I've tried to install this per directions on a 'fresh' ubuntu 16.04 load and it too ends in testing with a similar openblas error:
=== compiling la/oblas =============================================================

github.com/cpmech/gosl/la/oblas

./oblas.go:22:28: fatal error: openblas_cblas.h: No such file or directory
compilation terminated.

from gosl.

Hi, thanks for trying Gosl!!!

I've just installed Ubuntu 16.04.4-desktop on a virtual machine (VirtualBox) and proceeded with the installation instructions from here: https://github.com/cpmech/gosl/blob/master/doc/InstallationOnUbuntu.md

I didn't get any error though! (see figure)

Can you please provide more information about your system?

Cheers.
Dorival

from gosl.

Hello, I've just tried it again in my fresh Ubuntu installation and all seems fine. I'm happy to help regardless. Cheers. Dorival

from gosl.

Thanks for checking. Previously I think there was a conflict with OpenBLAS being install previously.
SO, I reformatted my disks and installed a fresh copy of Ubuntu 16.04 and did all the updates.
Then I installed go1.10.1. So far so good.
Then I followed your steps, but this is what I got when trying .all.bash:
--- FAIL: TestMatInv02 (0.00s)
t_matrix_ops_test.go:30: matrix "b⋅bi" is not diagonal. diff=1.3587010698815905e-10

from gosl.

Excellent!

This error is not an issue. I usually set the tolerances to a very small number and sometimes different CPUs produce slightly different results.

I've just increased the tolerance to 1e-9.

You may like to git pull and try again.

Cheers.
Dorival

from gosl.

Perfect:
=== compiling tsr =============================================================
. . . testing . . . Tensor2set01
. . . testing . . . Tensor4set01
PASS
ok github.com/cpmech/gosl/tsr 0.008s

=== SUCCESS! ============================================================

Best regards,
Dave

from gosl.

Thanks!

from gosl.