Skip to content

Latest commit

 

History

History
58 lines (35 loc) · 2.89 KB

README.md

File metadata and controls

58 lines (35 loc) · 2.89 KB
Raw

Code for: Sequential Neural Posterior and Likelihood Approximation

This repo contains the code for the paper Sequential Neural Posterior and Likelihood Approximation arxiv-link.

The results presented in the secound Arxiv version of this paper where generated with the code at tag preprint2

Computer environment

The models were implemented using PyTorch utilizing the packages nflows and sbi.

The models were trained and run on the LUNARC computer system http://www.lunarc.lu.se/, and the results were analysed on a local computer.

System settings

LUNARC Local computer
Operating system CentOS Linux 7 Ubuntu 16.04
Python version 3.7.4 3.7.4
Package manager pip conda
Requirements env_lunarc.txt env_local.yml

Code structure

  • /algorithms - source code for the snpla method
  • /util - source code for some utility functions
  • /mv_gaussian - source code, run scripts, and notebooks for the MV Gaussian examples
  • /two_moons - source code, run scripts, and notebooks for the two-moons examples
  • /lotka_volterra - source code, run scripts, and notebooks for the Lotka-Volterra example
  • /hodgkin_huxley - source code, run scripts, and notebooks for the Lotka-Volterra example

The code for each experiment is structured as following:

  • The files functions.py and CaseStudy.py contain various classes and functions that defined the model
  • The run_script_"algorithm".py files are the run scripts
  • The notebook analysis.py is used to produce all analysis and plots
  • The *.sh files in the /lunarc folder are the scripts used to run the algorithms on the LUNARC system

Model simulator for the Hodgkin-Huxley model

We used the Neuron software (https://neuron.yale.edu/neuron/) to simulate the Hodgkin-Huxley model. The Neuron software was installed on our local computer, and all simulations and calculations for the Hodgkin-Huxley were carried out on our local computer. When simulating the Hodgkin-Huxley model, we utilized the same Neuron set up as in Sequential neural likelihood (http://proceedings.mlr.press/v89/papamakarios19a.html)

Data

The data used for all case studies can be generated from the code.

How to replicate the results

The results for case study C and algorithm A are computed by running the scripts A_main.sh and the A_main_h.sh scripts in /lunarc folder for case study C. The script A_main_h.sh will run the hyper-parameter search scheme and the script A_main.sh
will run the algorithm for the different data sets that are considered for case study C.

Acknowledgements

The computations were enabled by resources provided by the Swedish National Infrastructure for Computing (SNIC) at LUNARC (http://www.lunarc.lu.se/).