This is a direct port of a C++ routine by Jonathan Goodman (NYU) called ACOR that estimates the autocorrelation time of time series data very quickly.

Dan Foreman-Mackey (NYU) made a few surface changes to the interface in order to write a Python wrapper (with the permission of the original author).

Just run

pip install acor

with sudo if you really need it.

Otherwise, download the source code as a tarball or clone the git repository from GitHub:

git clone https://github.com/dfm/acor.git

Then run

cd acor
python setup.py install

to compile and install the module acor in your Python path. The only dependency is NumPy (including the python-dev and python-numpy-dev packages which you might have to install separately on some systems).

Given some time series x, you can estimate the autocorrelation time (tau) using:

import acor
tau, mean, sigma = acor.acor(x)

• http://www.math.nyu.edu/faculty/goodman/software/acor/index.html
• http://www.stat.unc.edu/faculty/cji/Sokal.pdf