diff options
Diffstat (limited to 'unsupported/Eigen/src/SVD/doneInBDCSVD.txt')
| -rw-r--r-- | unsupported/Eigen/src/SVD/doneInBDCSVD.txt | 21 |
1 files changed, 21 insertions, 0 deletions
diff --git a/unsupported/Eigen/src/SVD/doneInBDCSVD.txt b/unsupported/Eigen/src/SVD/doneInBDCSVD.txt new file mode 100644 index 000000000..8563ddab8 --- /dev/null +++ b/unsupported/Eigen/src/SVD/doneInBDCSVD.txt @@ -0,0 +1,21 @@ +This unsupported package is about a divide and conquer algorithm to compute SVD. + +The implementation follows as closely as possible the following reference paper : +http://www.cs.yale.edu/publications/techreports/tr933.pdf + +The code documentation uses the same names for variables as the reference paper. The code, deflation included, is +working but there are a few things that could be optimised as explained in the TODOBdsvd. + +In the code comments were put at the line where would be the third step of the algorithm so one could simply add the call +of a function doing the last part of the algorithm and that would not require any knowledge of the part we implemented. + +In the TODOBdcsvd we explain what is the main difficulty of the last part and suggest a reference paper to help solve it. + +The implemented has trouble with fixed size matrices. + +In the actual implementation, it returns matrices of zero when ask to do a svd on an int matrix. + + +Paper for the third part: +http://www.stat.uchicago.edu/~lekheng/courses/302/classics/greengard-rokhlin.pdf + |