topic Please ask in the Intel in Intel® oneAPI Math Kernel Library

Intel MKL Error Parameter 4 was incorrect on entry to DGELSD.

Nothing__S — Mon, 03 Sep 2018 16:10:17 GMT

I get this error when training a neural net using using Theano and Lasagne.

In particular, I get this error when taking the square or square root of a variable e.g., theano.tensor.sqr(input)

In case it helps:

I've installed python packages: mkl 2018.0.3, mkl-service 1.1.2, mkl_fft 1.0.4, mkl_random 1.0.1

Please advise.

Please ask in the Intel

mecej4 — Mon, 03 Sep 2018 20:25:28 GMT

Please ask in the Intel Distribution for Python Forum, https://software.intel.com/en-us/forums/intel-distribution-for-python .

If you can record and report on the arguments passed to DGELSD, you can post the details and ask for help here.

Hi,

Ying_H_Intel — Tue, 04 Sep 2018 01:04:25 GMT

Hi,

How about change the input a little manually and see if any change.

Here is some information for your reference

gelsd

Computes the minimum-norm solution to a linear least squares problem using the singular value decomposition of A and a divide and conquer method.

https://software.intel.com/en-us/mkl-developer-reference-fortran-gelsd
call dgelsd(m, n, nrhs, a, lda, b, ldb, s, rcond, rank, work, lwork, iwork, info)

Return Values

This function returns a value info.

If info=0, the execution is successful.

If info = -i, the i-th parameter had an illegal value.

If info = i, then the algorithm for computing the SVD failed to converge; i indicates the number of off-diagonal elements of an intermediate bidiagonal form that did not converge to zero.

if possible, please provide the input of the function of gelsd and check the error.

or please submit the issue by Theano and Lasagne .

Best regards,
Ying