Skip to content
Configuration model for correlation/covariance matrices
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
matlab
python
DMCC-algorithm.pdf
README.md
motivation.txt

README.md

config_corr

Configuration model for correlation/covariance matrices

This package contains two algorithms.

  • Algorithm DMCC solves a convex optimization problem and usually better than the other naive gradient descent algorithm. Therefore, we recommend DMCC over the naive gradient descent algorithm.
  • DMCC, including the performance test, is documented in the accompanying DMCC-algorithm.pdf.
  • DMCC is provided in Python only.
  • The naive gradient descent algorithm is provided in Python and MATLAB.

Install using pip

pip install configcorr

Install from source

Package configcorr has the following dependencies:

Python >= 3.4
setuptools >= 1.4
NumPy >= 1.8
CVXPy >= 1.0

These packages need to be installed manually.

Clone the configcorr git repository by running

git clone https://github.com/naokimas/config_corr.git

Then, navigate to the top-level of the cloned directory and run

python setup.py install

DMCC algorithm in Python

python3 test_dmcc.py
  • test_dmcc.py calls max_ent_config_dmcc.py

Naive gradient descent algorithm in Python

python3 test_naive_gradient_descent.py
  • test_naive_gradient_descent.py calls max_ent_config_naive_gradient_descent.py

Naive gradient descent algorithm in MATLAB

test_naive_gradient_descent
  • test_naive_gradient_descent.m calls max_ent_config_naive_gradient_descent.m
  • The data set to be used should be specified within test_naive_gradient_descent.m
  • Variable curr_dir should be modified according to where you place the data set.
You can’t perform that action at this time.