Where to download the dataset used in the EQNN force field tutorial?

Demos
1

Hi PennyLane team,

I’m currently working through the “Symmetry-invariant quantum machine learning force fields” tutorial (link).

However, the code loads .npy files like this:

energy = np.load("eqnn_force_field_data/Energy.npy")
forces = np.load("eqnn_force_field_data/Forces.npy")
positions = np.load("eqnn_force_field_data/Positions.npy")

But I couldn’t find any download link or GitHub repository that contains these files. Could you kindly point me to the dataset used in this demo, or provide any guidance on where I can obtain or regenerate these files?

Thanks a lot!

2

Hi @Ciel_Phantomhive , welcome to the Forum!

You can find the dataset here.

Let us know if you find everything you need there!

3

Dear PennyLane Team,

Thank you for your previous response and for providing the dataset link. The files available in the dataset are in formats like .pdf, .rst, and .data.*, such as input.data.RPBE, input.data.BLYP, etc., but not the required .npy format. I understand these may be the raw training files, but the tutorial assumes preprocessed .npy arrays without showing how to extract or convert them.

Would it be possible for you to provide the exact .npy files used in the tutorial, or share the preprocessing script or notebook that converts the provided raw data files into Energy.npy, Forces.npy, and Positions.npy?

This would be extremely helpful in reproducing the results and continuing to learn from the demo.

Thank you very much for your time and support!

Best regards,
Lu