Just to clarify, JAMA is linear algebra library for Java (not primarily designed for Procesing) that can be used within P5. I’ve read that EJML (also P5 compatible like any other Java library I guess) is faster, the absolute winner being ojAlgo according to this benchmark. (curious to know how these compare to ND4J)
Incidentally, it would be interesting to have a place (somewhere on this forum or on a specific website / github page) dedicated to the most popular/convenient non-Processing libraries that are compatible with Processing Java and Python.
For example, in Python mode I like to use Bridson for Poisson Disk Sampling or OpenSimplex for simplex noise. These libraries are fairly simple but extremely useful in a lot of cases.