doctest that only integer +/- 1 gets special treatment

[williamstein]

This is just a doctest of a fix already in main Sage.

Original description:

This is really bad

sage: 1.0 * pi
pi
sage: sin(1.0 * pi)
0

This is good:

sage: maxima('sin(%pi*1.0)')
sin(1.0*%pi)
sage: maxima('1.0*%pi')
1.0*%pi

This is likely the fault of ginac doing something stupid with 1.0 == 1.

Ken Ribet pointed this out during the JMM 2012 meeting.

Depends on #18088
Depends on #18362

CC: @kcrisman

Component: calculus

Keywords: pynac

Author: Ralf Stephan

Branch/Commit: 906736b

Reviewer: Jeroen Demeyer

Issue created by migration from https://trac.sagemath.org/ticket/12257

[burcin]

comment:1

There are two sides to this. Doing the multiplication and printing the coefficient.

GiNaC takes a shortcut when the coefficient is 1 and skips multiplication. There is a Python callback function (sage.symbolics.pynac.py_get_parent_char) to check the characteristic of the parent. Pynac forces the operation for positive characteristic. If we return a positive integer from that function when the parent is inexact, the multiplication will go through.

If a coefficient is 1, we don't print it. So the printing code in pynac needs to be changed to print inexact coefficients. This is now pynac issue #4.

Even when all these are sorted out, sin(1.0*pi) might still return 0. Since bool(1.0*pi == pi) will probably be True.

[burcin]

Changed keywords from none to pynac

[williamstein]

comment:3

It should be that sin(1.0*pi) returns 0.0. We can fix this somehow. I'm not worried about printing.