CCL INTEGRATE failure to determine standard deviations #74

New Issue

2026-02-10T14:54:50+01:00

zaharko commented

2026-02-10 14:54:50 +01:00

Sometimes the fit algorithm fails to determine estimated standard deviations.
the program should
a) do a check, if any estimated standard deviation for a given fit is NAN,
b) in this case shift the initial peak position by 0.1 deg and refine once more

Sometimes the fit algorithm fails to determine estimated standard deviations. the program should a) do a check, if any estimated standard deviation for a given fit is NAN, b) in this case shift the initial peak position by 0.1 deg and refine once more

usov_i commented

2026-05-13 16:03:34 +02:00

There are a few edge cases to consider here:

what if there's no fit function component with a peak
what if the initial peak is not provided
how many time to refine in case there's still no good fit

What, I think, might be better is to highlight scans that don't have a good fit on the UI, so seeing and fixing them manually would be more convenient. I don't think that we can reach 100% automated fit quality, so proceeding without a manual check won't be possible anyways.

There are a few edge cases to consider here: 1) what if there's no fit function component with a peak 2) what if the initial peak is not provided 3) how many time to refine in case there's still no good fit What, I think, might be better is to highlight scans that don't have a good fit on the UI, so seeing and fixing them manually would be more convenient. I don't think that we can reach 100% automated fit quality, so proceeding without a manual check won't be possible anyways.

usov_i commented

2026-05-13 16:24:54 +02:00

Another suggestion is to refine the fitting constrains. For example, I noticed that we don't impose any restriction on sigma for peak-like functions. However, it would make sense to set: min = 0. Maybe there are other rules we can initially set that can be applied for all data.

Another suggestion is to refine the fitting constrains. For example, I noticed that we don't impose any restriction on `sigma` for peak-like functions. However, it would make sense to set: min = 0. Maybe there are other rules we can initially set that can be applied for all data.

Sign in to join this conversation.

2 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: zebra/pyzebra#74