lucianopaz/sensory_integration_time

This repository contains the code used to fit the models described in "A sensory integration account for time perception" (submission in progress to eLife).

sensory_integration_time

This package constains the source code used to fit the leaky integration model for tactile perception used in the Toso et al (submitted).

Installation

This package contains a custom c extension that uses a fortran library (fortran source codes are included in this package). In order to install, your system must have

1. A fortran compiler
2. The GNU scientific library (GSL) must be installed and available in the build environment.

Once you have those, you should be able to simply install this package by downloading its source code and running

pip install path_to_downloaded_package

To try to install everything in a single go, you can try to use conda like this:

conda create -n env_name python=3.7 gsl some_fortran_compiler_or_toolchain
./scripts/add_env_vars.sh -n env_name
conda activate env_name
pip install -r path_to_downloaded_package/requirements.txt
pip install path_to_downloaded_package

or use the bash script

scripts/install_in_conda.sh -n env_name

Be aware that conda can mess up pip on macos, the fortran compilers on windows are in non-default channels and once gsl is installed, its headers and lib directories are not added to environment variable. The bash script add_env_vars.sh, attempts to solve the last problem mentioned. The supplied install_in_conda.sh script attempts to solve all of these problems. It was tested and works in a local ubuntu desktop, but does not work in the azure-pipelines CI, so use it with a pinch of salt.