dll error while performing interpolation using pykrige

[dosel]

Hello,
I downloaded pykrige to interpolate the UCS point data to volume.

I followed this workflow

Interpolation/Kriging

An external Kriging python library can be applied using the Programmable Filter. For example, if you want to use PyKrige (PyKrige — PyKrige 1.6.1 documentation), first, prepare the same version of the Python environment that ParaView uses, and install Pykrige on it. Then, copy: Anaconda3/envs/pyKrige/Lib/site-packages/pykrige Anaconda3/envs/pyKrige/Lib/site-packages/scipy (This is an Anaconda example) under (ParaView install folder)/bin/Lib/site-packages. After this preparation, star…

But when I used some codes to use this module, this error occured.

But the dll file was in the path.

I searched why this error occurs, and I finaly found that it was because of the missing dependency files for that dll file.

The missing files are not in my computer. I re-download several times but all was the same.

The missing files were more than 600, so I couldn’t download them one by one.

Do you have some ideas about this problem?

[Kenichiro-Yoshimi (Kyoshimi)]

The version of numpy in ParaView 5.11.1 is 1.21.1, which is older than the one used by PyKrige, so you will get a Warning, but if you use the Anaconda environment on Windows, you can achieve this as follows.

1. Creating the pykrige environment:

conda create -n pykrige python=3.9
conda activate pykrige

2. Install PyKrige

pip install pykrige

3. Backup the numpy folder in ParaView
   In C:\Program Files\ParaView 5.11.1\bin\Lib\site-packages, rename the numpy folder name to numpy_bak.
4. Copy the folders
   Copy the folders under C:\Users\aaa\anaconda3\envs\pykrige\Lib\site-packages:
   ・numpy
   ・pykrige
   ・scipy
   ・scipy.libs
   to C:\Program Files\ParaView 5.11.1\bin\Lib\site-packages

[dosel]

Thank you for your help.

I tried, but I got this error message.

[dosel]

(I used anaconda3 powershell prompt and created environment from the start.
But I didn’t rename the numpy folder and copy the scipy.libs folder.
So I did those 2 in this try.)

[Kenichiro-Yoshimi (Kyoshimi)]

Your error message states that it is an MKL error with respect to numpy.
Are you installing using the pip command? If you are installing with pip, NumPy should be installed as built with OpenBLAS. On the other hand, if you install with the conda command from conda-forge, it will be built with Intel MKL, which seems to be difficult to integrate into ParaView.

[dosel]

Thank you for your help.

I installed the OpenBLAS and re-installed numpy and pykrige.

And I got the same DLL error message.

[Kenichiro-Yoshimi (Kyoshimi)]

Was pykrige installed with the conda command? This also seems to be hard to integrate into ParaView. Is it possible to install pykrige (and scipy) with the pip command?

[dosel]

Thank you for your help.

But I installed pykrige and scipy with pip command.

I re-installed scipy just in case. Then the scipy folder disappeared in the C:\Users\aaa\anaconda3\envs\pykrige\Lib\site-packages.

so I copied just 3 folders and there was the same dll error.

[Kenichiro-Yoshimi (Kyoshimi)]

I am not sure about the disappearance of the scipy folder after reinstalling scipy. Have you activated the virtual environment with the

activate pykrige

command and then installed pykrige or scipy?

[dosel]

Yes, I used the environment. (I named the environment “paraview”)

[dosel]

[dosel]

Hello, thank you for your help.
I resolved this problem by creating new environment.
But there is another problem.

I got this error message after running your code

(I used “RandomAttributes1” as the data for interpolation and “filtered_coordinates.vtk” as the space data to assign interpolation results.)

Thank you for reading!

[dosel]

I think this error means that the imported library dosen’t have the function which is in the code.
But I know your code worked well the other cases.
So I have no idea what was wrong.

[Kenichiro-Yoshimi (Kyoshimi)]

If you could share the data and state file(*.pvsm or *.py), I may be able to confirm the phenomenon.

[dosel]

https://drive.google.com/drive/folders/1pHwYalv8YU8RpawvQqLbiQwWJ__3Rtos?usp=sharing

You can download the data and state file.

Thank you.

[Kenichiro-Yoshimi (Kyoshimi)]

Adding a header line to borehole_data.csv seems to work. Also, the Kriging method requires a very large amount of memory, as can be seen in the semi-variogram definition. Hence, it is difficult to treat many observation points as input.

Attached below is the State file when Kriging is applied when the number of observation points and the points to be predicted are reduced.
pykrige_small.7z (23.8 KB)

Of course, to use Kriging properly, an empirical variogram must be created from the observation point data and fitted to the appropriate variogram model to determine the sill, range, and nugget.

[dosel]

Thank you for your help.

I understood the process.
When I load your state, the figure was same as the image you uploaded.

But if I do this work in my computer, the figure dosen’t created.
And I got this error message. (This error occured when I loaded your state, too.)

I thought this error occured because of the version of numpy.
So I updated numpy at anaconda powershell prompt and copied the numpy folder.
But this error occured again.

Do you have an idea about this error?

Again, I really appreciate for your help.

[Kenichiro-Yoshimi (Kyoshimi)]

The numpy messages are warnings and can be ignored. I think what is not working with your data is another cause, such as memory overflow.

[dosel]

Okay, thank you for your help!
I’ll find another solution.