View source on GitHub
|
TensorFlow Probability Optimizer python package.
Modules
convergence_criteria module: TensorFlow Probability convergence criteria for optimizations.
linesearch module: Line-search optimizers package.
Classes
class StochasticGradientLangevinDynamics: An optimizer module for stochastic gradient Langevin dynamics.
class VariationalSGD: An optimizer module for constant stochastic gradient descent.
Functions
bfgs_minimize(...): Applies the BFGS algorithm to minimize a differentiable function.
converged_all(...): Condition to stop when all batch members have converged or failed.
converged_any(...): Condition to stop when any batch member converges, or all have failed.
differential_evolution_minimize(...): Applies the Differential evolution algorithm to minimize a function.
differential_evolution_one_step(...): Performs one step of the differential evolution algorithm.
lbfgs_minimize(...): Applies the L-BFGS algorithm to minimize a differentiable function.
nelder_mead_minimize(...): Minimum of the objective function using the Nelder Mead simplex algorithm.
nelder_mead_one_step(...): A single iteration of the Nelder Mead algorithm.
proximal_hessian_sparse_minimize(...): Minimize using Hessian-informed proximal gradient descent.
proximal_hessian_sparse_one_step(...): One step of (the outer loop of) the minimization algorithm.