Skip to content
/ mfm Public

Code to reproduce experiments in Markovian Flow Matching: Accelerating MCMC with Continuous Normalizing Flows

Notifications You must be signed in to change notification settings

albcab/mfm

Repository files navigation

Code and data to replicate experiments in "Markovian Flow Matching: Accelerating MCMC with Continuous Normalizing Flows".

Run experiments:

example="4-mode"
echo "$example..."

python3 multi_modal.py --example $example --learning_iter 1000 --mcmc_per_flow_steps -1
python3 multi_modal.py --example $example --learning_iter 1000 --mcmc_per_flow_steps 1000
python3 multi_modal.py --example $example --learning_iter 1000 --mcmc_per_flow_steps 100
python3 multi_modal.py --example $example --learning_iter 1000 --mcmc_per_flow_steps 100 --hutch
python3 multi_modal.py --example $example --learning_iter 1000 --mcmc_per_flow_steps 10
python3 multi_modal.py --example $example --learning_iter 1000 --mcmc_per_flow_steps 10 --hutch
python3 multi_modal.py --example $example --learning_iter 1000 --mcmc_per_flow_steps 1
python3 multi_modal.py --example $example --learning_iter 1000 --mcmc_per_flow_steps 1 --hutch

python3 multi_modal.py --example $example --learning_iter 1000 --mcmc_per_flow_steps 10 --do_fab
python3 multi_modal.py --example $example --learning_iter 1000 --mcmc_per_flow_steps 10 --do_dds
python3 multi_modal.py --example $example --learning_iter 1000 --mcmc_per_flow_steps 10 --do_flowmc


example="gaussian-mixture"
echo "$example..."

python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps -1
python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 10000
python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 1000
python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 1000 --hutch
python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 100
python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 100 --hutch
python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 10
python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 10 --hutch

python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 10 --do_fab
python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 10 --do_dds
python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 10 --do_flowmc


example="phi-four"
echo "$example..."

python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 10000 --hutch
python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 1000
python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 1000 --hutch
python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 100
python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 100 --hutch

python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 10 --do_fab
python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 10 --do_dds
python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 10 --do_flowmc


example="pines"
echo "$example..."

python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 10000 --hutch
python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 1000
python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 1000 --hutch
python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 100
python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 100 --hutch

python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 10 --do_fab
python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 10 --do_dds
python3 multi_modal.py --example $example --learning_iter 10000 --mcmc_per_flow_steps 10 --do_flowmc

About

Code to reproduce experiments in Markovian Flow Matching: Accelerating MCMC with Continuous Normalizing Flows

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages