Saturday, November 12, 2011

la4j-0.2.0 has come!

You can download it form official page: http://code.google.com/p/la4j/
Please, feel free in touch me if you are using la4j in yours projects. I would like to feel output from my work.

Here is full changelog:
  • ~55 classes, ~6700 loc, ~90 tests, ~50 kb (bin);
  • new package la4j.io for reading/writing matrices in MatrixMarket format;
  • matrices decomposition (LU, QR, Cholesky, SVD, Eigenvalues) support;
  • matrix inversion support;
  • sparse (CSR) and dense matrices and vectors support;
  • matrices and vectors can be serialized;
  • la4j uses Maven and jUnit now;


3 comments:

  1. Каким алгоритмом SVD считаете?

    ReplyDelete
  2. Самым очевидным, который описан в http://en.wikipedia.org/wiki/Singular_value_decomposition разделе SVD Calculation. Такой-же метод использует JAMA и LAPAC.

    ReplyDelete
  3. LAPACK? То есть статью Деммеля и Кахана вы тоже прочитали и улучшения оттуда реализовали)?

    ReplyDelete