Why don't we have the physics MathJax extension enabled? Should we?

Question

As a recent question points out, it would be nice if the site's MathJax environment included something similar to the physics LaTeX package.

Fortunately, the list of third-party extensions for MathJax does include a physics extension. I'm not sure if this has been considered before and discarded (and, if so: when and where? and why was this considered not feasible?), but if we haven't already, then we should.

I note, in particular, that the list includes the mhchem extension, which the Chemistry site has had enabled for the past six years.

Can we do the same? Should we?

Answer 1

Just from a brief look at this, it seems to me like a bad idea to enable it, for several reasons.

1. It changes the behavior of various commands. Example:

   \sin\left(\frac{1+\frac{1}{1+\frac{1}{2}}}{2}\right)

   $\sin\left(\frac{1+\frac{1}{1+\frac{1}{2}}}{2}\right)$

   \sin(\frac{1+\frac{1}{1+\frac{1}{2}}}{2})
   

   $\sin(\frac{1+\frac{1}{1+\frac{1}{2}}}{2})$

   Currently, when I look at this in my browser the results are the same as I would expect in standard latex. The parens are big in the first example and small in the second example. But if I load the physics package and compile this document in latex, it automagically supplies the big parens.

   I kind of doubt that this standard behavior was a mistake by Donald Knuth because he wasn't smart enough to make it work better. I'm sure it was a conscious design decision. If we were going to change it at this point, we would break the behavior of old code, and people would start writing code on physics.SE that would not be compatible with standard latex.

   If you're using the latex implementation of the physics package, then you have the option of turning off this behavior and reverting to standard tex math by doing \usepackage[notrig]{physics}, but in mathjax on physics.SE we wouldn't have any way to do that, would we?

2. New commands like \trace are defined. I doubt that these are compatible with adaptive technologies for the blind, which probably assume standard-ish tex/mathjax math.

3. A lot of the functionality wouldn't be usable in the context of mathjax on SE. E.g., if you want an arrow over your nabla operator, you can do \usepackage[arrowdel]{physics}, but that won't exist on the SE implementation.

4. It would break compatibility. For instance, if I post an answer on SE and later want to use it in a latex document, I don't want to have to worry about the fact that it may not work because I'm using two different dialects. Although mathjax is different from standard tex/latex math, they are similar enough that the differences almost never crop up.

5. The physics package for latex is fairly new, and the mathjax extension is newer. We would be effectively acting as large-scale testers this mathjax extension. I don't know how good the code quality is or how well supported it will be. What if we have a lot of problems with bugs and incompatibilities and decide that it was a bad idea? We can't switch back without breaking content that people wrote using the extension.

6. It breaks compatibility between SE sites. We want to be able to say on mathoverflow, for example, "According to this answer on physics.SE, \$...\$" and have the quoted math display properly. When a question is closed and moved to another site where it's more on topic, we want that to work.

Answer 2

Note:

This answer's a placeholder for votes, to gauge the community's opinion regarding whether it's a god idea or not. As pointed out in the comments, there are a few pretty nasty bugs, and I don't support the addition of the physics package any more.

I'm actually quite fond of the physics $\LaTeX$ package.

1. It has a neat support for derivatives: \dv[n]{f}{x} gives $\frac{\mathrm{d}^nf}{\mathrm{d}x^n}$, and it's much more concise than \frac{\mathrm{d}^nf}{\mathrm{d}x^n}. Also, it makes the d straight automatically, which is perfect.

2. It has \grad and \curl, which are cleaner and more readable than \nabla\cdot and \nabla\times. (I'm not sure if this supports arrows over the nabla, but you could still use \vec{\nabla} if you wanted to)

3. There's the advantage of easy/intuitive bra-ket notation, as mentioned in the post you linked in the question.

These are 3 pretty common stuff, which I think is good enough evidence that it's going to be used plenty. So if it's stable enough, I think we should go ahead and set it up!