Electronic dielectric tensors: coordinate frame and matching structure

Materials Project
1

Hello Materials Project team,

Could you clarify the coordinate convention and recommended structure pairing for electronic dielectric tensors returned by the current materials.dielectric API?

For structure-to-tensor applications, I would like to retrieve a structure and its electronic dielectric tensor in exactly the same Cartesian frame. The dielectric methodology documentation describes the electronic and ionic contributions, but I would appreciate guidance on the following:

  1. Are the electronic dielectric tensor components expressed in the Cartesian frame of the originating DFPT calculation, an IEEE-standardized frame, or another convention?
  2. Which structure is guaranteed to match that frame? Can a structure from materials.summary be directly paired with a tensor from materials.dielectric, or should the exact originating DFPT task structure be retrieved?
  3. If the structures differ by standardization or rotation, is a transformation matrix available in the API? What is the recommended procedure for obtaining an aligned pair while retaining the correct task provenance?

For a coordinate change using column vectors, r_new = R @ r_old, I expect the tensor to transform as epsilon_new = R @ epsilon_old @ R.T. A short example showing how this relates to the API structures would be very helpful.

I have seen the historical origins/task-mapping report materialsproject/api#960. I understand that its fix concerned provenance mapping rather than the underlying property values, and I am not claiming that the previous bug persists.

This is a coordinate-convention and data-access clarification request, not a confirmed bug report. Could you point me to the authoritative way to obtain a frame-consistent structure and electronic dielectric tensor?

Thank you for your help.