# Activate the --kurtdir flag

This is code originally from @saad and edited by me. It extends the previous --kurt flag to compute the kurtosis along the 3 eigenvalues (i.e., allows the eigenvalues to change linearly as a function of b-value). This is a simplified version of the kurtosis model, which also allows deviations from a diffusion tensor model within a single shell (i.e., for a single b-value).

To prevent the kurtosis from going to +/-infinite, when the diffusion is very small, a prior is added to the kurtosis with a mean of 0 and variance of 1.

I'll open a seperate pull request to pyfeeds-tests to add some tests.

TODO: enable --wls flag to work with --kurtdir (currently it crashes) <- not doing this yet, added an error if both --wls and --kurtdir flags are set