Add option to save first samples of peak(lets) waveform #867
Conversation
for more information, see https://pre-commit.ci
There was a problem hiding this comment.
Thanks, @HenningSE . The PR looks good I would add comments about this artificial threshold 6 of downsample_factor.
for more information, see https://pre-commit.ci
for more information, see https://pre-commit.ci
There was a problem hiding this comment.
Thanks @HenningSE . If you are sure this PR works well with your proposed change in straxen, please merge it after autotest finishes.
|
Thanks @dachengx, it looks like I'm not an authorized user and can not merge. |
| p["data_start"] = wv_buffer[: len(p["data_start"])] | ||
| else: | ||
| p["data_start"][: p["length"]] = wv_buffer[: p["length"]] | ||
|
|
||
| else: |
There was a problem hiding this comment.
Maybe here you should also assign data_start.
There was a problem hiding this comment.
Hi @dachengx, I think this case is not needed as we do not downsample the waveform. The data field contains all needed information.
There was a problem hiding this comment.
But if you look at data_start, it will be zeros or np.nan. People might misunderstand it.
What is the problem / what does the code in this PR do
This PR adds the option to save the first n (default to 200) samples of the merged peak(lets) waveform without downsampling. In case of very wide S1s or S1s with a lot of afterpulses strax will downsample the peak waveforms to fit into the 200 sample summed waveform that we save with peaklets and peaks. This process removes important information of the S1 waveform. With this PR the first n samples of the waveforms are saved for peaklets that 1. get downsampled and 2. have a downsampling factor smaller 7. This should cover basically all S1s without introducing a large file size overhead.