All gradients equal zero about auto-diff HOT 4 CLOSED

I am very sorry, this repository is just a toy project, it does not have a good data structure, it can only generate right results under gfortran, and uses features not standardized by the Fortran standard.

As for fazang, unfortunately it seems that the author stopped developing it about a year ago and introduced Sundials, out of curiosity I moved fazang and removed the ODE support and now it is compilable.

It's worth noting that Fortran lacks enough new open source developers and fazang is in early alpha, so if you need to use fazang it may mean you need extra work for it.

from auto-diff.

Thank you for your response and the fixed fazang fork!!
I'm new into using meson, I've tried compiling with:

mkdir build
cd build; meson setup; meson compile

Which gives me both a libfazang.so and a libfazang.a

I've tried compiling a simple main.f90 just copying the lines on the readme file. And compiling with either

gfortran -o main main.f90 -L. -I. -lfazang

gfortran -o main main.f90 libfazang.a -L. -I.

In either case I get

main.f90:2:7:

    2 |   use fazang ! load Fazang library
      |       1
Fatal Error: Cannot open module file ‘fazang.mod’ for reading at (1): No such file or directory
compilation terminated.

Sorry this isn't your library, but just in case you could give a little insight in what I'm doing wrong.

Thanks!

from auto-diff.

It looks like you didn't move the *.mod files, which are Fortran binary header files used to access interfaces to the link library at compile time.

from auto-diff.

Thanks! That was it, thought that the *.mod files were also included in the library

from auto-diff.