The final structure is completely same as the initial structure

I seem that the final structure coordinate is completely same as the initial structure by using recent API. This situation is confirmed by other many data. Is this correct…? I seem this is a bug for API.

When I used three months ago, I could get the different coordinate.

HI @nd-02110114

It looks like our switch to static calculations to get the energy is the source of this issue. We’ll remedy this shortly.

In the mean-time, can you discuss your use case for initial_structure? There might be another way to get what you need and this will help us in our development of the next API.

@shyamd

Recently, I’m building the CGCNN model with final structure(relaxed structure) data of Materials Project Dataset. What I wanted to do is to check how different the CGCNN predict when inputing initial structure and final structure.

Is there any workaround to get both initial structure and final structure…?

Hi @nd-02110114

I’ve dumped the initial_structure, structure which is the final structure, and the energy_per_atom from the old build here: https://drive.google.com/file/d/1WZF05jGtBCy2fF0LXW5Mir-u-20gOMTA/view?usp=sharing

In general, though, the best way to get data for CGNN is not to go by the materials docs. The material’s doc is intended to mix information. The idea is that a ‘material’ is agnostic of where we get that info from, but is the best quality data for that material that we have. Rather, you should grab the data from the task’s endpoint.

2 Likes

Thank you for a quick response! Currently, I got both initial structure and final structure by using task’s endpoint!