Plotting histogram with 1d data when fig argument is given

[SuperKam91]

Hi, When trying to plot two sets of 1d histograms (1d data xs1 and xs2, with weights weights1 and weights2) on the same figure e.g.

fig = corner.corner(xs1, weights1) fig = corner.corner(xs2, weights2, fig = fig)

EDIT: I think this bug generalises to plotting 1d histograms and supplying a fig argument, rather than the more specific case of plotting two 1d histograms on the same figure

on the second call to corner.corner I get an error in corner.py on line 239:

n, _, _ = ax.hist(x, bins=bins[i], weights=weights, range=np.sort(range[i]), **hist_kwargs)

that the object ax is a np.array not an axes object. This is because the proceeding statement at line 233 is:

if np.shape(xs)[0] == 1: ax = axes else: ax = axes[i, i]

and sets ax = axes in the case of 1d data. But axes is an array in the case of supplying corner.corner() a fig argument since in line 206 we have:

if fig is None: fig, axes = pl.subplots(K, K, figsize=(dim, dim)) else: try: axes = np.array(fig.axes).reshape((K, K)) except: raise ValueError("Provided figure has {0} axes, but data has " "dimensions K={1}".format(len(fig.axes), K))

However I think I rectified this buy by replacing the conditional at line 233 with: try: ax = axes[i, i] except TypeError: ax = axes

Hopefully, this does not introduce new bugs. From what I've seen it looks ok but I have in no way tested it thoroughly.

Thanks

Kamran

[dfm]

Thanks for this. We definitely don't test for the 1-D case, but it would be good to support it.

If anyone feels up to it, please write a test and the fix for this bug. If not, I'll get to it at some point!