Hi,

I am looking for a program that can do Principle Component Analysis with a very high dimensional input space (thousands of input variables). Since PCA requires the calculation of the covariance matrix, which has a few thousands times a few thousands entries, and not sparse. Do you know if any program exists that can do PCA with limited memory? I tried to use matlab but it needs to calculate and store the whole covariance matrix in memory, which may require up to a few hundred megabytes!

Thanks.

Savio

-- /\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\ Real world: Lam Lai Yin, Savio | E-mail: lsavio@cs.ust.hk | "Live the La Salle WWW: http://dbcl13.cs.ust.hk:8001/savio | Spirit!" | M.Phil. Student, Computer Science Department | - '89 La Salle old boy Hong Kong University of Science and Technology | \/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/

-- /\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\ Real world: Lam Lai Yin, Savio | E-mail: lsavio@cs.ust.hk | "Live the La Salle WWW: http://dbcl13.cs.ust.hk:8001/savio | Spirit!" | M.Phil. Student, Computer Science Department | - '89 La Salle old boy Hong Kong University of Science and Technology | \/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/

-- /\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\ Real world: Lam Lai Yin, Savio | E-mail: lsavio@cs.ust.hk | "Live the La Salle WWW: http://dbcl13.cs.ust.hk:8001/savio | Spirit!" | M.Phil. Student, Computer Science Department | - '89 La Salle old boy Hong Kong University of Science and Technology | \/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/