Hi Jake, I took a 10X10 random matrix, A and ran Lanczos to produce the Eigen Vectors. You are correct that the Eigen values are closer to each other now. However, there is still a problem with the Eigen Vectors corresponding to Eigen Values as in Eigen Vectors are being associated by Mahout to wrong Eigen Values. Here is the output. http://evijayakumar.pastebin.mozilla.org/1249324
Thanks Esh
