vertical space between the
> text and the formula such that we get the following output instead:

>   blah blah blah blah blah blah blah 
>   blah blah 
>                formula

> The question is now: How do I do that?

You don't :\ Sadly, Hans decided that the difference between
abovedisplayskip and abovedisplayshortskip was unnecessary.
Hans?

-- 
Giuseppe "Oblomov" Bilotta

Re: Vertical alignment of formulas

[Hans Hagen]

At 22:32 16/09/2003 +0200, you wrote:

>The question is now: How do I do that?

i'll have a look at it,

one complication is that the last line width is not available outside math 
mode (which is a pain)

Hans
-------------------------------------------------------------------------
                                   Hans Hagen | PRAGMA ADE | pragma@wxs.nl
                       Ridderstraat 27 | 8061 GH Hasselt | The Netherlands
  tel: +31 (0)38 477 53 69 | fax: +31 (0)38 477 53 74 | www.pragma-ade.com
-------------------------------------------------------------------------
                        information: http://www.pragma-ade.com/roadmap.pdf
                     documentation: http://www.pragma-ade.com/showcase.pdf
-------------------------------------------------------------------------

Re: Vertical alignment of formulas

[Hans Hagen]

At 17:59 17/09/2003 +0200, you wrote:

>You don't :\ Sadly, Hans decided that the difference between
>abovedisplayskip and abovedisplayshortskip was unnecessary.
>Hans?

that's not true -)

the problem is that it is quite hard to get the spacing ok in situations like

text

formula

text

this has to do with the fact (and i spent days on that topic) that tex only 
knows the length of the last line when in display math mode and even then 
not always

this means that when one wants control over spacing, and to take the length 
of the last lin einto account, one has to go through a lot of trouble; i 
once had it working but somehow lost something in the code, so i need to 
look at it

suggestion for gbetex: lastlinelength always available

Hans

-------------------------------------------------------------------------
                                   Hans Hagen | PRAGMA ADE | pragma@wxs.nl
                       Ridderstraat 27 | 8061 GH Hasselt | The Netherlands
  tel: +31 (0)38 477 53 69 | fax: +31 (0)38 477 53 74 | www.pragma-ade.com
-------------------------------------------------------------------------
                        information: http://www.pragma-ade.com/roadmap.pdf
                     documentation: http://www.pragma-ade.com/showcase.pdf
-------------------------------------------------------------------------

Re: Vertical alignment of formulas

[Pawel Jackowski na Onet]

Hi

> one complication is that the last line width is not available outside math
> mode (which is a pain)

If really needed one can use quite old plain macro which try to soften this
pain... However for short
paragraphs it needs some changes (\prevgraf can't be less then zero). Meybe
this is not very nice but...

\newif\ifmeasdisp \measdispfalse
\newcount\lineno
\newdimen\linesize
\def\measline{\ifhmode
              \bgroup
              \parfillskip=0pt plus 1fil
              \predisplaypenalty=10000
              \postdisplaypenalty=10000
              \displaywidowpenalty=\widowpenalty
              \abovedisplayskip=-\baselineskip
              \belowdisplayskip=-\baselineskip
              \abovedisplayshortskip=-\baselineskip
              \belowdisplayshortskip=-\baselineskip
              $$\global\linesize=\predisplaysize$$
              \global\advance\linesize by -2em
              \global\lineno=\prevgraf
              \global\advance\lineno by -3
              \ifmeasdisp\immediate\write16{>>\the\lineno: \the\linesize}\fi
              \global\advance\lineno by -1
              \global\prevgraf=\lineno
              \kern\linesize
              \egroup
              \else
              \immediate\write16{! I'M IN VMODE---MEASURING IGNORED}%
              \fi}
% ------------------------------------------------------------------------
% Author: B. Jackowski, 22 V 1990.
% ------------------------------------------------------------------------

Regards

Re: Vertical alignment of formulas

[Hans Hagen]

[-- Attachment #1: Type: text/plain, Size: 723 bytes --]

Hi,

I fixed/extended some experimental code. See attached file. Of course you 
need to move the core file to base and regenerate the format.

If you want autospacing, you need the \epar (this will trigger some line 
width calculations, somethign that needs to take place inside a par).

You can also \moveformula ([-top] but that is the default and onlyh option 
in not grid mode)

\optimizedisplayspacingtrue % needed

test \par \dorecurse{10}{test } \moveformula \startformula test 
\stopformula test \endgraf test
test \par \dorecurse{10}{test } \startformula test \stopformula test 
\endgraf test

\dorecurse{30}{\bpar \dorecurse\recurselevel{test } \epar \startformula 
formula \stopformula}

no guarantees -)

Hans  

[-- Attachment #2: new.zip --]
[-- Type: application/zip, Size: 12028 bytes --]

[-- Attachment #3: Type: text/plain, Size: 597 bytes --]

-------------------------------------------------------------------------
                                   Hans Hagen | PRAGMA ADE | pragma@wxs.nl
                       Ridderstraat 27 | 8061 GH Hasselt | The Netherlands
  tel: +31 (0)38 477 53 69 | fax: +31 (0)38 477 53 74 | www.pragma-ade.com
-------------------------------------------------------------------------
                        information: http://www.pragma-ade.com/roadmap.pdf
                     documentation: http://www.pragma-ade.com/showcase.pdf
-------------------------------------------------------------------------

Re: Vertical alignment of formulas

[Hans Hagen]

At 19:24 17/09/2003 +0200, you wrote:
>Hi
>
> > one complication is that the last line width is not available outside math
> > mode (which is a pain)
>
>If really needed one can use quite old plain macro which try to soften this
>pain... However for short
>paragraphs it needs some changes (\prevgraf can't be less then zero). Meybe
>this is not very nice but...
>
>\newif\ifmeasdisp \measdispfalse
>\newcount\lineno
>\newdimen\linesize
>\def\measline{\ifhmode
>               \bgroup
>               \parfillskip=0pt plus 1fil
>               \predisplaypenalty=10000
>               \postdisplaypenalty=10000
>               \displaywidowpenalty=\widowpenalty
>               \abovedisplayskip=-\baselineskip
>               \belowdisplayskip=-\baselineskip
>               \abovedisplayshortskip=-\baselineskip
>               \belowdisplayshortskip=-\baselineskip
>               $$\global\linesize=\predisplaysize$$
>               \global\advance\linesize by -2em
>               \global\lineno=\prevgraf
>               \global\advance\lineno by -3
>               \ifmeasdisp\immediate\write16{>>\the\lineno: \the\linesize}\fi
>               \global\advance\lineno by -1
>               \global\prevgraf=\lineno
>               \kern\linesize
>               \egroup
>               \else
>               \immediate\write16{! I'M IN VMODE---MEASURING IGNORED}%
>               \fi}
>% ------------------------------------------------------------------------
>% Author: B. Jackowski, 22 V 1990.
>% ------------------------------------------------------------------------

this 'using a fake formula' is more or less the way i determine the last 
line length, but keep in mind that this only works in situations like:

   some text \measline

and not in

   some text \par (or empty line) \measline

Hans

-------------------------------------------------------------------------
                                   Hans Hagen | PRAGMA ADE | pragma@wxs.nl
                       Ridderstraat 27 | 8061 GH Hasselt | The Netherlands
  tel: +31 (0)38 477 53 69 | fax: +31 (0)38 477 53 74 | www.pragma-ade.com
-------------------------------------------------------------------------
                        information: http://www.pragma-ade.com/roadmap.pdf
                     documentation: http://www.pragma-ade.com/showcase.pdf
-------------------------------------------------------------------------

Re[2]: Vertical alignment of formulas

[Giuseppe Bilotta]

Thursday, September 18, 2003 Hans Hagen wrote:

> this 'using a fake formula' is more or less the way i determine the last
> line length, but keep in mind that this only works in situations like:

>    some text \measline

> and not in

>    some text \par (or empty line) \measline

Hans, that is *not* supposed to work in those cases because a
\par or an empty line determine an end-of-paragraph, so the
formula starts a new paragraph and the lenght of the previous
line is *irrelevant* --you don't have a mid-paragraph display.

I know that you like those freeform spacey sources, but they
are not really 100% compatible with the empty line => paragraph
concept of TeX :)

-- 
Giuseppe "Oblomov" Bilotta

Re[2]: Vertical alignment of formulas

[Hans Hagen]

At 10:47 18/09/2003 +0200, you wrote:
>Thursday, September 18, 2003 Hans Hagen wrote:
>
> > this 'using a fake formula' is more or less the way i determine the last
> > line length, but keep in mind that this only works in situations like:
>
> >    some text \measline
>
> > and not in
>
> >    some text \par (or empty line) \measline
>
>Hans, that is *not* supposed to work in those cases because a
>\par or an empty line determine an end-of-paragraph, so the
>formula starts a new paragraph and the lenght of the previous
>line is *irrelevant* --you don't have a mid-paragraph display.
>
>I know that you like those freeform spacey sources, but they
>are not really 100% compatible with the empty line => paragraph
>concept of TeX :)

well, the problem is that in both cases:

   text $$math$$ text

   text \par $$math$$ \par text

one wants to optimize spacing, so the limitation of not having access to 
the last line length is a pitty (i understand that it can be tricky in 
cases like:

   text \par somevboxmanipulation $$math$$

since then one may wonder what the last line length applies to

Hans
-------------------------------------------------------------------------
                                   Hans Hagen | PRAGMA ADE | pragma@wxs.nl
                       Ridderstraat 27 | 8061 GH Hasselt | The Netherlands
  tel: +31 (0)38 477 53 69 | fax: +31 (0)38 477 53 74 | www.pragma-ade.com
-------------------------------------------------------------------------
                        information: http://www.pragma-ade.com/roadmap.pdf
                     documentation: http://www.pragma-ade.com/showcase.pdf
-------------------------------------------------------------------------

Re[3]: Vertical alignment of formulas

[Giuseppe Bilotta]

>>Hans, that is *not* supposed to work in those cases because a
>>\par or an empty line determine an end-of-paragraph, so the
>>formula starts a new paragraph and the lenght of the previous
>>line is *irrelevant* --you don't have a mid-paragraph display.
>>
>>I know that you like those freeform spacey sources, but they
>>are not really 100% compatible with the empty line => paragraph
>>concept of TeX :)

Thursday, September 18, 2003 Hans Hagen wrote:

> well, the problem is that in both cases:

>    text $$math$$ text

>    text \par $$math$$ \par text

> one wants to optimize spacing, so the limitation of not having access to
> the last line length is a pitty (i understand that it can be tricky in
> cases like:

>    text \par somevboxmanipulation $$math$$

> since then one may wonder what the last line length applies to

The point is that it's not true that in both cases one wants to
optimize spacing because the two cases have a totally different
meaning: in the first case you have a displayed equation within
a paragraph, so it makes sense to "optimize spacing" to put the
text and the formula as close together as possible (within
aesthetical limits). In the second case you have a formula
which is on a paragraph of its own, and is therefore "absolute"
from the surrounding text. In this case you *don't* want to put
the formula and text as close together as possible because they
are (supposedly) logically *independent* concepts and having
them more spaced in the printout helps noticing this
difference.

In the end it always end up on the totally different views on
displayed material that you and TeX have. We discussed this for
the "autoindent" stuff, remember? You always consider displays
a "separate entity" from the surrounding text, whereas TeX
allows it to be separate or not (depending on \pars around it).
TeX approach is both more flexible and more "correct".

Let's say for example that we're discussing Einstein's
equation. The following three examples have different logical
meaning and should therefore be formatted appropriately
differently:

=== EXAMPLE 1: Display part of the paragraph ===

Einstein's equation $$E=mc^2$$ ties energy and mass of a body
to etc

=== EXAMPLE 2: Display part of the previous paragraph but not
    of the following one ===

A relation between mass and energy of a body is expressed by
Einstein's equation $$E=mc^2.$$

We shall see further on that etc

=== EXAMPLE 3: Display part of the following paragraph but not
    of the previous one ===

some text etc.

$$E=mc^2$$ is Einstein equation and it ties energy and mass
of a body etc.

================================================

In Ex1, you want the displayed equation to have as little
spacing as possible both above and below, and the following
text should NOT start with an indent.

In Ex2, you want the displayed equation to be as close as
possible to the previous text, and to have some spacing below,
to ensure that it's clearly distinguishable from the next
paragraph, which is a new entity. Also, the following text
should start with an indent.

In Ex3, you want there to be more space above than below
because the previous text is independent from the equation. On
the contrary, it should be close together with the following
text which is in the same paragraph as the equation. The
following text should NOT be indented.

Now, while we solved the indenting issue with the introduction
of the indentnext=auto key=value, which automatically takes
care of the various cases, you now have to convince yourself
that the same must hold true for spacing around displayed
material:

text $$math$$ text

and

text\par $$math$$\par text

are NOT the same thing and should be treated differently!

-- 
Giuseppe "Oblomov" Bilotta

Re[3]: Vertical alignment of formulas

[Hans Hagen]

At 14:05 18/09/2003 +0200, you wrote:

>In the end it always end up on the totally different views on
>displayed material that you and TeX have. We discussed this for
>the "autoindent" stuff, remember? You always consider displays
>a "separate entity" from the surrounding text, whereas TeX
>allows it to be separate or not (depending on \pars around it).
>TeX approach is both more flexible and more "correct".

hm, in this case i more tend to look at the grayness of a page and then 
decide on a space correction like this; (which is why i also have a 
threshold)

>Let's say for example that we're discussing Einstein's
>equation. The following three examples have different logical
>meaning and should therefore be formatted appropriately
>differently:
>
>=== EXAMPLE 1: Display part of the paragraph ===
>
>Einstein's equation $$E=mc^2$$ ties energy and mass of a body
>to etc
>
>=== EXAMPLE 2: Display part of the previous paragraph but not
>     of the following one ===
>
>A relation between mass and energy of a body is expressed by
>Einstein's equation $$E=mc^2.$$
>
>We shall see further on that etc
>
>=== EXAMPLE 3: Display part of the following paragraph but not
>     of the previous one ===
>
>some text etc.
>
>$$E=mc^2$$ is Einstein equation and it ties energy and mass
>of a body etc.
>
>================================================
>
>In Ex1, you want the displayed equation to have as little
>spacing as possible both above and below, and the following
>text should NOT start with an indent.
>
>In Ex2, you want the displayed equation to be as close as
>possible to the previous text, and to have some spacing below,
>to ensure that it's clearly distinguishable from the next
>paragraph, which is a new entity. Also, the following text
>should start with an indent.

yes, but a very short last line will make uit look quite ugly, i.e. there 
is already quite some space there; i'd rather tend to have a different 
threshold then (say 4em instead of 2em, which boils down to:

some long text
                 formula part of paragraph

some text
                 formula part of paragraph

some long text

                 formula not part of paragraph

some text
                 formula not part of paragraph

so, to let the horizontal distance between end of line text and begin of 
text also play a role.

text $$math$$ text

>and
>
>text\par $$math$$\par text
>
>are NOT the same thing and should be treated differently!

i know, but that's on the agenda; and even then, i'll add options to 
control the look and feel -)

(in practice i think that it also depends on the taste and specific 
formula, say x=2 vs x=sqrt(2) may demand their own spacing rules)

Hans

-------------------------------------------------------------------------
                                   Hans Hagen | PRAGMA ADE | pragma@wxs.nl
                       Ridderstraat 27 | 8061 GH Hasselt | The Netherlands
  tel: +31 (0)38 477 53 69 | fax: +31 (0)38 477 53 74 | www.pragma-ade.com
-------------------------------------------------------------------------
                        information: http://www.pragma-ade.com/roadmap.pdf
                     documentation: http://www.pragma-ade.com/showcase.pdf
-------------------------------------------------------------------------

Re[4]: Vertical alignment of formulas

[Giuseppe Bilotta]

Thursday, September 18, 2003 Hans Hagen wrote:

> yes, but a very short last line will make uit look quite ugly, i.e. there
> is already quite some space there; i'd rather tend to have a different
> threshold then (say 4em instead of 2em, which boils down to:

> some long text
>                  formula part of paragraph

> some text
>                  formula part of paragraph

> some long text

>                  formula not part of paragraph

> some text
>                  formula not part of paragraph

> so, to let the horizontal distance between end of line text and begin of
> text also play a role.

actually a very short last time should NOT happen full stop. :)
There was a discussion on this recently on c.t.t., about
"French" orphans or "Russian" paragraphs. You can look them up
with GoogleGroups. To prevent those lines, Donald Arseneau
proposes something like

\parfillskip = .5\hsize plus .092\hsize minus .5\hsize

He says:

"This not only prohibits the short final lines, but heavily penalizes
moderately short final lines.

Another bad feature is that \abovedisplayshortskip will never be used,
leading to excessive white space around some formulas."

Given that you never use \abovedisplayshortskip, we're safe ;)

>> text $$math$$ text

>>and

>>text\par $$math$$\par text

>>are NOT the same thing and should be treated differently!

> i know, but that's on the agenda; and even then, i'll add options to
> control the look and feel -)

> (in practice i think that it also depends on the taste and specific
> formula, say x=2 vs x=sqrt(2) may demand their own spacing rules)

Don't exaggerate. The human eye is slack enough to give those
spacing some tolerance :)

-- 
Giuseppe "Oblomov" Bilotta

Re[4]: Vertical alignment of formulas

[Hans Hagen]

At 17:31 18/09/2003 +0200, Giuseppe Bilotta wrote:

>proposes something like
>
>\parfillskip = .5\hsize plus .092\hsize minus .5\hsize

hm, we need a way to parametrize this, will think of it

>Another bad feature is that \abovedisplayshortskip will never be used,
>leading to excessive white space around some formulas."

these skips around math are tricky anyway; also, i noticed that 
occasionally some baselineskip is added; another painful thing is the 
automatic 2em displayindent

 > Don't exaggerate. The human eye is slack enough to give those spacing 
some tolerance :)

i noticed that in many tex docs on the web -) also, i am quite sensitive to 
it, so, sorry to all those users who suffer from the side effects in context

Hans
-------------------------------------------------------------------------
                                   Hans Hagen | PRAGMA ADE | pragma@wxs.nl
                       Ridderstraat 27 | 8061 GH Hasselt | The Netherlands
  tel: +31 (0)38 477 53 69 | fax: +31 (0)38 477 53 74 | www.pragma-ade.com
-------------------------------------------------------------------------
                        information: http://www.pragma-ade.com/roadmap.pdf
                     documentation: http://www.pragma-ade.com/showcase.pdf
-------------------------------------------------------------------------

Re[5]: Vertical alignment of formulas

[Giuseppe Bilotta]

Thursday, September 18, 2003 Hans Hagen wrote:

> At 17:31 18/09/2003 +0200, Giuseppe Bilotta wrote:

>>proposes something like
>>
>>\parfillskip = .5\hsize plus .092\hsize minus .5\hsize

> hm, we need a way to parametrize this, will think of it

I was hoping you would :)

>>Another bad feature is that \abovedisplayshortskip will never be used,
>>leading to excessive white space around some formulas."

> these skips around math are tricky anyway; also, i noticed that
> occasionally some baselineskip is added; another painful thing is the
> automatic 2em displayindent

 >> Don't exaggerate. The human eye is slack enough to give those spacing
> some tolerance :)

> i noticed that in many tex docs on the web -) also, i am quite sensitive to
> it, so, sorry to all those users who suffer from the side effects in context

Make it an option, so that both you and "those who suffer" can
be happy about it :)

-- 
Giuseppe "Oblomov" Bilotta

Re: Vertical alignment of formulas

[Emil Hedevang Lohse]

Hans Hagen <pragma@wxs.nl> writes:

> Hi,
>
> I fixed/extended some experimental code. See attached file. Of course
> you need to move the core file to base and regenerate the format.
>
> If you want autospacing, you need the \epar (this will trigger some
> line width calculations, somethign that needs to take place inside a
> par).
>
> You can also \moveformula ([-top] but that is the default and onlyh
> option in not grid mode)
>
> \optimizedisplayspacingtrue % needed
>
> test \par \dorecurse{10}{test } \moveformula \startformula test
> \stopformula test \endgraf test
> test \par \dorecurse{10}{test } \startformula test \stopformula test
> \endgraf test
>
> \dorecurse{30}{\bpar \dorecurse\recurselevel{test } \epar
> \startformula formula \stopformula}

Thank you for the patch \moveformula now works fine.

I have not tested the \bpar-\epar thing since it looks like there is a
small bug: when I typeset your file test.tex, the ninth formula on
page 2 is incorrectly move up.

Regards,

-- 
Emil Hedevang Lohse <http://home.imf.au.dk/emil/> 

Alle spørgsmål er lige dumme. 
Og spørgsmålet "Kan ænder flyve?" er ikke dumt.