Computing Library › Fusion Codes
Fusion Codes

Coupling Strategies and Operator Splitting

Operator splitting advances coupled physics one process at a time, a simple scheme whose accuracy and stability depend on how the pieces are ordered and iterated.

Splitting the coupled problem

When several physical processes act together, solving them all in one monolithic system is often impractical. Operator splitting advances each process separately in sequence, using the output of one as the input to the next. This lets each subproblem be solved by a specialized code, at the cost of introducing a splitting error from not solving them simultaneously.

Splitting is the backbone of most integrated modeling: equilibrium, then transport, then heating, then back, each solved by its own module and coupled in a loop.

Kronos motion — fusion

First-order and symmetric splitting

The simplest scheme applies the operators in fixed order once per step, giving first-order accuracy in the timestep. Symmetric (Strang) splitting applies them in a palindromic order, half a step of one, a full step of another, half a step of the first, recovering second-order accuracy for the same components. The ordering is not cosmetic; it changes the error.

Fixed-point iteration and stability

For steady-state coupling the loop is iterated to a fixed point. Strong nonlinear feedback between modules can make the iteration oscillate or diverge, so under-relaxation, blending the new and old states, is applied to stabilize convergence. Choosing the relaxation factor is a practical art in coupled fusion workflows.

Design relevance

The self-consistent operating points computed for the Hyperion breeder depend on a stable, converged splitting loop; a poorly chosen scheme can produce a spurious answer or fail to converge. Sound splitting is quietly foundational to trustworthy simulation, and its convergence is documented as part of the honest design record.