Inputs, sweeps and cases
Inputs expose the parameters that a study or reusable model permits callers to change. Sweeps bind those inputs into named cases; they do not rewrite source declarations or mutate a model.
Declare an input
study bracket_response {
input service_load : force {
default = 1 kN
minimum = 0 N
maximum = 5 kN
}
geometry = geometry.bracket
mesh = mesh.bracket
model = model.loaded_bracket(service_load = input.service_load)
analysis static { structural static {} }
}default, minimum, and maximum are optional. Without a default, a consuming root or case must
bind the input. Bounds have the same physical type as the input and are enforced before geometry or
solver work begins.
Inputs may belong to geometry, model, or study declarations. A study input can bind a geometry parameter, a model parameter, or both. A geometry change correctly creates a different geometry and mesh identity; it is not treated as a load-only solve.
Product sweeps
sweep envelope over study.bracket_response {
axis service_load {
half = 0.5 kN
nominal = 1.0 kN
overload = 1.5 kN
}
axis damping {
low = 0.01
nominal = 0.02
}
combine = product
}product creates the Cartesian product of all axis values. Case identities are derived from axis
and value labels in a stable canonical order, not from task completion order.
Zipped sweeps
With combine = zip, every axis must use the same labels. Values with matching labels form one
case:
sweep paired over study.bracket_response {
axis service_load {
low = 0.5 kN
high = 1.5 kN
}
axis damping {
low = 0.01
high = 0.04
}
combine = zip
}Explicit cases
Use case blocks when the parameter combinations are irregular or deserve domain names:
sweep qualification over study.bracket_response {
case limit_load {
service_load = 1.25 kN
damping = 0.02
}
case proof_load {
service_load = 1.5 kN
damping = 0.03
}
}A sweep uses axes or explicit cases, never both. Every required study input must be bound after defaults are applied. Unknown inputs, duplicate case names, out-of-range values, unit mismatches, and unequal zip labels fail checking.
Sweep outputs
Sweep outputs operate on each case's declared study outputs:
output peaks = collect(case.output.peak_displacement.value)
output governing = max(case.output.peak_displacement.value, over = cases)
output fields = retain_each(case.output.displacement)collect retains compact values with case coordinates. min and max retain the governing value
and its case attribution. retain_each requests each large output object; use it deliberately
because field retention can dominate storage.
Parallel execution
Independent sweep cases may execute concurrently locally or as cluster fan-out. That is different from partitioning one numerical operator across workers. The sweep definition remains the same in both placements, and results are restored to canonical case order before publication.