Cannot execute examples/pinn_inverse/fractional_Poisson_2d_inverse.py

[AndresCenteno]

Hello! I get this error message, literally got no clue on what to do, got python 3.8.10, paddle 2.6.1

Using backend: paddle
Other supported backends: tensorflow.compat.v1, tensorflow, pytorch, jax.
paddle supports more examples now and is recommended.
Traceback (most recent call last):
File "trial.py", line 2, in
import deepxde as dde
File "/home/aner/.local/lib/python3.8/site-packages/deepxde/init.py", line 20, in
from . import backend
File "/home/aner/.local/lib/python3.8/site-packages/deepxde/backend/init.py", line 130, in
load_backend(get_preferred_backend())
File "/home/aner/.local/lib/python3.8/site-packages/deepxde/backend/init.py", line 61, in load_backend
mod = importlib.import_module(".%s" % mod_name.replace(".", "_"), name)
File "/usr/lib/python3.8/importlib/init.py", line 127, in import_module
return _bootstrap._gcd_import(name[level:], package, level)
File "/home/aner/.local/lib/python3.8/site-packages/deepxde/backend/paddle/init.py", line 1, in
from .tensor import * # pylint: disable=redefined-builtin
File "/home/aner/.local/lib/python3.8/site-packages/deepxde/backend/paddle/tensor.py", line 7, in
raise RuntimeError("DeepXDE requires PaddlePaddle==0.0.0(develop).")
RuntimeError: DeepXDE requires PaddlePaddle==0.0.0(develop).

Installed the 0.0.0 and got then

RuntimeError: (NotFound) The kernel matmul_coo_dense is not registered. [Hint: Expected iter != kernels_.end(), but received iter == kernels_.end().] (at /paddle/paddle/phi/core/kernel_factory.cc:225)

[lululxvi]

@lijialin03 @HydrogenSulfate Any idea?

[HydrogenSulfate]

Can you provide more about the os version and environment information by: python -c "import paddle; paddle.version.show()" ? I have installed the develop version as below

and I can run examples/pinn_inverse/fractional_Poisson_2d_inverse.py sucessfully by running: DDE_BACKEND=paddle python examples/pinn_inverse/fractional_Poisson_2d_inverse.py

you can check the installation script of paddle.

[HydrogenSulfate]

RuntimeError: (NotFound) The kernel matmul_coo_dense is not registered. [Hint: Expected iter != kernels_.end(), but received iter == kernels_.end().] (at /paddle/paddle/phi/core/kernel_factory.cc:225)

Another reason is that fractional_Poisson_2d_inverse.py use sparse tensor API which requires CUDA>=11.2,
RuntimeError: (NotFound) The kernel matmul_coo_dense is not registered. [Hint: Expected iter != kernels_.end(), but received iter == kernels_.end().] (at /paddle/paddle/phi/core/kernel_factory.cc:225) will be raised if you installed a paddle of cpu version.

In summary, check if CUDA>=11.2 and install a corresponding gpu version paddle.