Closed rsibille-psi closed 3 years ago
Currently, the errors are calculated as sqrt(n)
, where n
is the number of counts. The number of counts scales linearly to a monitor value. What would be the right way to scale errors in that case? I guess, sqrt(original_n) * monitor_scale
?
yes, with monitor_scale the ratio between the original and new monitors. thank you very much.
let me know if it works, then I will update the main server too
yes it does work properly I think. Thank you!
From: Ivan Usov @.> Reply-To: paulscherrerinstitute/pyzebra @.> Date: Wednesday, 29 September 2021 at 16:54 To: paulscherrerinstitute/pyzebra @.> Cc: "Sibille Romain Franck (PSI)" @.>, Author @.***> Subject: Re: [paulscherrerinstitute/pyzebra] monitor normalisation and error bars (#40)
let me know if it works, then I will update the main server too
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHubhttps://github.com/paulscherrerinstitute/pyzebra/issues/40#issuecomment-930254480, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AO23BBPZDMSCDFA7DX7ULLLUEMSBTANCNFSM5E4YJF7Q. [ { @.": "http://schema.org", @.": "EmailMessage", "potentialAction": { @.": "ViewAction", "target": "https://github.com/paulscherrerinstitute/pyzebra/issues/40#issuecomment-930254480", "url": "https://github.com/paulscherrerinstitute/pyzebra/issues/40#issuecomment-930254480", "name": "View Issue" }, "description": "View this Issue on GitHub", "publisher": { @.": "Organization", "name": "GitHub", "url": "https://github.com" } } ]
The error bars are not properly normalised in pyzebra it seems. For instance, I open dat file 6349 of experiment 20211272, the natural monitor is 6 millions. The plot produced by pyzebra is normalised to monitor 100000 with unrealistically large error bars (image 1), while if I impose the natural monitor to pyzebra the error bars look correct (image 2) and consistent with our old "fit" tool (image 3).