log plots blank with points, but fine with lines

[jason]

Inspired by ​http://ask.sagemath.org/question/1743/list_plot_semilogy-not-working, look at this. The following three plots are all blank (and the log axes also seem blank)

import mpmath
ber = [0.5*mpmath.erfc(i)/sqrt(2) for i in range(10)]
points(enumerate(ber)).show(scale='semilogx')
points(enumerate(ber)).show(scale='semilogy')
points(enumerate(ber)).show(scale='loglog')

while the following (obtained by changing points to line) seem to work fine.

import mpmath
ber = [0.5*mpmath.erfc(i)/sqrt(2) for i in range(10)]
line(enumerate(ber)).show(scale='semilogx')
line(enumerate(ber)).show(scale='semilogy')
line(enumerate(ber)).show(scale='loglog')

[ppurka]

The problem seems more generic. This works:

sage: list_plot_semilogy([10^-i for i in range(3)])

This doesn't work:

sage: list_plot_semilogy([10^-i for i in range(4)])

So, the bug might be in points itself, or in matplotlib.

[ppurka]

About the second plot in comment:1 - this is fixed on adding #13528. Don't ask me why. ^_^

About the original question, the problem with the plot is that the point x=0 is present in the plot. This is a documented fail situation - look at the doc list_plot? or point2d?, both of which mention that non-positive points should not be used in the log scale. This is the reason why point works with semilogy scale and not with semilogx or loglog.

Instead, both these commands work:

sage: points(zip(*(range(1,len(ber)+1), ber))).show(scale='loglog')
sage: points(zip(*(range(1,len(ber)+1), ber))).show(scale='semilogx')

If you want to use the x=0 point, then use list_plot along with plotjoined=True, linestyle='', marker='o':

sage: list_plot(list(enumerate(ber)), plotjoined=True, linestyle='', marker='o').show(scale='semilogx')
sage: list_plot(list(enumerate(ber)), plotjoined=True, linestyle='', marker='o').show(scale='semilogy')
sage: list_plot(list(enumerate(ber)), plotjoined=True, linestyle='', marker='o').show(scale='loglog')

This latter command runs through the line command in Sage, which uses the usual plot of matplotlib. plotjoined=False runs through the point command of Sage, which uses the scatter command of matplotlib. So, the failure of x=0 in log scale is from the scatter command in matplotlib.

[ppurka]

Works for me (TM)