tplotxy (or similar tool) should work with 1-D non-timeseries data

[jameswilburlewis]

Vassilis had to drop down to matplotlib in one of his homework solutions, which involved a log-log plot of power vs frequency for a particular variable:

import matplotlib.pyplot as plt
freq = thd_scw_fac_z_clip_pwrspc.v[0,:]   # frequency (Hz)
psd  = thd_scw_fac_z_clip_pwrspc.y[0,:]   # power spectrum

plt.figure(figsize=(6,4))
plt.loglog(freq, psd)

plt.xlim(1., 4e3)
plt.ylim(1e-10, 1e-5)

plt.xlabel('f [Hz]')
plt.ylabel('thd_scw_fac_z_pwrspc [nT²/Hz]')
plt.title('Quiet time dynamic power spectrum')

plt.grid(True, which='both', ls='--', alpha=0.4)
plt.show()

tplotxy could probably be modified to handle this, but we'd want some extra features:

• allow passing X and Y as arrays instead of tplot variables
• options for log scaling on both X and Y axes
• coordinate grids on/off/transparency/linestyle

It's also worth thinking about how we might support non-time-series data in store_data/get_data. Right now we assume the X coordinate is always time, it always gets converted to np.datetime64 going in and Unix seconds coming out, and these assumptions are baked in throughout the code.