Parameter estimation for complex physical problems often suffers from finding ‘solutions’ that are not physically realistic. The PEUQSE software provides tools for finding physically realistic parameter estimates, graphs of the parameter estimate positions within parameter space, and plots of the final simulation results.
13
stars
5
forks
source link
Get continueSampling to work with doe (and parallel processing of doe) #154
MH MCMC currently waits until the very end until writing to file. There should be an option to write immediately. Currently, when a simulation crashes due to server crashing, it is not possible to continue sampling.
As of Oct 2020 , continueSampling is not compatible with doe.
It should be possible with the parallel processing of conditions because that exports files like this: conditionsPermutationAndInfoGain_mod1_cond3
In the example (runfile_Example13_doe_function_meshgrid_1_parallel_conditions_exploration.py), there are 26 process ranks (of which 25 are used for simulations) and each modulation for each condition has a separate rank.
So it should be 'relatively' straightforward to add an if statement at the top of MetropolisHastings to load the appropriate modulation and condition mcmc samples and continue from there.
MH MCMC currently waits until the very end until writing to file. There should be an option to write immediately. Currently, when a simulation crashes due to server crashing, it is not possible to continue sampling.
We have already gotten parallel processing to work with doe (as noted in https://github.com/AdityaSavara/CheKiPEUQ/issues/8 ) along with continueSampling
As of Oct 2020 , continueSampling is not compatible with doe.
It should be possible with the parallel processing of conditions because that exports files like this: conditionsPermutationAndInfoGain_mod1_cond3 In the example (runfile_Example13_doe_function_meshgrid_1_parallel_conditions_exploration.py), there are 26 process ranks (of which 25 are used for simulations) and each modulation for each condition has a separate rank.
So it should be 'relatively' straightforward to add an if statement at the top of MetropolisHastings to load the appropriate modulation and condition mcmc samples and continue from there.