fix conversion of sign function to fricas

[fchapoton]

which apparently has no symbolic sign function

CC: @mantepse

Component: interfaces: optional

Keywords: FriCAS

Author: Frédéric Chapoton

Branch/Commit: 65695c1

Reviewer: Martin Rubey

Issue created by migration from https://trac.sagemath.org/ticket/31085

[fchapoton]

New commits:

9cdcb3f

fix conversion of sign function to fricas

[fchapoton]

Commit: 9cdcb3f

[fchapoton]

Branch: u/chapoton/31085

[mantepse]

comment:2

I don't quite understand the patch. If you want a sign function for symbolic expressions in fricas, you'll have to use x/abs(x), I guess. But I don't think that fricas would be able to do anything with it. There is a sign function for integers and integer fractions, but I don't see the point of providing a translation, because it should never show up in a result that fricas provides.

[fchapoton]

comment:3

I was trying the integral in #11590 with fricas, and it did not work. The existing conversion of the symbolic sign function is "sign", which cannot handle symbolic arguments.

But then apparently, Fricas cannot integrate functions containing abs.

[mantepse]

comment:4

Yes, that's correct (as far as I know).

So, do you think it makes sense to convert sign to x/abs(x)? (I don't)

[fchapoton]

comment:5

Well, at least by converting to abs(x)/x (I prefer this way, and did that), one can go to fricas and back to sage without error. Why do you think that it's not a good idea ?

[mantepse]

comment:6

Sorry, I missed that! Yes, I agree with that! A doctest might be good, I'll try to think of one.

[mantepse]

Reviewer: Martin Rubey

[mantepse]

Changed keywords from none to FriCAS

[mantepse]

comment:9

Here is a trival test, do you want to add it?

sage: fricas(sign(x)).eval(x=-3)
- 1

Other than that, it looks good. There is currently a giac failure, which I cannot say anything about.

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. New commits:

65695c1

add a doctest for fricas sign

[sagetrac-git]

Changed commit from 9cdcb3f to 65695c1

[fchapoton]

comment:11

Voilà !

[mantepse]

comment:12

Great, looks good. I cannot reproduce the giac failure, I guess it's unrelated.

[vbraun]

Changed branch from u/chapoton/31085 to 65695c1