qq compare command to compare two qibocal reports #714
Conversation
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #714 +/- ##
========================================
Coverage 97.60% 97.61%
========================================
Files 119 121 +2
Lines 9260 9417 +157
========================================
+ Hits 9038 9192 +154
- Misses 222 225 +3
Flags with carried forward coverage won't be shown. Click here to find out more.
|
|
Missing |
|
Hi @GabrielePalazzo, what is the status of this PR ? |
Comparison of single sweeper experiments (like resonator spectroscopy) works. |
I agree. If you want, you can just put the plots of the two reports near each other. |
I was going to suggest the same. Another possible idea would be to come up with some sort of ratio plot to see the differences between the two outcomes. I don't know how feasible it is. |
|
The new code is not covered by tests. I'll try to write some. |
|
@GabrielePalazzo, could you resolve the conflicts? Thanks |
For conflicts let me know if you need help. I recently changed some stuff related to the report generation. |
There was a problem hiding this comment.
Thanks @GabrielePalazzo I have a few comments.
I start playing with qq compare and I have noticed that if I combined together two monitoring report it feels very slow: http://login.qrccluster.com:9000/nmg4AkeKRuCBKYp3M-NnrA==/
I think because of the fact that the HTML along is 20 MB. We don't have to address it here but I was wondering if there could be a way to produce less heavy reports. Any ideas @alecandido?
| if any(isinstance(trace, go.Heatmap) for trace in fig0.data): | ||
| # TODO: check if this is valid for all protocols |
There was a problem hiding this comment.
There are still problems with same 2D plots like Chevron, flux dependence or readout characterization. Of course we cannot have something that works for everything. It would be nice to support the readout characterization out of the box given that it would used for the monitoring most likely.
If it is too much of an issue we can postpone it for a later PR.
There was a problem hiding this comment.
I'll have a look. In case it takes too long, we can do it later.
There was a problem hiding this comment.
Ok, actually I would say that if it works just for readout_characterization it might be better because as we already discussed people will not compare much Heatmaps. I can also help with that.
GabrielePalazzo
requested review from
Edoardo-Pedicillo,
andrea-pasquale and
alecandido
I'm pretty sure that the size of the HTML is all driven by plots (the rest of the page is incredibly minimal). To be fair, even However, regarding performances, I'd really advise putting here the threshold: this is definitely not the best way to optimize, and a black hole in terms of time required to improve. |
|
Another thing @GabrielePalazzo, you could add just a couple of sentences in the documentation that would be perfect. |
Example of a combined report containing a resonator spectroscopy and a resonator flux dependence:
http://login.qrccluster.com:9000/T5rbRzsmTRmKLtOSaa03DQ==
Another example:
http://login.qrccluster.com:9000/o73U5W5dTuKIjl_8wCkTzg==/
Only experiments with the same id are compared. All other experiments are discarded.
Tables contain results from both experiments side-by-side.
Experiments with scatterplots (e.g.: resonator spectroscopy) are shown in the same figure (with overlapping points and lines).
Plots of experiments with heatmaps (e.g.: resonator punchout) are vertically stacked.
Checklist:
mastermainmain