Documentation menu

Draft. This manual is new and still growing. If something here does not match what you see in the software, the software is right — tell us and we will fix the page.

Study

Where it is

Setup tree: Study

Controls

WHAT TO VARY

＋ Factor

Control: action (ToolButton2)

DESIGN

Design

Control: choice (ComboRow)

Samples

Control: value (ValueField)
Value modes: none (plain entry)
Shown when: root.lhs

The in-app guidance depends on the case:

The size of each base sample, NOT the number of runs — a Sobol design asks for this many times the number of factors plus two. The indices are differences of means, so below a few hundred they carry real noise.
How many runs. A Latin hypercube covers the space in whatever number you choose, which is why it does not explode.

Seed

Control: value (ValueField)
Value modes: none (plain entry)
Shown when: root.lhs

Saved with the case. The same seed gives the same points on any machine — without that a study's results cannot be checked.

THE RUNS

Generate the runs

Control: action (ToolButton2)
Enabled when: StudyState.solvable && StudyState.pointCount > 0 && root.allFactorsResolve

OPTIMISE

Goal

Control: choice (ComboRow)
Options: Minimise the result, Maximise the result

Budget

Control: value (ValueField)
Value modes: none (plain entry)

The most runs the search may spend. Each one is a mesh and a solve, so this is a cost ceiling, not a quality knob.

Searching…

Label changes with the case; the source gives these forms: Searching…, Start the search.

Control: action (ToolButton2)
Enabled when: !CaseState.optimisationRunning && root.allFactorsResolve && StudyState.factors.length > 0

Stop

Control: action (ToolButton2)
Shown when: CaseState.optimisationRunning

THE SPREAD

Judge against a limit

Control: on/off (ToggleRow)
Shown when: root.spreadResult !== null

The reliability question: what fraction of the plausible inputs put the answer past a threshold - a yield stress, a temperature rating, a drag target.