Skip to content

v1.03

Choose a tag to compare

View all tags
@VMLC-PV VMLC-PV released this 14 Aug 10:47
· 52 commits to main since this release
v1.03
9de06ea

v1.03 - 2025-08-14 - VMLC-PV

  • Upgrading the ax-platform version to 1.0.0.
  • axBOtorchOptimizer: Removing the optimize(batch=True) that used the runner in older versions of optimPV. Mostly because it did not provide any significant benefits and added unnecessary complexity and that the new ax-platform broke the compatibility with the old code. Multiple small fixes to the code to make it compatible with the new ax-platform version 1.0.0. Removed the parallel_agent option and replaced it with a parallel option that will run everything in parallel.
  • PymooOptimizer: Initial implementation of the PymooOptimizer class for single and multi-objective optimization using the pymoo library. This class provide new optimization capabilities withe the evolutionary/genetic algorithms and other algorithms available in the Pymoo library. The class is designed to have a similar interface to the AxOptimizer class, allowing for easy integration with existing code. The class supports both single and multi-objective optimization, and can be used with the same parameters and agents as the AxOptimizer class. The class also supports parallel optimization using the Pymoo library's parallelization capabilities. However, for not it only support float as value_type for the parameters.
  • posterior: added the 'pymoo' optimizer as an options for the plot_density_exploration function. This allows to plot the density exploration of the optimization process using the Pymoo library's optimization results.
  • New Notebooks and created tests for the new functionalities.
  • logger: Added logger inspired but the one in the Ax library.
  • SuggestOnlyAgent: Added the SuggestOnlyAgent class to provide a simple way to suggest new parameters without running the simulation/experiment and using a known dataset. This is useful when doing design of experiments (DoE) in multiple steps. It can be used with the axBOtorchOptimizer and the PymooOptimizer classes to suggest new parameters based on the previous results.
  • EGBO: Updated the EGBOAcquisition class to work with the new ax-platform version 1.0.0.
  • RateEqAgent: small fixes for the case Gfracs = None we where not taking the right time axis to take into account the pump frequency, this was done properly when we had several Gfracs.
  • RateEqModel: Added two new model based on the work by Kober-Czerny et al. 2025 and M. Simmonds.
  • Pumps: fixing the initial_carrier_density to convert the N0 into proper generation rate.
Assets 2
Loading