Releases: AdityaSavara/PEUQSE
Version 0.4.6
The contour plot settings have been standardized into contour_plot_settings in UserInput.
Version 0.4.5
Added simulatedResponses bounds feature so that a person can intentionally restrict their posterior to only physically realistic regions. This is desirable for optimizing surrogate models, for example.
Made multistart more robust against errors when only non-finite probability points are found.
Version 0.4.3
continueSampling has been added for multiStart uniform distribution getLogP
Version 0.4.2
Exporting permutations to samples statistics to log file improved & changed things so that the 'auto' choice for continueSampling will default to False, to prevent unexpected behavior.
Version 0.4.0
This is one of the most important upgrades. Now posterior distributions can be obtained from multistart gridsearch and multistart uniform distribution searches with getLogP. This is a feature that pymc3 does not have in a readily accessible manner. Here, it is in example 00f.
Version 0.3.2
Warns user if the simulated responses and observed response have different shapes, casts the scalars to arrays if they are numbers, and also includes automatic abscissa generation when an abscissa is not provided.
Version 0.3.0
This is a major release. We now have continued sampling for ESS and for MH, and the continued sampling works for parallelized runs as well (with MPI).
Version 0.2.0
Parallelized design of experiments has been added.
Staggered responses are now completely supported. A significant update of the internal data structures has been made in this release. This has also removed some of the internal 'casting' and 'case specific' indexing.
Version 0.1.2
A major update: now MPI is available for parallel sampling during mcmc.
Version 0.1.1
Improvements in the implementation of gridsearch and mcmc.