You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have an idea of performance enhancement when we are using some matrix parameters (i.e. rank, eigen decomposition, SVD) simultaneously. The idea is to store some parameters and change while we are using some algorithms. I.e. if we perform SVD decomposition and then call rank() method, we don't need to perform rank computation directly if we compute inner parameter rank while doing decomposition (we know that count of nonzero singular values IS a rank, so we can use direct search of nonzero singular values instead of O(n^3) algorithm for rank finding).
The text was updated successfully, but these errors were encountered:
Right. Caching the intermediate result (aka memoization) is the most common practice of performace improvements. Since, the la4j library is a single-threaded solution we could do that w/o worrying on thread-safity. Let's discuss this technique in 0.6.0 release.
I have an idea of performance enhancement when we are using some matrix parameters (i.e. rank, eigen decomposition, SVD) simultaneously. The idea is to store some parameters and change while we are using some algorithms. I.e. if we perform SVD decomposition and then call
rank()
method, we don't need to perform rank computation directly if we compute inner parameterrank
while doing decomposition (we know that count of nonzero singular values IS a rank, so we can use direct search of nonzero singular values instead ofO(n^3)
algorithm for rank finding).The text was updated successfully, but these errors were encountered: