The BayesianCalibration with HMC often takes minutes to complete (and is significantly slower compared to e.g., metropolis sampling). At least some of this is to be expected as the method is doing a lot of additional computations. Nevertheless, it would be good to check what is taking the longest and ensure that there isn't an obvious bottleneck in our implementation (e.g., gradient computations) or other choice of default settings that might help speed things up.
This issue can be closed once profiling has been done even if the conclusion is that nothing needs changing.
The BayesianCalibration with HMC often takes minutes to complete (and is significantly slower compared to e.g., metropolis sampling). At least some of this is to be expected as the method is doing a lot of additional computations. Nevertheless, it would be good to check what is taking the longest and ensure that there isn't an obvious bottleneck in our implementation (e.g., gradient computations) or other choice of default settings that might help speed things up.
This issue can be closed once profiling has been done even if the conclusion is that nothing needs changing.