Control-toolbox interface

[ocots]

GUI for the control-toolbox ecosystem.

Some starting examples

• ct playground: a Pluto notebook where we can define an optimal control problem, solve it and display the solution. The notebook is interactive and reactive.

• web app: the result of a ``projet long'' from N7 students. We can also define an optimal control problem, solve it and display the solution. The app is not reactive. It is a web app with React.js for the frontend and with `C#` backend.

• Bocop: the bocop software aims to solve optimal control problems by direct method. With this software, you can define the problem, choose the discretization scheme, define the initial guess, tune the solver's parameters and visualize the solution.

[ocots]

I would suggest to create a web-app with components. I recommend to develop the components into separate repositories.

The components I can imagine.

• a code editor specialized to define the optimal control problem.
• a component to specify the options to transcribe the OCP into a DOCP (Discretized Optimal Control Problem), that is the Runge-Kutta scheme for instance.
• a component to define the initial guess that we give to the solve.
• a component to set the solver's options like the tolerances, the AD backend, etc.
• a component to define the process of resolution: single, batch...
• a textual component to visualize the output of the solver in real-time.
• a graphical component to display the solution. We could imagine to display also in real-time each iteration.
• a diagnostic component to help the user to analyze the solution.
• a component to load / save a problem
• a component to load / save a solution. We could imagine that a problem may have different solutions depending on the options, etc.

[ocots]

I note the repo where @PierreMartinon started to code a GUI: https://github.com/control-toolbox/CTGui.