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
We are planning a new release with the follow breaking changes
New High-Level Workflow
As before, the user constructs an MCP from F, lb, ub.
Unlike before, the MCP can be converted into different transcriptions
We new support multiple solvers. A solver is generated by handing an MCP transcription to solver constructor
The solver is called as solver(θ[; initial_guess]). It's output format depends on the input format
TODO: think about this --- maybe we need a MCPVariables type that is associated with each problem format
Features to Implement
Must have
There is a new abstract problem type AbstractParametricMCP with a nominal concrete version that only takes in F(z,\theta), lb, ub and no gradients. Gradients are now considered solver-internals
Beyond the nominal problem type, we provide different transcription types + conversion scheme; e.g G-H-slack formulation.
There will be a AbstractMCPSolution type: each concrete problem struct has a concrete solution struct associated with it; we provide conversions
Maybe
The user can specify which output format they want
The text was updated successfully, but these errors were encountered:
We are planning a new release with the follow breaking changes
New High-Level Workflow
F, lb, ub
.solver(θ[; initial_guess])
. It's output format depends on the input formatFeatures to Implement
Must have
AbstractParametricMCP
with a nominal concrete version that only takes inF(z,\theta), lb, ub
and no gradients. Gradients are now considered solver-internalsAbstractMCPSolution
type: each concrete problem struct has a concrete solution struct associated with it; we provide conversionsMaybe
The text was updated successfully, but these errors were encountered: