Skip to content

apjanke/fieldtrip-parallel-support

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 

Repository files navigation

fieldtrip-parallel-work

This repo contains work in progress for Andrew Janke's 2021 FieldTrip Parallelization project.

General Stuff

This is not intended to be a long-lived repo or software project in itself: it documents the work done for a specific work project in 2021. Useful code and results will be either upstreamed to FieldTrip itself or spun out to a separate repo, if they're considered sufficiently valuable.

The code here is intended to work as far back as Matlab R2016b, but was mostly tested on R2019b.

Repo Contents

Tutorial and Example Functions

The main contents of this repo are the code from the FieldTrip Tutorials converted to functions which can be run without arguments, and with a common environment configuration, to support benchmarking of the FieldTrip code.

These are effectively versions of the test_tutorial_*.m test scripts in the FieldTrip test/ directory which can be run outside of the environment of the Donders network.

These tutorial functions are structured so that they can be run without depending on your Matlab session's current working directory. This allows you to run them while you're cd'ed to their source code while working on them, or cd'ed to the FieldTrip repo to look at its code easily, or anywhere else. This means that the path references in the code have all been modified to replace relative paths with full paths using ft_tut_datadir (a new function introduced specifically for this scaffolding code).

I have left the tutorial code pretty much unmodified. I have not attempted to make the code Mlint-clean.

The tutorial functions do not return anything useful. If you want to see their results, you will generally want to put a Matlab debugger breakpoint at the final end of the function and then run it.

Running the tutorial functions may have the following side effects:

  • They may use blocking GUI functions, so manual interaction may be required to complete them.
  • They may leave new figures up.
  • They may write or delete any data under ft_tut_workdir.

The function sets are:

Usage

All the Matlab code is in the Mcode directory. Add that to your Matlab path. You'll also need FieldTrip itself.

The data used by the tutorial code is sourced from the FieldTrip FTP site at ftp://ftp.fieldtriptoolbox.org/pub/fieldtrip/tutorial/. (Note that some modern browsers do not support FTP, so you'll need a separate FTP client.) You'll need to download all the relevant data files for the tutorial code you want to run to a directory on your disk and unzip them. Then, to let this repo's tutorial code know where they are, run ft_tut_datadir('/path/to/my/datadir'). This data set is kind of large; like 45 GB once unzipped.

Some additional data used by some tutorial functions is found at ftp://ftp.fieldtriptoolbox.org/pub/fieldtrip/workshop. If you download this locally, set the path to it with ft_tut_workshopdir('/path/to/my/workshop/copy').

Author and Project

Developed by Andrew Janke.

The project website is the this apjanke/fieldtrip-parallel-work repo on GitHub. Andrew's work on parallelizing FieldTrip itself can be found on the apjanke/fieldtrip fork repo on GitHub on the apjanke/parfor-ify branch.

Coding powered by "The Roar of Love" by 2nd Chapter of Acts.

About

Support files for FieldTrip parallelization work

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages