Scoping and Optimization Codes
Optimization codes search the design or operating space for points that maximize an objective subject to physics and engineering constraints.
Searching the design space
A fusion device has many free parameters and many constraints. Scoping and optimization codes automate the search for good designs by wrapping a physics or systems model in an optimization loop: propose a set of parameters, evaluate the objective and constraints, and iterate toward an optimum.
The objective might be to maximize confinement or fusion power density in simulation, or to find the smallest machine meeting a performance target, while constraints enforce stability limits, magnet stress bounds, and power balance.
Algorithms
Gradient-based methods converge fast when the model is smooth and derivatives are available. Derivative-free and population-based methods, genetic algorithms, particle swarm, Bayesian optimization, are used when the objective is noisy, discontinuous, or expensive. Bayesian optimization with a surrogate is popular when each evaluation is costly.
Multi-objective trade-offs
Real design pits objectives against each other. Multi-objective optimization returns a Pareto front, the set of designs where improving one objective requires sacrificing another, letting designers see the trade-offs explicitly rather than collapsing them into a single weighted score prematurely.
Design relevance
Optimization over stability, transport, and geometry helps locate operating points like the negative-triangularity, -0.30, configuration chosen for the Hyperion breeder, balancing confinement against exhaust and stability. Because the machine is not built, these are simulation-derived design choices, refined as higher-fidelity models mature.
- Wraps a model in an optimization loop
- Gradient-based or derivative-free algorithms
- Bayesian optimization for costly evaluations
- Pareto fronts expose multi-objective trade-offs