[AstroPy] matplotlib histograms and astropy quantites

[John K. Parejko]

On 27Jul 2016, at 05:34, federica <fb55 at nyu.edu> wrote:

> This a numpy version compatibility issue with versions 1.9.x of numpy:
> 
> https://github.com/numpy/numpy/issues/6469

This appears to be a different bug, and I'm on numpy 1.10.4.

> On Wed, Jul 27, 2016 at 5:15 AM, Daniel Evans <d.f.evans at keele.ac.uk> wrote:
> 
> My version of astropy is far too ancient to have this feature (I should bug my sysadmin!) so I can't test it myself, but does calling astropy.visualization.quantity_support() fix your errors?

Ah, thanks! It does get rid of the exception, and produce somewhat more helpful axis labels. I'll have to poke around some to fully understand how the units interact with the plot windows, but I think I'm good now.

Thanks a bunch,
John

> On 27 July 2016 at 10:00, John K. Parejko <parejkoj at uw.edu> wrote:
> I wonder if I'm misusing astropy quantities, or if there's a bug with how it interacts with matplotlib's pyplot.hist (and thus np.histogram). I've attached a minimal working example that represents a histogram I was attempting to create. If you try to make the plot range be a unit, plt.hist will fail. I would have thought that I could specify a range with units and get the plot axis to match the units I want to show the data with.
> 
> Also, trying to draw a vertical line with axvline() on a Quantity fails, though I guess I shouldn't be surprised at that. But given the above, this was also surprising to me.
> 
> I feel like there could be some more documentation about the interaction between astropy quantities and matplotlib?
> 
> Am I missing something here?
> 
> Thanks in advance,
> John
> 
> 
> import numpy as np
> from astropy import units as u
> import matplotlib.pyplot as plt
> 
> range1 = (-3, 3)
> range2 = range1*u.arcsecond
> np.random.seed(100)
> data = np.random.normal(scale=3e-6, size=100)*u.radian
> plt.figure()
> # Not well centered and wrong X scale.
> plt.hist(data)
> # Works: I guess I shouldn't be surprised that I have to extract a value here.
> plt.axvline(x=data[0].value, color='red')
> plt.figure()
> # Wanted the range in arcseconds, so this isn't right!
> plt.hist(data, range=range1)
> # Dimensionalizing it seems to work correctly.
> plt.axvline(x=data[0]/u.arcsecond, color='red')
> plt.figure()
> # UnitsError: Can only apply 'less_equal' function to dimensionless quantities...
> plt.hist(data, range=range2)
> # UnitsError: Can only apply 'greater' function to dimensionless quantities...
> plt.axvline(x=data[0])
>