Hi Frank, I'm sorry for problems we have provoked you. It was my fault. The problem is of one of the students that we have in our lab. I simple resend his message because I have been in the gsl-discuss list from long time ago. I didn't know the existence of the gsl-help list. Anyway, here is a new version that compiles and the same example in matlab code. The cpp file could be compiled in a standard linux box by: g++ codi_gsl.cpp -lgsl -lcblas -o codi_gsl or with gcc gcc codi_gsl.cpp -lgsl -lcblas -lstdc++ -o codi_gsl the file produces the result of the matrix: 0.0155688 8.90712e+17 -4.45356e+17 0.000937403 -8.50932e+17 4.25466e+17 -0.00800181 1.5173e+18 -7.58652e+17 0.0028235 -4.49257e+17 2.24628e+17 -0.000897874 -4.2349e+17 2.11745e+17 -0.000242821 1.73714e+17 -8.68568e+16 -0.000316232 -6.16453e+15 3.08227e+15 0 0 0 a similar code in matlab produces: 0.0156 0.0012 0.0024 0.0009 0.0070 0.0140 -0.0080 0.0119 0.0239 0.0028 0.0139 0.0277 -0.0009 0.0180 0.0360 -0.0002 0.0036 0.0072 -0.0003 0.0000 0.0001 0 0 0 the studend mail is attached below. Also, I would like to note that I have had problems with my email (lepalom@wol.es) so I didn't notice your mail since yesterday. Best regards, and thanks. Leo ----------------------------------------------------------------------------- Hi, good morning I have been lately working with the gsl library, using the SVD decomposition, but I am having real trouble, so i thought that i was probably doing something wrong and you could help me. I need to calculate the pseudoinverse of a mxn matrix with n>m (more columns than rows), and with that matrix being rank-deficient. I have created the code to compute the pseudoinverse of A. As the gsl library isn't able to make the SVD decomposition of n A = VDU^T pinv(A) = US^(-1)V^T The thing is that I have computed the pseudoinverse both using gsl and matlab, and with matlab I get more accurate results than with gsl. And my question is if that difference in the results is because I am coding something wrong or itis something else. The result i get from matlab for the pseudoinverse of the same matrix as the one in the code is: 0.0156 0.0012 0.0024 0.0009 0.0070 0.0140 -0.0080 0.0119 0.0239 0.0028 0.0139 0.0277 -0.0009 0.0180 0.0360 -0.0002 0.0036 0.0072 -0.0003 0.0000 0.0001 0 0 0 I'd really apreciate your help. Thanks.