java-matrix.md

java matrix tools

whishlist

• multiplication
• addition
• matrix decomposition
• not only java based (option for BLAS)
• vectorized (SIMD)
• fast

benchmarks:

• https://github.com/optimatika/ojAlgo/wiki/Java-Matrix-Benchmark
• https://java-matrix.org
• https://www.reddit.com/r/scala/comments/5fcjo3/breeze_vs_nd4s/

available libraries

• https://github.com/scalanlp/breeze

  • included into spark
  • netlib java blas backend for labpack and netlib which optimally are compiled with MKL support

• https://nd4j.org

  • switching backend is transparent (BLAS, GPU)
  • commercially supported

• http://ejml.org/wiki/index.php?title=Main_Page

  • no BLAS, pure java
  • free
  • 202 stars
  • apache 2.0
  • recommendation: nice, but only pure java. probably easy to get started with, not the most performance. a lot of algorithms implemented

• https://github.com/optimatika/ojAlgo

  • no native code
  • 189 stars

• https://scicomp.stackexchange.com/questions/2486/parallel-scientific-computation-software-development-language

• intel MKL https://software.intel.com/en-us/articles/performance-tools-for-software-developers-how-do-i-use-intel-mkl-with-java/

  • commercial licensing

• https://ujmp.org

• https://github.com/fommil/netlib-java/

• https://github.com/fommil/matrix-toolkits-java and https://github.com/andreas-solti/matrix-toolkits-java

• https://github.com/hughperkins/jeigen

  • java wrapper around C library