Skip to content

[Backlog Item]: Build a differentiable wofost72_pp #39

@SarahAlidoost

Description

@SarahAlidoost

In #3, we created a prototype, where two modules leaf_dynamics and root_dynamics are differentiable wrt only TDWI and SPAN parameters.
To develop a differentiable module, we check for look-up tables, hard thresholds, and mathematical operations, and replace them with differentiable alternatives. In addition to differentiability, the implementation also focuses on efficiency, by leveraging vectorized operations.

We want to apply the same changes to all parameters of 9 modules as well as wofost72 module.

Requirements:

  • tests should be there to check correctness of the model using all test data, see [Task]: use all test data for unit tests #27
  • tests should be there to check gradients
  • operations should be vectorized for both parameters and weather data (drv variable)
  • documentation and docstrings should be fixed.

Examples:
See current implementations of leaf_dynamics and root_dynamics .

Note:
We use EngineTestHelper from utils to run individual model, until engine is fixed in #25

Metadata

Metadata

Assignees

No one assigned

    Labels

    Backlog itemA collection of tasksEpic 1Optimizing parameters in Wofost72_PP using pytorch

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions