manta executable file for centos 7 x86 64

[joseph pareti]

is this available for download? I have significant troubles building it on my VM, but for my project it is just an intermediate element, so it would help to just use what is available.

Thanks.

[Nils Thuerey]

Hi Joseph, unfortunately, we can't provide pre-compile executables at the moment. Compilation with linux should be pretty straight forward though. What errors are you getting?

Cheers, Nils

[joseph pareti]

Hi Nils and everybody,

my latest attempt to build manta on centos 7 terminates as follows:

./manta ../scenes/_trainingData.py --dim 3 --addModelGeometry True --addSphereGeometry True

Version: mantaflow 64bit fp1 commit dd3bb0c0a65cc531d3c33487bde5edcb4aa6784f from Mar 29 2018, 15:21:34

Loading script '../scenes/_trainingData.py'

Traceback (most recent call last):

  File "../scenes/_trainingData.py", line 13, in <module>

    from voxel_utils import VoxelUtils

  File "/home/joepareti54/FluidNet/manta/scenes/voxel_utils.py", line 1, in <module>

    import numpy as np

ModuleNotFoundError: No module named 'numpy'

Script finished.

More details in the attached document.

Regards,

--
You received this message because you are subscribed to the Google Groups "mantaflow Forum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mantaflow-forum+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mantaflow-forum/1861b3d0-228b-4fc1-aae3-195e6aefc3ad%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Nils Thuerey]

Right, it looks like mantaflow links against 2.7, and not 3.6 which has numpy. You could try forcing the version with:

find_package(PythonLibs 3.6 EXACT REQUIRED)

for the second (!) find_package(PythonLibs...) call. (We're planning to update the cmake script there anyway, that's a common problem.)

Otherwise, you could manually set PYTHON_INCLUDE_DIRS and PYTHON_LIBRARIES to the right anaconda directory.

[Nils Thuerey]

I just pushed a few updates in the master branch that should address this problem. If I saw it correctly, you're using the mantaflow version from the "Accelerating Eulerian Fluid Simulation With Convolutional Networks" paper, though - that might require a few changes.