Skip to content

popt⚓︎

Optimisation methods.

Python Optimization Problem Toolbox (POPT)⚓︎

POPT is one part of the PET application. Here we solve the optimization problem. Currently, the following methods are implemented:

  • EnOpt: The standard ensemble optimization method
  • GenOpt: Generalized ensemble optimization (using non-Gaussian distributions)
  • SmcOpt: Gradient-free optimization based on sequential Monte Carlo
  • LineSearch: Gradient based method satisfying the strong Wolfie conditions

The gradient and Hessian methods are compatible with SciPy, and can be used as input to scipy.optimize.minimize. A POPT tutorial is found here.