search

control-toolbox / CTFlows.jl

Flows: classical, Hamiltonian, from OCP and more
http://control-toolbox.org/CTFlows.jl/
MIT License
1 stars 1 forks source link

Hamiltonian flows #13

Open jbcaillau opened 11 months ago

jbcaillau commented 11 months ago

@ocots The previous package to compute Hamiltonian flows is deprecated, is it OK to use stuff from this package and hamiltonian.jl to retrieve the same functionality (= flow of a Hamiltonian, not of an ocp with a given maximising control)?

Seems to be OK from the code, not in the documentation. Any example somewhere?

ocots commented 11 months ago

We can of course compute a flow from a Hamiltonian. Some doc is needed indeed.

ocots commented 11 months ago

See here for examples: https://github.com/control-toolbox/CTFlows.jl/blob/main/test/test_flow_hamiltonian.jl

ocots commented 11 months ago

I repeat that we need to define an AbstractFlowModel with methods and use it to define flows. It permits to write a generic code and to instantiate many computations of flows simply defining a subtype of the abstract type and new to overload the methods.

Like this we can benefit from the concatenate function, plot and more.

If a user wants to define its own flow he can. It is a nice feature to develop new algorithms.

jbcaillau commented 11 months ago

@ocots thanks for the pointer to the tests (should've checked.) neat.

agree about having an abstraction for flows: any open issue on this?