You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
[Metadata] Shared initial guess in the metadata, read by every modeling #377
Agents: read the guide for agents in #310 first. Specification: this description, #306 and its comments.
Sub-issue of #306, part 3 of 5. Depends on #368 and #369 (they rewrite the JuMP models and the number of controls per scheme).
Scope
As in #306, section :init: format (state=…, control=…, variable=…), each component a constant, a vector or a function of t, never data on a grid; OptimalControl modelings pass it as is; JuMP evaluates it at the nodes with the controls of the scheme; overridable (init=). No initial guess left in the model files. Maintainer's rule: the guess must make the solvers converge without being too close to the solution. Documentation of the initial guess (#151).
Done when
Initial guesses read from the metadata by every modeling; none left in the model files.
comparison/test_init checks the guesses coincide at the nodes.
Activity
ocots commentedon Nov 6, 2026
Taken by an agent; a plan is proposed to the maintainer before any code.
feat: the initial guess of every problem in its metadata
feat: the OptimalControl models read the shared initial guess, init= …
feat: JuMP start values from the shared initial guess, at the nodes o…
docs: the shared initial guess and how to replace it (#151)
refactor: drop the start value of Component, the initial guess is the…
Merge pull request #395 from control-toolbox/feat/377-shared-init
test(mtk): build the Corleone problem once, from the initial guess, a…
test(mtk): the time span from the parameters, the control start check…