The tutorials in this directory cover many optimization domains. Together they show how fcmaes can be used for practical gradient-free problems in engineering, science, planning, simulation, and machine learning.
Dimensional Analysis
-
BuckinghamExamples Compute π-groups for dimensional analysis.
Quality Diversity
-
Quality Diversity Run a multi-algorithm improvement emitter in parallel with CVT MAP-Elites.
-
Map Elites Combine CVT MAP-Elites with an adapted CMA-ES.
-
Map Elites 2 Apply quality diversity to ODE-based control problems.
Space Flight Trajectory Design
-
Introduction Solve simple example problems from space flight dynamics.
-
Space Flight Revisit space flight mission design.
-
Pagmo results GTOPX benchmark results using Pagmo.
-
Pykep gym results Benchmark results for the Pykep gym problems.
-
Mixed Integer Mixed-integer flight mission design.
-
ESAChallenge The ESA Optimization Challenge.
-
Spherical t-design Weighted spherical t-design.
Quantum Computing
-
Variational Qubit Optimize a variational qubit and a variational quantum eigensolver.
Service - Demand Network Planning (p-center problem)
-
5G network planning Solve the p-center problem for irregular shapes with holes.
-
Service Locations Solve the α-neighbor p-center optimization problem.
Scheduling
-
Employee Scheduling Optimize an employee schedule.
-
JobShop Solve the flexible job shop scheduling problem.
-
Scheduling Solve a complex scheduling problem from the GTOC11 competition.
Routing
-
One For All Work on MMKP and VRPTW problems.
-
Multi-UAV Solve a multi-UAV task assignment problem.
-
Routing Capacitated Vehicle Routing.
-
Noisy TSP Solve the noisy Traveling Salesman Problem.
Trading
-
Crypto Trading Optimize a crypto trading strategy.
Water Management
-
Water Management Optimize water resource management.
Social Media
-
Social Media Analyze social media user data.
Game Design
-
gbea TopTrumps Benchmark The TopTrumps game optimization benchmark.
Simulation
-
Moran Process Optimize a Moran process, a real case from aging research.
-
Parameter Sweep Optimize parameters of a biochemical stochastic model.
-
Surrogate Optimize the Mazda car design problem.
-
Robot Pushing and Navigating Rovers Robot pushing and rover navigation.
-
CFD Optimize problems based on CFD simulations.
-
Power Plant Power plant efficiency.
-
Vaccination Model vaccination.
-
Hospital Managing hospital resources during a pandemic.
Machine Learning
-
EvoJax Hardware-accelerated neuroevolution.
-
Hyper Parameters Hyperparameter optimization.
-
Optimization Assistant An open source AI optimization assistant.
Miscellaneous
-
Subset selection Select an optimal subset.
-
Clustering Out-of-the-box clustering.
-
ODE Use differential equation solvers.
fcmaes-Details
-
MO-DE MO-DE, a new multi-objective algorithm.
-
Multi-Objective Solve multi-objective problems using variable weights.
-
Constraints Optimize with constraints.
-
Expressions Use sequences and random choices of optimizers.
-
Delayed Update Asynchronous parallel function evaluation.
Log output of the parallel retry
The log output of the parallel retry contains the following rows:
Parallel retry
-
time (in sec)
-
evaluations / sec
-
number of retries - optimization runs
-
total number of evaluations in all retries
-
best value found so far
-
mean of the values found by retries below the defined threshold
-
standard deviation of the values found by retries below the defined threshold
-
list of the best 20 function values in the retry store
-
best solution (x-vector) found so far
Mean and standard deviation would be misleading when using coordinated retry because some retries are initiated by crossover. Therefore, the rows in the log output differ slightly:
Smart retry
-
time (in sec)
-
evaluations / sec
-
number of retries - optimization runs
-
total number of evaluations in all retries
-
best value found so far
-
worst value in the retry store
-
number of entries in the retry store
-
list of the best 20 function values in the retry store
-
best solution (x-vector) found so far