Bumps design

[RossBoylan]

Several things seem odd about the current bumps design. This ticket is to see if there is a good reason for any of them.

1. Maximum 10 bumps per type. Seems completely arbitrary. See

   bifs/bifs/priors/AbstractPrior.py

   Line 195 in d4a786a

   raise RuntimeError("No more than 10 bump filters of a type allowed. Exceeded for type "+key_orig)

   for the error message; the code above it hardcodes looking for 10.
2. Why are keys being created with the original key, a dot, and a counter?

   bifs/bifs/priors/AbstractPrior.py

   Line 190 in d4a786a

   new_key = key_orig+"."+str(i)

3. It is not entirely clear to me if the order in which one applies the bumps matters. If each transforms the k-space representation then they would be order dependent. I think that is not the case because the code applies each bump to the original data, x, and adds up the results. Can anyone confirm?

It seems to me a more natural data structure would be to make the bumps a list of Bump objects, with each object having all relevant info including the signal key.

[kyoung21b]

Het Ross, Sorry to continue my general delinquency re. BIFS issues but I’ve essentially been on another planet lately and having a hard time recalling the justification for any of my original choices (I know the combination of lousy code combined with poor documentation re. something I considered a rough prototype isn’t making life easy for you). What I recall about the maximum bump limit is something like that I wanted to allocate a bump array and just choose some ridiculously high limit re. bumps (if one is using more than a couple of bumps in k-space that would make over fitting look like a conservative strategy). At this point I’m not sure why I was concerned about a limit; have to go back and look. Presumably one could just append as many bumps as one wanted so it must have been something in the down the line processing that caused me to set a limit. I’ll have to take the proverbial look at the code… — KY Karl Young

…

On Feb 22, 2022, at 8:31 PM, RossBoylan ***@***.***> wrote: Several things seem odd about the current bumps design. This ticket is to see if there is a good reason for any of them. Maximum 10 bumps. — Reply to this email directly, view it on GitHub <#33>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AFTQY7VCP5X5C5HO4CMXWP3U4RPKJANCNFSM5PDIJK3Q>. Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. You are receiving this because you were assigned.