Re: least-squares problem: B ~ X.A

*To*: mathgroup at smc.vnet.net*Subject*: [mg57381] Re: [mg57329] least-squares problem: B ~ X.A*From*: Sseziwa Mukasa <mukasa at jeol.com>*Date*: Thu, 26 May 2005 04:31:24 -0400 (EDT)*References*: <200505251002.GAA03952@smc.vnet.net>*Sender*: owner-wri-mathgroup at wolfram.com

On May 25, 2005, at 6:02 AM, Pascal wrote: > Hi, > > I would like to find the solution of the least-squares problem: B ~ > X.A, > that is, given matrices B and A, find matrix X which minimizes the > difference between B and X.A > (typical dimensions, B: 500x60, X: 500x5, A: 5x60) > Can Mathematica do that? > I believe the pseudoinverse satisfies your requirement, in particular X = B.PseudoInverse[A]. Regards, Ssezi

**References**:**least-squares problem: B ~ X.A***From:*"Pascal" <Pascal.Plaza@ens.fr>