Set up design variables

A design variable is a value a study holds, searches or computes. Each one has one of three roles:

Name a variable

A variable is addressed <asset>.<entry>: the asset id from the project space, a dot, and the name of a varset entry in that scene. hull.span is the span entry of the scene added as hull. A variable that belongs to the study rather than to any scene uses the reserved asset id study, as in study.cruiseSpeed. An unqualified name is refused. A guess at which scene you meant would drive the wrong one and read like a modeling mistake.

Set one

{"task": "study.variable.set", "args": {"study": "s1", "variable": "hull.span",
  "role": "free", "min": 1.8, "max": 3.6, "steps": 18}}
{"task": "study.variable.set", "args": {"study": "s1", "variable": "study.liftToDrag",
  "role": "pinned", "value": 14}}

• Steps are intervals. A free variable swept from 1.8 to 3.6 in 18 steps is evaluated at 19 points, because both ends are sampled.
• Linked variables cannot be set through study.variable.set yet; the task refuses the role. When they arrive, a linked expression reads other variables by address, and a variable linked to itself, or a circle of linked variables, is refused: unlike a coupled pair of components, a circular definition has nothing to converge to.
• Changing a pinned value keeps the study's results table. Adding a variable, or changing a role, starts a new one, because the study is now asking a different question. study.variable.set returns revisionChanged so you can tell which happened.

Where a variable is read

A variable does nothing until the model graph reads it: bind it to a component input. The study refuses a variable that nothing reads, and a graph that reads a variable the study does not declare.

The Variables section of the Study activity lists every variable with its role, and shows bounds only for the free ones.