 Posted Wednesday, November 11, 2009 7:05 AM
|
|
|
|
| Hi I am trying to claculate something (as below) Theta = YYt.PseudoInverse() * B; I have to admit it has been a while since I did matrices but I am getting some 'strange' results. If Theta, YYt and B are all from the Matrix class the result is what I am expecting , but If they are all CMatrix (imaginary part of YYt and B set to 0) , I get a completely different result. Am I doing something wrong here ?
|
|
|
Posted Friday, November 20, 2009 4:14 AM
|
|
|
|
| Hello, I have checked the computations of complex pseudoinverse and I confirm that there is bug which was indroduced in version 4.3 You may still compute the correct pseudoinverse for a compex matrix A as follows: CSVD csvd = CSVD(A);
CMatrix Api = csvd.V * csvd.S.Inverse() * csvd.U.Adjoint(); This issue will be fixed in the next version of NML. Thank you for informing us about.
Trifon Triantafillidis | Lead Developer | 
|
|
|
|
|