oscillation tests #6
Conversation
…look up table of oscillation probabilities to be loaded.
mastbaum
left a comment
mastbaum
left a comment
There was a problem hiding this comment.
Hi @kgilje! Thanks so much for putting this together, it looks great. I just had a couple of small questions/comments throughout. One overall question is whether it'd be possible to add a flag to use 1-Pee, to oscillate the numus? B8 numu ES isn't big (and the variations here are small, on top of that!) but it'd be nice to be consistent. Let me know what you think. Testing now and will let you know if I run into any issues!
| // energy: min-8, max-9, nstep-10, step-11 | ||
| double energy = fields[syst->obs]; | ||
| // Skip crazy energies seen in setup. | ||
| if (energy > 40) return; |
There was a problem hiding this comment.
Are there neutrinos with energies higher than the endpoint?
There was a problem hiding this comment.
For the first setup of fake data I was getting segfaults. I tracked it down to dummy values being passed during the setup of the histograms. The easiest way for me to avoid was to add the energy > 40 remove.
| int &reject_event, RNGState* rng) { | ||
| if (pee_osc_lut_size == 0) { | ||
| printf("pdfz::apply_systematic: Something is seriously wrong with the look up table!\n"); | ||
| return; |
There was a problem hiding this comment.
Since there are some really bad conditions here that return, I wonder if it'd be worth setting some error flag that gets checked when samples are periodically copied back to the host memory, and asserting at that point.
There was a problem hiding this comment.
I'm no 100% sure how to do that? Do you have any sample code?
|
|
||
| void EvalHist::EvalAsync(bool do_eval_pdf) { | ||
|
|
||
| // Create a random number generator to pass to oscillation systematic if needed. |
There was a problem hiding this comment.
Does this need to happen for every call of EvalHist::EvalAsync, or just once e.g. in the EvalHist constructor?
There was a problem hiding this comment.
I tried to do this, but got segfaults. I'm chalking that up to me not knowing how to code GPU style. This may be messy but it works...?
2 participants
The majority of the oscillation commits are in: src/pdfz.cpp src/pdfz.h src/signal.cpp src/systematic.cpp src/systematic.h src/osctable.cpp src/osctable.h
I think this caught a few missing updates from synching with Andy's local commits rather than via the general sxmc on github