A genuine bug in numerical evaluation ?

[Emmanuel Charpentier]

This ask.sagemath  question seems to raise a genuine bug. Perusing Trac doesn't raise muc relevant, possibly except for Trac#24428 and Trac#21754.

From my comment to the ask.sagemath question :

Indeed :

    sage: var("y,z")

    (y, z)

    sage: cauchy(z)=solve(z*y^3 +y^2 - 2*z*y+2 ,y)[0].rhs()

    sage: def foo(u):return(arg(cauchy(u)).n())

    sage: [foo(t) for t in (1,1.1..2)]

    [0.828222717321238,

    ## Snip...

     0.377665318352514]

Therefore, this function *can* be evaluated. But :

    sage: plot(foo,(1,2))

    verbose 0 (3635: plot.py, generate_plot_points) WARNING: When plotting, failed to evaluate function at 200 points.

    verbose 0 (3635: plot.py, generate_plot_points) Last error message: 'negative number cannot be raised to a fractional power'

    Launched png viewer for Graphics object consisting of 0 graphics primitives

... not in `plot`. Trying  to plot symbolically yelds :

    TypeError: Cannot evaluate symbolic expression to a numeric value.

A bug, IMHO.

Can someone confirm this and, possibly, give me hints as how to flag this ticket efficiently ?

[Emmanuel Charpentier]

According to Serge Lelièvre, the proble doesn't occur when ran under a Python3-based Sage. He advises to run such a Sage, hich seems a bit early, IMHO?

Advice ?

[John H Palmieri]

I can confirm that it works for me on OS X with Python 3, gives the error you described with Python 2.

[John H Palmieri]

Oh, and Sage with Python 3 is in pretty good shape these days. Still some doctest failures, but I think it is mostly very functional.

  John