Baunsgaard commented on pull request #1153: URL: https://github.com/apache/systemds/pull/1153#issuecomment-760750295
> One thing where I was quite unsure was the unit tests. Since calculating determinants and the inverse of the covariance matrices can lead to floating point errors, I was not quite sure how to compare the results. I did compare most of them, as suggested in the mailing list, with the avg. bit distance, with a quite high maxUnitsOfLeastPrecssion. Maybe since the correction value is overwritten in DML and therefore not the same parameter used for the algorithm, but still as long as the difference is this small i consider it fine. > Although the values from the inverse covariance matrices can differ a lot (systemDS vs R), i am pretty sure that the computation is correct, since multiplying it with the covariance matrix itself, leads to the identity (which I tested during development). please do add such a test back, since it reflects "a function of the algorithm", not "similarity to R" since in the end what we want is the first, not necessarily the second. ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected]
