On Thu, 12 Jan 2006, Hans Hagen wrote:

Kumar Appaiah wrote:

...

2. The tutorial tells us to refer to other sources (The TeX Book) for more on Math typesetting. Is there any online resource which would explain TeX (as opposed to LaTeX) math?

most math (also in latex) is tex math, macro packages add additional symbols (context has them), some matrix stuff (is available in math modules) and font handling (depends on font system used)

Most math in latex is *not* same as tex math. Though, most that can be done with latex can also be done in tex, but latex does provide a nice interface which sadly is missing in context. amsl and nath modules provide some of this functionality but a lot still needs to be done to make Context math handling as easy as it is in Latex. Having said that, the most commonly used math functionality of latex+amsmath is available in context. Unfortunately it is not well documented. You need to read some latex amsmath tutorial to get familiar with the syntax, then you can simply replace \begin{...} \end{...} pairs with \start .... \stop pairs and the basic features will work.

the best resource is still "the tex book"

What is the development status of amsl and nath modules? Aditya

On Thu, 12 Jan 2006, Hans Hagen wrote:

Aditya Mahajan wrote:

...

Most math in latex is *not* same as tex math. Though, most that can be done with latex can also be done in tex, but latex does provide a nice interface which sadly is missing in context. amsl and nath modules provide some of this functionality but a lot still needs to be done to make Context math handling as easy as it is in Latex.

Do you mean that everything between $ $ and \begin{math} \end{math} is different?

a + b \int ... \sqrt

etc is not different is it?

No, I mean the complicated math is much harder in context. Consider \begin{align} a &= b \\ c &= d \notag \\ &= f \notag \\ &= g \end{align} which will typeset as a = b (1) c = d = f = g (2) There is no Context way to do it, and one has to resort to plain tex \placeformula $$ \eqalignno{ a &= b &\formulanumber \cr c &= d \cr &= f \cr &= g \formulanumber} $$ Context gives the same result, but the input syntax is much more verbose than the latex syntax. Latex also has a lot of other useful enviromnemts like \begin{equation} \begin{split} a &= b \\ &= c \end{split} \end{equation} Again, the same effect can be achieved in plain tex, but it is more verbose. amsmath also has some useful environments like multalign and aligned, gathered, faligned, alignat, etc. Some of their functionality can be achieved using \start stop array from amsl but the amsmath environments have a lot more features. There are certain features that are much more difficult to achieve using context. Consider equation subnumbering. In latex, it is as simple as \begin{subequations} \begin{align} a &= b \\ c &= d \end{align} \end{subequations} Compare this from how to do this in context (see the wiki). You have to *manually* set the number of the subequation. Actually, for equation numbering and refering, the context way is rather limited. Consider something like an align environment a &= b\\ c &= d\\ e &= f Suppose, I want to refer to the second equation. In latex, I can simply add \label{eq:2} and the end of c&= d and then \ref{eq:2}. For context, the reference label goes at the top, with \placeformula[eqs]. But I am not sure, how to give individual labels to each equations. Browse through http://www-sop.inria.fr/miaou/latex/voss-math.pdf which explains almost all math tricks in latex. For some of the more complicated stuff, compare the latex code and the context code to achieve that result. Context has all the features, but for complicated maths, they are low level tricks. There is no upper layer like amsmath making life easier for the user. It will be nice to have such a layer. Aditya

<--- On Jan 13, Hans Hagen wrote --->

Aditya Mahajan wrote:

...

No, I mean the complicated math is much harder in context. Consider

\begin{align} a &= b \\ c &= d \notag \\ &= f \notag \\ &= g \end{align}

which will typeset as

a = b (1) c = d = f = g (2)

\begin{subequations} \begin{align} a &= b \\ c &= d \end{align} \end{subequations}

Compare this from how to do this in context (see the wiki). You have to *manually* set the number of the subequation. Actually, for equation numbering and refering, the context way is rather limited. Consider something like an align environment

a &= b\\ c &= d\\ e &= f

Suppose, I want to refer to the second equation. In latex, I can simply add \label{eq:2} and the end of c&= d and then \ref{eq:2}. For context, the reference label goes at the top, with \placeformula[eqs]. But I am not sure, how to give individual labels to each equations.

this 'loose label' is one of the ugliest concept i can think of -)

If you come up with any other way, I will be glad to use it. Frankly, I do not think that the latex syntax is the best, but plain tex looks too indimidating and context does not have anything yet.

btw, defaulting to numbers and then using \notag is messy; i prefer readable code, even if it takes more bytes; inventing a formula takes more time than keying it in. Also, more structure, means more hooks for configurability

Fine by me. Your method is perfectly acceptable.

much if this 'complicated' math is not that complicated to support,see attached file

Wonderful. I did not know that something like this can be done so easily in context.

i'm willing to implement anything reasonable but since i hardly use such math i only act on 'i want to achieve this' kind of specs (i have no time to read tons of tex documents)

Here are features that I would want context math to have. I do not care about the input syntax (whether it is same as latex or not) as long as the features are there. 1. Have align support with variable number of columns. a &= b &= c &+ d & &+ e should be typeset as a = b = c + d + e 2. Have a mechanism to individually number/not-number an equation. 3. Have a gather environment. Depending on the user option should either center, left justify or right justify all equations. Useful when you have a bunch of equation together and do not want to have startforumla and stopformula after each of them. 4. In align you should be able to specify the separation between columns. There should be some input syntax that allows you to typeset an output like a = b c = d e = f g = h It will be great if you can number each equation (i.e. four number in the above example). There should be an option to only number each line. 5. have a splitformula environment that allows a =& b & + c \\ & + d to come out as a = b + c + e (1) with only one formula number. The formula number should be at the last line or the center of the whole formula, configurable by an option. 6. Allow, someway of typesetting a = b + c Using some result = d + e that is, you should be able to come out of the align mode temporarily and then go back. Latex calls this intertext. Have some means to adjust the before and after skips of intertext. Something like \setupintertextskip[small or medium or big] 7. Allow some way of typesetting multilined formula where first line is left aligned second line after some hskip third line after some hskip ... last line right alinged This should have only one formula number, either on the last line or in the center of the display. 8. Have a version of align, gather and multiline that can be used inside a formula. So one should be able to type \startformula \left.\startaligned \NC a \NC= b \NR \NC c \NC= d \NR \stopaligned\right\} \implies \startaligned \NC b \NC= a \NR \NC d \NC= c \NR \stopaligned \stopformula to get a = b } => b = a c = d } d = c where the implies is centered between the two lined. Have an option to make these environments come at the top, middle or bottom of the baseline. 9. Allow easy input of multiline subscripts. Suppose I want \sum a < b c < d e < f 10. Allow the big subscripts of math operators to be smashed to the left or right. Suppse, I have a \times \lim_{a very long equation} b_n There should be a command, say \smashoperator such that a \times \smashoperator\lim_{a very long equation} b_n comes as a lim b_n a very long eq where (a very long equation) in centered beneath the center of lim but the "bounding box" is the bb of lim not the bb of the subscript Also a \times \smashoperator[r]\lim_{a very long equation} b_n should come as a lim b_n a very lone eq where the "bounding box" of a lim on the right is the bb of the lim, on the left is the bb of the subscript. Similarly of \smashoperator[l] 11. Have commands like \medop and \bigop which typeset the following operator slightly smaller or slightly larger. 12. Have a command say \novheight (or whatever) so that while typing \left\{\novheight \sum_{i=0}^n a_n \right\} the braces do not take the size of the sum into account while scaling. 13. Have a command say \tag so that \startformula a = b \tag{*} \stopformula comes out as a = b (*) 14. Allow a framed command that works with align and aligned. So that a \implies\mframed{ \startaligned \NC b \NC= c \NR \NC e \NC= f \NR \NC g \NC= h \NR \stopaligned} Comes as +---------+ | b = c | a => | e = f | | g = h | +---------+ 15. Automatic numbering in subequations. In the attachment that you sent, one has to number the subequations by hand. The trouble with this is that if you rearrange the equations at a later stage, you need to keep track of equation subnumbering. These are all that I can think of right now. All these features are available in latex. It will be great if context can support these features (or a subset of thereof). I promise to document them if they get incorporated :-) Aditya -- Aditya Mahajan, EECS Systems, University of Michigan http://www.eecs.umich.edu/~adityam || Ph: 7342624008

Just to give you some hints what's already there... Aditya Mahajan writes: < 63 lines deleted by Adam Duck >

1. Have align support with variable number of columns. a &= b &= c &+ d & &+ e

should be typeset as

a = b = c + d + e

At least this should be possible with \wall and \return (using the "nath" module), like this: #v+ a \wall = b \\ = c \wall + d \\ + e \return\return #v- < 20 lines deleted by Adam Duck >

5. have a splitformula environment that allows

a =& b & + c \\ & + d

to come out as

a = b + c + e (1)

with only one formula number. The formula number should be at the last line or the center of the whole formula, configurable by an option. < 142 lines deleted by Adam Duck >

Again, this is a \wall-\return thingy, but \placeformula places the number vertically centered... #v+ \usemodule[nath] \starttext \placeformula \startformula a \wall = b \\ +c \\ +e \return \stopformula \startformula a \wall = b \\ = c \wall + d \\ +e \return\return \stopformula \stoptext #v- cu, Adam.

Hello,

%D \startbuffer %D \placeformula \startformula \startalign %D \NC a \NC \eq b \NR[+] %D \NC c \NC \neq d \NR %D \NC \NC \neq f \NR[for:hans] %D \NC \NC \geq g \NR[for:whoelse][a] %D \NC \NC \leq h \NR[for:whomore][b] %D \NC \NC \neq i \NR %D \stopalign \stopformula %D \stopbuffer Can this be enhanced to allow more than one alignment points? For instance f(x) = x²+b² = g(x) (1) h(x,y) = x²-5y² = g_2(x,y) (2) should be aligned at the each "=", currently - using \equalign(no) or \startalign - one can only align at one.

Tobias