Time grids

A solution stores the discretisation points at which its trajectories were computed. CTModels supports either one shared grid or one grid per component, and chooses the representation automatically.

Type	When	Stores
UnifiedTimeGridModel	all components share a grid	a single vector
MultipleTimeGridModel	state/control/costate/path differ	a named tuple of grids
EmptyTimeGridModel	no discretisation yet	nothing

using CTModels

pre = CTModels.PreModel()
CTModels.variable!(pre, 0)
CTModels.time!(pre; t0=0.0, tf=1.0)
CTModels.state!(pre, 2)
CTModels.control!(pre, 1)
CTModels.dynamics!(pre, (r, t, x, u, v) -> (r[1] = x[2]; r[2] = u[1]; nothing))
CTModels.objective!(pre, :min; lagrange=(t, x, u, v) -> 0.5u[1]^2)
CTModels.time_dependence!(pre; autonomous=true)
ocp = CTModels.build(pre)

One shared grid

When the four grids passed to build_solution are identical, CTModels collapses them into a UnifiedTimeGridModel to save memory:

T = collect(range(0.0, 1.0; length=101))
X = [1.0 - t/100 for t in 1:101, i in 1:2]
U = [sin(t/50) for t in 1:101, i in 1:1]
P = zeros(101, 2)

sol_u = CTModels.build_solution(ocp, T, T, T, T, X, U, Float64[], P;
    objective=0.5, iterations=10, constraints_violation=1e-6,
    message="ok", status=:optimal, successful=true)

julia> CTModels.time_grid_model(sol_u) isa CTModels.UnifiedTimeGridModel
true

julia> length(CTModels.time_grid(sol_u))
101

One grid per component

Passing different grids for state, control, costate and path yields a MultipleTimeGridModel; each component is then queried by name through time_grid:

T_state   = collect(range(0.0, 1.0; length=101))
T_control = collect(range(0.0, 1.0; length=51))
T_costate = collect(range(0.0, 1.0; length=76))
T_path    = collect(range(0.0, 1.0; length=61))

X = [1.0 - t/100 for t in 1:101, i in 1:2]
U = [sin(t/25) for t in 1:51, i in 1:1]
P = zeros(76, 2)

sol_m = CTModels.build_solution(ocp, T_state, T_control, T_costate, T_path,
    X, U, Float64[], P;
    objective=0.5, iterations=10, constraints_violation=1e-6,
    message="ok", status=:optimal, successful=true)

julia> CTModels.time_grid_model(sol_m) isa CTModels.MultipleTimeGridModel
true

julia> length(CTModels.time_grid(sol_m, :state))
101

julia> length(CTModels.time_grid(sol_m, :control))
51

julia> length(CTModels.time_grid(sol_m, :costate))
76

Component aliases

time_grid(sol, :states), time_grid(sol, :duals), … accept many synonyms. They are normalised to the four canonical grids (:state, :control, :costate, :path) by clean_component_symbols:

julia> CTModels.clean_component_symbols((:states, :controls, :costate, :constraint, :duals))
(:state, :control, :costate, :path)

This is why box-constraint duals share the state/control grids, and path-constraint duals the :path grid — the mapping is centralised in one place rather than scattered across accessors.

Checking for emptiness

is_empty returns true for an EmptyTimeGridModel and false otherwise. The convenience is_empty_time_grid calls is_empty on the solution's time grid model directly.