You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
in PESTPP-IES, I am trying the multimodal upgrade. I set alpha at 0.1 and have 1000 realizations, 1044 parameters and 245040 obs. I run PESTPP-IES in parallel with 1000 agents. I do not use localization. My model takes about 10-15 minutes for one run.
In this framework I find that the bottleneck as regards the computaiton time is by far the step where the .log file says :
'
2022-11-15 20:56:54,0.00357351,multimodal solve for inflation factor 1
2022-11-15 20:56:54,3.576e-05,multimodal upgrade using 100 realizations
2022-11-15 20:56:54,0.0108185,calculating multimodal upgrade for 0
2022-11-15 20:56:55,0.834885,upgrade thread progress: 0 of 1 parts done
2022-11-15 20:56:55,0.566427,calculating multimodal upgrade for 1
2022-11-15 20:56:56,0.699594,upgrade thread progress: 0 of 1 parts done
2022-11-15 20:56:56,0.562412,calculating multimodal upgrade for 2
2022-11-15 20:56:57,0.700173,upgrade thread progress: 0 of 1 parts done
...etc
'
Please find my .log file below as an example (N.B. I killed the optimization process before its end) case.log
Is there a way to speed up this step ? I think about multithreading, but I know it is normally used ofr localization so it may be hard to also use it here?
Thank you for your help.
Best regards,
The text was updated successfully, but these errors were encountered:
Sorry for the delay. This is possible but would require a specialized solver instance - currently its the same solver for localized solves being called repeatedly for the mutlimodal solves. But I'm using this option more and more also and it could be faster. I'll look into this but might be a bit before its ready for primetime...
Hello,
in PESTPP-IES, I am trying the multimodal upgrade. I set alpha at 0.1 and have 1000 realizations, 1044 parameters and 245040 obs. I run PESTPP-IES in parallel with 1000 agents. I do not use localization. My model takes about 10-15 minutes for one run.
In this framework I find that the bottleneck as regards the computaiton time is by far the step where the .log file says :
'
2022-11-15 20:56:54,0.00357351,multimodal solve for inflation factor 1
2022-11-15 20:56:54,3.576e-05,multimodal upgrade using 100 realizations
2022-11-15 20:56:54,0.0108185,calculating multimodal upgrade for 0
2022-11-15 20:56:55,0.834885,upgrade thread progress: 0 of 1 parts done
2022-11-15 20:56:55,0.566427,calculating multimodal upgrade for 1
2022-11-15 20:56:56,0.699594,upgrade thread progress: 0 of 1 parts done
2022-11-15 20:56:56,0.562412,calculating multimodal upgrade for 2
2022-11-15 20:56:57,0.700173,upgrade thread progress: 0 of 1 parts done
...etc
'
Please find my .log file below as an example (N.B. I killed the optimization process before its end)
case.log
Is there a way to speed up this step ? I think about multithreading, but I know it is normally used ofr localization so it may be hard to also use it here?
Thank you for your help.
Best regards,
The text was updated successfully, but these errors were encountered: