Consider using concurrency in MatrixProfile implementation
barrybecker4 opened this issue · 3 comments
barrybecker4 commented
Not sure how well this will work, but I would like to experiment with using concurrency to speed execution of the for loops in MPdistance.
ensozos commented
That's a good idea. You may find useful STOMP and SCRIMP algorithms (presentation and article) .
We can also speed up the execution of MPdistance with MASS v3 algorithm (current version in our code is mass v2).
barrybecker4 commented
It was actually the loop in MatrixProfile class and not MBdistance that I modified. The loop over all the series steps can now be done sequentially or with concurrent threads.
barrybecker4 commented
Closing this, since may change to add concurrency was accepted into master.
Thank you.