Meeting agenda, notes and actions for 2025-03-26 at 2pm ET

Organizer
: Daniel Wheeler

Attendees
: - Daniel Wheeler
- Add name

## Links

- [Repository][repository]

## Agenda

1. Any questions or items to raise for discussion (please add below)
2. [Actions from last week](https://github.com/usnistgov/am-dt-modeling/blob/main/resources/meetings/DT%20AM%20Meet%202025-03-19.md#action-items)
4. Issues with Adamantine and using TorchDA
- Why do I want to do this?
- Control the parameter updates
- Have a wide range of parameter updates
- e.g. have a temperature dependent material parameter as update
- control the H function (state -> observable)
- no provision for this currently in Adamantine
- Data assimilation should be programmed in high level language due to the many customizations that are required
- Hoping to use the restart mechanism but restart files are just serialized dumps of state.
- Not reasonable to read/write C++ update serialized dump from within Python
- For now test Adamantine / TorchDA with single runs without continous updates
- This is adequate for feasiblity study
- Other options for TorchDA / Adamantine coupling
- Update Adamantine to allow initial imposed fields
- lose current mesh refinement
- many drawbacks with this approach
- Update Adamantine to restart from VTK files
- Update Admantine to read updated fields from files in the same place where current DA is used
- Call Python from Adamantine C++ and use in place of current DA
- Use Adamantine as a C++ library from Python
- Just use current DA capability
- Just use Adamantine for paramteer estimation
- https://github.com/adamantine-sim/adamantine/issues/379
5. Dilip updates?
6. Shengyen updates?

## Notes

Notes

## Action Items

- [ ] action item

<!-- links -->

[repository]: https://github.com/usnistgov/am-dt-modeling