Documentation: Explain that min, max do not work for SR

[alexjbest]

On 9.2beta0

sage: a = var('a')
sage: min(a,+Infinity)
a
sage: min(+Infinity,a)
+Infinity

returning a in both cases would be best, but at the very worst this should consistently throw an error imo.

Component: documentation

Keywords: symbolics, infinity, minimum, maximum

Issue created by migration from https://trac.sagemath.org/ticket/29802

[mkoeppe]

comment:1

This is Python's built-in min. Try min_symbolic

[alexjbest]

comment:2

@mkoeppe, thanks, sorry for the noise, I searched pretty hard for this information (using Google and trac search). I wonder how we could make this information more discoverable?

Perhaps adding a min method to sage.symbolic.expression.Expression to compute the minimum of self and other in this way?

[kcrisman]

comment:4

I am pretty sure we tried doing something along the lines of making this easier to find, but apparently not. For instance, it could be useful in plot documentation too - see here.

See also:

• repr of max_symbolic and min_symbolic when passed a list argument #26650
• reference manual

As for the suggestion, are there any other builtins that we are able to treat in this manner? Other than things like repr() = __repr__()? I don't see anything about min or max in the Python docs on custom classes. The actual CPython source seems to implement a generic function that then gets a max via gt and min via lt, but I don't see an easy way (or desire) to hack this. Unfortunately improving doc - a lot - may be the best way to handle this. (I don't immediately see how the key argument would solve this, for instance.)

[kcrisman]

comment:5

Interestingly, we actually do do this for abs (see builtins list), and it causes a problem, because

sage: abs?
Signature:      abs(x, /)
Docstring:      Return the absolute value of the argument.
Init docstring: Initialize self.  See help(type(self)) for accurate signature.
File:           
Type:           builtin_function_or_method

But presumably we want something more than that, such as the content here. I'm not sure why this is what is returned.

Anyway - maybe that means we can overload min/max somehow, maybe even as naively as for abs. Though note #6949 comment:2.

[mkoeppe]

comment:6

See also #24296

[mkoeppe]

comment:8

Setting new milestone based on a cursory review of ticket status, priority, and last modification date.