Hi, for nice linebreaking we can adjust math formulas by the use of NC, NR alignment. But if someone only wants long formulas to fit in a defined area, and want them to break automatically ... is there a handy solution? (Below are examples that don't work) Thanks, Steffen ------- \starttext \placeformula \startformula[9pt] I=\big\{\lambda u(x_0-\pi^H K)+(1-\lambda)[\pi^L u(x_0-K+(1-\pi^L)Z_{L'}) +(1-\pi^L)u(x_0-\pi^L Z_{L'})]\big\}-\big\{\pi^U u(x_0-K+(1-\pi^U)Z_{U'}) +(1-\pi^U)u(x_0-\pi^U Z_{U'})\big\}. \stopformula \defineframedtext [defbackground] [width=10cm, frame=on, location=paragraph] \startdefbackground \placeformula \startformula[9pt] I=\big\{\lambda u(x_0-\pi^H K)+(1-\lambda)[\pi^L u(x_0-K+(1-\pi^L)Z_{L'}) +(1-\pi^L)u(x_0-\pi^L Z_{L'})]\big\} -\big\{\pi^U u(x_0-K+(1-\pi^U)Z_{U'}) +(1-\pi^U)u(x_0-\pi^U Z_{U'})\big\}. \stopformula \stopdefbackground \definelayer[mylayer] \setlayerframed [mylayer] [x=25mm, y=85mm, height=43.5mm, width=10cm,align=block] { \placeformula \startformula[9pt] I=\big\{\lambda u(x_0-\pi^H K)+(1-\lambda)[\pi^L u(x_0-K+(1-\pi^L)Z_{L'}) +(1-\pi^L)u(x_0-\pi^L Z_{L'})]\big\} -\big\{\pi^U u(x_0-K+(1-\pi^U)Z_{U'}) +(1-\pi^U)u(x_0-\pi^U Z_{U'})\big\}. \stopformula } \setupbackgrounds[page][background=mylayer] \stoptext
On Thu, 9 Aug 2012, Steffen Wolfrum wrote:
Hi,
for nice linebreaking we can adjust math formulas by the use of NC, NR alignment.
But if someone only wants long formulas to fit in a defined area, and want them to break automatically ... is there a handy solution?
Short answer: No easy solution exists. Aditya
Steffen Wolfrum wrote:
them to break automatically ... is there a handy solution?
Aditya Mahajan wrote:
Short answer: No easy solution exists.
Is this because allowing automatic line breaks in math mode is itself hard to implement, or is the basic mechanism possible but preventing stupid/ugly breakpoints is hard? Because I can imagine a 70 percent solution being a big help already. --Sietse
On Thu, 9 Aug 2012, Sietse Brouwer wrote:
Steffen Wolfrum wrote:
them to break automatically ... is there a handy solution?
Aditya Mahajan wrote:
Short answer: No easy solution exists.
Is this because allowing automatic line breaks in math mode is itself hard to implement, or is the basic mechanism possible but preventing stupid/ugly breakpoints is hard? Because I can imagine a 70 percent solution being a big help already.
Partly both. The question is not so much the difficulty in choosing break-points (which is easy: before all \mathbin atoms, after \mathpunct if needed, etc). The hard part is how to align the split up portions, and choosing different alignments changes the line widths, and therefore affect future line-breaks. The old manual of breqn (the one by Michel Jones) had some discussion on the different alignment of display math. They are also explained nicely in the book on typesetting math by AMS (I forget the author's name). In principle, one can follow the same approach as Knuth did for line-breaking text (define penalties and then solve a dynamic program to find the layout that gives the lowest score). The breqn package does implement such an algorithm that works well in most cases. I wish that the breqn algorithm were described somethere; then one could play around implementing it in luatex. Trying to read breqn code to understand the algorithm is very difficult. Aditya
Hi Aditya,
I wish that the breqn algorithm were described somethere; then one could play around implementing it in luatex.
Have you seen the new User's Guide yet? It's from last May, very recent, so it may not have been there yet when last you looked. It includes the entire source code, and comments that quite thoroughly: the ratio works out to 16 lines of comment per 26 lines of code. (A page is about 42 lines.) And the comments go into the "how", and not just the "what". Might be useful at some point in the future, if it is indeed new. http://ctan.unixbrain.com/macros/latex/contrib/mh/breqn.pdf --Sietse P.s. The lack of an index and a table of contents is criminal, although probably due to this being a beta version (v0.98b) of the manual. I've extracted the sections and subsection titles, albeit without the page numbers. == User's guide == 1 A bit of history 2 Package loading 3 To do 4 Introduction 5 Principal features 6 Shortcomings of the package 6.1 Incompatibilities 6.2 Indention of delimited fragments 6.3 Math symbol subversion 6.4 Subscripts and superscripts 7 Incomplete 8 Package options 9 Environments and commands 9.1 Environments 9.2 Commands 10 Various environment options 11 The \@exisym package 12 Caution! Warning! 13 Examples == Implementation == 14 Introduction 15 Strategy 16 Prelim 17 Package options 18 Required packages 19 Some useful tools 20 Debugging 21 The \listwidth variable 22 Parameters 23 Measuring equation components 24 The dmath and dmath* environments 25 Special processing for end-of-equation 26 Preprocessing the equation body 27 Choosing optimal line breaks 28 Equation layout options 29 Centered Right-Number Equations 30 Framing an equation 31 Delimiter handling 32 Series of expressions 33 Equation groups 34 The darray environment 35 Miscellaneous 36 Wrap-up 37 To do
Addendum: I just bumped into Morten Høgholm's bachelor's thesis. "Automatic line-breaking of displayed math expressions" https://sites.google.com/site/mortenhoegholm/breqn-thesis.pdf That thesis contains a conceptual explanation of the breqn algorithm (or at least the algorithm he used when reimplementing breqn as l3breqn). Might be useful if/when somebody decides to have a go at improving ConTeXts math breaking. The thesis also contains the code of l3breqn, but that is rather more sparsely commented than the code in the breqn.pdf manual. --Sietse
On 9-8-2012 20:53, Aditya Mahajan wrote:
On Thu, 9 Aug 2012, Sietse Brouwer wrote:
Steffen Wolfrum wrote:
them to break automatically ... is there a handy solution?
Aditya Mahajan wrote:
Short answer: No easy solution exists.
Is this because allowing automatic line breaks in math mode is itself hard to implement, or is the basic mechanism possible but preventing stupid/ugly breakpoints is hard? Because I can imagine a 70 percent solution being a big help already.
Partly both.
The question is not so much the difficulty in choosing break-points (which is easy: before all \mathbin atoms, after \mathpunct if needed, etc). The hard part is how to align the split up portions, and choosing different alignments changes the line widths, and therefore affect future line-breaks.
The old manual of breqn (the one by Michel Jones) had some discussion on the different alignment of display math. They are also explained nicely in the book on typesetting math by AMS (I forget the author's name).
In principle, one can follow the same approach as Knuth did for line-breaking text (define penalties and then solve a dynamic program to find the layout that gives the lowest score). The breqn package does implement such an algorithm that works well in most cases. I wish that the breqn algorithm were described somethere; then one could play around implementing it in luatex. Trying to read breqn code to understand the algorithm is very difficult.
When I'm done ceaning up the pending tex and lua code, I'll look into these issues. It's probably not that hard to come up with a solution in lua (one of the pet projects on my list). Hans ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | voip: 087 875 68 74 | www.pragma-ade.com | www.pragma-pod.nl -----------------------------------------------------------------
On 09/08/12 09:37, Steffen Wolfrum wrote:
Hi,
for nice linebreaking we can adjust math formulas by the use of NC, NR alignment.
But if someone only wants long formulas to fit in a defined area, and want them to break automatically ... is there a handy solution?
(Below are examples that don't work)
Thanks, Steffen -------
\starttext
\placeformula \startformula[9pt] I=\big\{\lambda u(x_0-\pi^H K)+(1-\lambda)[\pi^L u(x_0-K+(1-\pi^L)Z_{L'}) +(1-\pi^L)u(x_0-\pi^L Z_{L'})]\big\}-\big\{\pi^U u(x_0-K+(1-\pi^U)Z_{U'}) +(1-\pi^U)u(x_0-\pi^U Z_{U'})\big\}. \stopformula
\defineframedtext [defbackground] [width=10cm, frame=on, location=paragraph]
\startdefbackground \placeformula \startformula[9pt] I=\big\{\lambda u(x_0-\pi^H K)+(1-\lambda)[\pi^L u(x_0-K+(1-\pi^L)Z_{L'}) +(1-\pi^L)u(x_0-\pi^L Z_{L'})]\big\} -\big\{\pi^U u(x_0-K+(1-\pi^U)Z_{U'}) +(1-\pi^U)u(x_0-\pi^U Z_{U'})\big\}. \stopformula \stopdefbackground
\definelayer[mylayer]
\setlayerframed [mylayer] [x=25mm, y=85mm, height=43.5mm, width=10cm,align=block] { \placeformula \startformula[9pt] I=\big\{\lambda u(x_0-\pi^H K)+(1-\lambda)[\pi^L u(x_0-K+(1-\pi^L)Z_{L'}) +(1-\pi^L)u(x_0-\pi^L Z_{L'})]\big\} -\big\{\pi^U u(x_0-K+(1-\pi^U)Z_{U'}) +(1-\pi^U)u(x_0-\pi^U Z_{U'})\big\}. \stopformula }
\setupbackgrounds[page][background=mylayer]
\stoptext
When I needed something like that, I used \allowbreak with inlinemath. In your case, you could do: \bTABLE[frame=on,align=right,width=10cm] \bTR \bTD $I=\big\{\lambda u(x_0-\pi^H K)\allowbreak+(1-\lambda)[\pi^L u(x_0-K+(1-\pi^L)Z_{L'})\allowbreak+(1-\pi^L)u(x_0-\pi^L Z_{L'})]\big\}\allowbreak -\big\{\pi^U u(x_0-K+(1-\pi^U)Z_{U'})\allowbreak+(1-\pi^U)u(x_0-\pi^U Z_{U'})\big\}.$ \eTD \eTR \eTABLE This will just flow it like regular text. I don't know if this is good enough for your needs. If not, you can add another way to your list of ways ;-). -- Prash
Am 09.08.2012 um 22:21 schrieb Prashanth:
On 09/08/12 09:37, Steffen Wolfrum wrote:
Hi,
for nice linebreaking we can adjust math formulas by the use of NC, NR alignment.
But if someone only wants long formulas to fit in a defined area, and want them to break automatically ... is there a handy solution?
(Below are examples that don't work)
Thanks, Steffen -------
\starttext
\placeformula \startformula[9pt] I=\big\{\lambda u(x_0-\pi^H K)+(1-\lambda)[\pi^L u(x_0-K+(1-\pi^L)Z_{L'}) +(1-\pi^L)u(x_0-\pi^L Z_{L'})]\big\}-\big\{\pi^U u(x_0-K+(1-\pi^U)Z_{U'}) +(1-\pi^U)u(x_0-\pi^U Z_{U'})\big\}. \stopformula
\defineframedtext [defbackground] [width=10cm, frame=on, location=paragraph]
\startdefbackground \placeformula \startformula[9pt] I=\big\{\lambda u(x_0-\pi^H K)+(1-\lambda)[\pi^L u(x_0-K+(1-\pi^L)Z_{L'}) +(1-\pi^L)u(x_0-\pi^L Z_{L'})]\big\} -\big\{\pi^U u(x_0-K+(1-\pi^U)Z_{U'}) +(1-\pi^U)u(x_0-\pi^U Z_{U'})\big\}. \stopformula \stopdefbackground
\definelayer[mylayer]
\setlayerframed [mylayer] [x=25mm, y=85mm, height=43.5mm, width=10cm,align=block] { \placeformula \startformula[9pt] I=\big\{\lambda u(x_0-\pi^H K)+(1-\lambda)[\pi^L u(x_0-K+(1-\pi^L)Z_{L'}) +(1-\pi^L)u(x_0-\pi^L Z_{L'})]\big\} -\big\{\pi^U u(x_0-K+(1-\pi^U)Z_{U'}) +(1-\pi^U)u(x_0-\pi^U Z_{U'})\big\}. \stopformula }
\setupbackgrounds[page][background=mylayer]
\stoptext
When I needed something like that, I used \allowbreak with inlinemath. In your case, you could do: \bTABLE[frame=on,align=right,width=10cm] \bTR \bTD $I=\big\{\lambda u(x_0-\pi^H K)\allowbreak+(1-\lambda)[\pi^L u(x_0-K+(1-\pi^L)Z_{L'})\allowbreak+(1-\pi^L)u(x_0-\pi^L Z_{L'})]\big\}\allowbreak -\big\{\pi^U u(x_0-K+(1-\pi^U)Z_{U'})\allowbreak+(1-\pi^U)u(x_0-\pi^U Z_{U'})\big\}.$ \eTD \eTR \eTABLE
This will just flow it like regular text. I don't know if this is good enough for your needs. If not, you can add another way to your list of ways ;-).
Hey, you skipped start/stopformula and used a $$ environment?! Why not, that's a quite clever approach (... actually, it's a quite efficient workaround for my needs). Thank you Prash, thank you very much! Steffen
participants (5)
-
Aditya Mahajan -
Hans Hagen -
Prashanth -
Sietse Brouwer -
Steffen Wolfrum