Defining Adjoint Stopping Criteria

Aside from the general stopping criteria for the primal and adjoint solvers, you can create individual stopping criteria for each cost function. With this option, you can run various cost functions with different conditions for stopping the adjoint solver. When either global or cost function specific criteria are met, the adjoint solver moves to the solution of the next cost function, or stops completely if there are no more cost functions to solve.

The global adjoint solver stopping criteria apply to all the cost functions. Therefore, you must ensure that any global adjoint solver stopping criteria that you define are adequate for all the active cost functions. In particular, the global criteria should not depend on values associated to a specific cost function (for example the residual of a specific cost function), because they could potentially stop the solution of the other defined cost functions. The most commonly used stopping criterion for the global adjoint solver is to set a fixed number of steps (Fixed Steps criterion).

By default, there are no global stopping criteria for the adjoint solver. Instead, there are two automatically-generated stopping criteria upon creation of a cost function. These cost function specific stopping criteria are Fixed Steps, and Residual.

To define adjoint stopping criteria:
  1. To create global stopping criteria for the adjoint solver, right-click the Solvers > Adjoint > Solvers > Adjoint Steady > Stopping Criteria node and select New Criterion or New Monitor Criterion. Set the properties appropriate for your simulation.
  2. To create stopping criteria specific to a cost function, right-click the Adjoint > Adjoint Cost Functions > [cost function] > Stopping Criteria node and select New Criterion or New Monitor Criterion.