Wrong LaTeX for products of numbers

[novoselt]

E.g.

sage: e = 2 * 2^(1/3)
sage: print latex(e)
2 \, 2^{\left(\frac{1}{3}\right)} 

the output looks like power of 22, there should be \cdot.

(And there should be no parenthesis around the power, but that does not change the meaning at least.)

Depends on #15198

CC: @burcin @eviatarbach

Component: symbolics

Keywords: latex, pynac, sd48

Author: Burcin Erocal

Branch/Commit: 2c9b4e6

Reviewer: Jean-Pierre Flori

Issue created by migration from https://trac.sagemath.org/ticket/13356

[novoselt]

comment:1

My investigation shows that the issue is in GEx_to_str_latex, but I have no clue how to fix it.

[novoselt]

comment:2

#3202 was about the same problem and claims to be fixed, but I am not sure what is/was the problem with the example there. On the other hand, there is a test in the docstring of symbolic expressions claiming that 2 \, 2 is great.

[novoselt]

comment:3

Burcin, any ideas on whether it is in Pynac or Ginac and how to fix it?

[novoselt]

comment:4

Slight change in Sage-5.10.rc1:

sage: sage: e = 2 * 2^(1/3)
sage: sage: print latex(e)
2 \, 2^{\frac{1}{3}}

[burcin]

Attachment: trac_13356-latex_coeff_sep.patch.gz

[burcin]

comment:5

Patch for pynac available on my work-in-progress repo. Doctest fixes in attachment: trac_13356-latex_coeff_sep.patch

Now the output is:

sage: latex(2 * 2^x)
2 \cdot 2^{x}

We still keep \, if the next character is not a digit:

sage: latex(2* x^2)
2 \, x^{2}

[burcin]

Author: Burcin Erocal

[burcin]

Changed keywords from latex to latex, pynac, sd48

[burcin]

comment:6

In this sage-support@ thread the consensus seems to be to remove \, and only use \cdot when both sides are numerals.

I am planning to make this change if there are no objections. I'm afraid this will mean fixing doctests all over the tree yet again.

[novoselt]

comment:7

I certainly would not object - I think Sage-generated code should be as close to human-written as possible, so that it is relatively easy to copy-paste into manually written papers. I've never seen anyone manually inserting \, for every product. (It would be good also to have x_0^2 rather than x_{0}^{2}, but that's not related to this ticket.)