I am doing multiple operations of type A.dot(B), where A and B are dense matrices of the size of the order 10e6. What methods can I use to not run into memory issues while computing such matrices?
That should be transparent to you as the user, since that's a memory management task performed by the operating system, unless you've not allocated sufficient virtual memory. The only real issue is that going to SSD or HD is substantially slower than RAM. Obviously, you should populate your computer with as much RAM as it can handle for problems like this
TTFN (ta ta for now)
I can do absolutely anything. I'm an expert!
I did see that the numpy linalg package has special dot product function, looks like its made to improve calculations that include a sparse vector vs a non-sparse vector though.