Residual errors are 0

[ZvejysMegejas]

Hi,

I'm a new user of Rascal. Trying to use it to calibrate the spectrometer. After messing around with custom atlas lines and all other setting I managed to get satisfactory line matches. However I noticed that the Residual error and RMS were shown as zeros in c.plot_fit. I looked in the code of the plot.fit in plotting.py, and I have feeling that the error sits over there. As I'm a new Python user as well, I can not pinpoint where is that exactly as the code for the Residual Error calculations seems complicated for me. A simpler code gives "proper" numbers for Residual error.

peaks_angstr = c.polyval(matched_peaks, fit_coeff1)
diff = []

for i, j in zip(peaks_angstr, matched_atlas):

diff.append(i - j)

print(diff)

# RMS calculations

rms = np.sqrt(np.mean(np.array(diff) ** 2.0))

print(rms)

[cylammarco]

Hi @ZvejysMegejas, would you please share the code and the full error message?

[ZvejysMegejas]

Please find at https://github.com/ZvejysMegejas/rascalcodeexample
Three files: the code, the data, and the result of 'c.plot_fit'

I noticed that if plots are send to a browser (plotly feature?), then the residual error is shown correctly. However, if plotting is done with pyplot (?), the error is flat.

I'll try to write my own code with 'plotly' using your examples.

[cylammarco]

I noticed that if plots are send to a browser (plotly feature?), then the residual error is shown correctly. However, if plotting is done with pyplot (?), the error is flat.

Thanks, that's a really good place to start!

[cylammarco]

This is also fixed in the dev-0.3.X branch.

[ZvejysMegejas]

This is also fixed in the dev-0.3.X branch.

Thus, I downloaded a new file. Now I'm getting a funny result :)

Labels and seems positions are used twice now.

[ZvejysMegejas]

Using
c.use_plotly()

The labels are shown too however the lines are not used in the fitting.

[ZvejysMegejas]

OK, my bad.
In my file, the spectral lines had a missing digit:
435.8335 was used as 35.8335.
[404.6565, 407.7837**, 35.8335,** 491.6068 ...]

With an improved atlas, still the plot does not act properly.

[cylammarco]

Ha! @jveitchmichaelis, is that an old bug creeping back or a wontfix in v0.3?

[jveitchmichaelis]

Did we figure out this one @cylammarco ?