utils
templates.pde.utils
Resolve field_dynamics into the assembly plan the FEM template emits.
The PDE backend used to read only operators[].coefficient, sum them into one scalar and assemble M + dt*D*K — so the equation.rhs an author wrote was decorative, and a reader could not tell which of the two descriptions ran. Everything here exists to make the declared equation be the executed operator.
A field system is lowered to the block form
M du/dt = A u + M f(u, t)with one block row per state variable. A collects every term that is LINEAR in the state variables (mass and stiffness blocks, assembled once); f collects whatever remains, evaluated at the current state each step. That split is what lets an implicit scheme run: the linear part is solved implicitly and is unconditionally stable, while a nonlinear reaction term stays explicit (IMEX). Callers get told which terms went where.
Recognised operators inside an RHS: laplacian(v), div(c*grad(v)) and bare state variables. A coefficient may be a scalar parameter or a per-vertex field, which is what makes a spatially varying propagation scale expressible.
Classes
| Name | Description |
|---|---|
| FieldPlanError | A field system that cannot be lowered to the block form above. |
FieldPlanError
templates.pde.utils.FieldPlanError()A field system that cannot be lowered to the block form above.
Functions
| Name | Description |
|---|---|
| field_assembly_plan | Everything the FEM template needs, resolved from the experiment datamodel. |
field_assembly_plan
templates.pde.utils.field_assembly_plan(experiment)Everything the FEM template needs, resolved from the experiment datamodel.
Raises
| Name | Type | Description |
|---|---|---|
| FieldPlanError | when the declared system cannot be lowered — an explicit failure in place of the old behaviour, which silently ignored the equation entirely. |