[BUG] IndexError when plotting

[ryanhammonds]

This issue is came up here. Whenever no peaks are found and plot_peaks='line' is passed to the plot method of a fm, an IndexError is raised. It may be useful turn this error into a warning?

[TomDonoghue]

Is this still the case on main, or just 1.0.0? I think I fixed this recently. If it's still happening, then yeh, I don't think it should even be a warning, it should just proceed as normal, without annotating any peaks since there aren't any.

[ryanhammonds]

I tried from main too, and the same error was raised.

[TomDonoghue]

Ahh, it looks like I fixed it somewhere in #205 then, as this seems to work now on that branch.

Just to double check - this is what I ran:

from specparam import PSD
from specparam.sim import gen_power_spectrum

freqs, pows = gen_power_spectrum([3, 40], [1, 1], [1, 1, 1])

fm = PSD(max_n_peaks=0)
fm.fit(freqs, pows)
fm.plot(plot_peaks='line')

Which (on name) runs, and spits out a plot with no annotated peaks (since there are none) and no warning or error.
Does this seem like it addresses the error, and seem like good behaviour to you?

Sidenote: if I remember right, fixing this ended up being cleaning up how gauss_params_ gets initialized (or similar), so if it's empty it plays well with downstream processes like grabbing peaks for plotting. I don't think there are any relevant changes to the actual plot functions (in case you are having a look at what got changed).

[ryanhammonds]

I was wrong and this is fixed on main too! I was getting an unrelated error from using plot_style=None which seems to be deprecated on main.

Since this is already fixed, I'll go ahead and close this.

[TomDonoghue]

Ahh, okay cool that makes sense! I think the plot_style is deprecated since #176 got merged. Anyways, glad it should be fixed now - thanks for checking through!