basic load/save solution in julia JLD2 and JSON format (#104)

* basic load/save solution in julia JLD2 format

* added functions for load/save; move to CTBase once finished

* add new struct for OCP solution in discrete form for export

* load / save in JSON format (interpolated solution)

.gitignore

@@ -6,8 +6,9 @@
 *.jl.mem
 *.jl.*.mem
 
-# Julia data files
+# Save / load solution data files
 *.jld2
+*.json
 
 # System-specific files and directories generated by the BinaryProvider and BinDeps packages
 # They contain absolute paths specific to the host computer, and so should not be committed

Project.toml

@@ -8,9 +8,11 @@ ADNLPModels = "54578032-b7ea-4c30-94aa-7cbd1cce6c9a"
 CTBase = "54762871-cc72-4466-b8e8-f6c8b58076cd"
 DocStringExtensions = "ffbed154-4ef7-542d-bbb7-c09d3a79fcae"
 JLD2 = "033835bb-8acc-5ee8-8aae-3f567f8a3819"
+JSON3 = "0f8b85d8-7281-11e9-16c2-39a750bddbf1"
 LinearAlgebra = "37e2e46d-f89d-539d-b4ee-838fcccc9c8e"
 NLPModels = "a4795742-8479-5a88-8948-cc11e1c8c1a6"
 NLPModelsIpopt = "f4238b75-b362-5c4c-b852-0801c9a21d71"
+StructTypes = "856f2bd8-1eba-4b0a-8007-ebc267875bd4"
 SymbolicUtils = "d1185830-fcd6-423d-90d6-eec64667417b"
 Symbolics = "0c5d862f-8b57-4792-8d23-62f2024744c7"
 

src/CTDirect.jl

@@ -6,6 +6,9 @@ using Symbolics                 # for optimized AD
 using ADNLPModels               # docp model with AD
 using NLPModelsIpopt            # NLP solver
 using LinearAlgebra             # norm
+using JLD2
+using JSON3
+using StructTypes
 
 # Other declarations
 const __grid_size_direct() = 100
@@ -30,6 +33,9 @@ export getNLP
 export setDOCPInit
 export OCPSolutionFromDOCP
 export OCPSolutionFromDOCP_raw
+export save_OCP_solution
+export load_OCP_solution
+export OCP_Solution_discrete
 
 # CTBase reexports
 export @def
@@ -46,4 +52,6 @@ export remove_constraint!
 export OCPInit
 export plot
 
+
+
 end

src/utils.jl

@@ -199,4 +199,103 @@ function DOCP_initial_guess(docp, init::OCPInit=OCPInit())
     end
 
     return xuv
-end
+end
+
+
+
+#+++ to be moved to CTBase !
+
+#struct for interpolated ocp solution with only basic data types that can be exported as json
+# +++todo:
+# - pass time grid / grid size
+# - add more fields from OptimalControlSolution
+# - constructor to recreate OptimalControlSolution from this one
+mutable struct OCP_Solution_discrete
+
+    grid_size
+    objective
+    times
+    #initial_time_name::Union{String, Nothing}=nothing
+    #final_time_name::Union{String, Nothing}=nothing
+    #time_name::Union{String, Nothing}=nothing
+    control_dimension
+    #control_components_names::Union{Vector{String}, Nothing}=nothing
+    #control_name::Union{String, Nothing}=nothing
+    control
+    state_dimension
+    #state_components_names::Union{Vector{String}, Nothing}=nothing
+    #state_name::Union{String, Nothing}=nothing
+    state
+    variable_dimension
+    #variable_components_names::Union{Vector{String}, Nothing}=nothing
+    #variable_name::Union{String, Nothing}=nothing
+    variable
+    costate
+    #objective::Union{Nothing, ctNumber}=nothing
+    #iterations::Union{Nothing, Integer}=nothing
+    #stopping::Union{Nothing, Symbol}=nothing # the stopping criterion
+    #message::Union{Nothing, String}=nothing # the message corresponding to the stopping criterion
+    #success::Union{Nothing, Bool}=nothing # whether or not the method has finished successfully: CN1, stagnation vs iterations max
+    #infos::Dict{Symbol, Any}=Dict{Symbol, Any}()
+    #OCP_Solution_discrete() = new() # for StructTypes / JSON
+
+    function OCP_Solution_discrete(solution::OptimalControlSolution)
+        solution_d = new()
+
+        # raw copy
+        solution_d.objective = solution.objective
+        solution_d.times = solution.times
+        solution_d.state_dimension = solution.state_dimension
+        solution_d.control_dimension = solution.control_dimension
+        solution_d.variable_dimension = solution.variable_dimension
+        solution_d.variable = solution.variable
+
+        # interpolate functions into vectors
+        # +++ ther *must* be a quicker way to do this -_-
+        solution_d.grid_size = length(solution_d.times) - 1
+        solution_d.state = zeros(solution_d.grid_size+1, solution_d.state_dimension)
+        solution_d.control = zeros(solution_d.grid_size+1, solution_d.control_dimension)
+        solution_d.costate = zeros(solution_d.grid_size+1, solution_d.state_dimension)
+        for i in 1:solution_d.grid_size
+            solution_d.state[i,:] .= solution.state(solution_d.times[i])
+            solution_d.control[i,:] .= solution.control(solution_d.times[i])
+            solution_d.costate[i,:] .= solution.costate(solution_d.times[i])
+        end
+        return solution_d
+    end
+end
+
+"""
+$(TYPEDSIGNATURES)
+  
+Save OCP solution in JLD2/JSON format
+"""
+function save_OCP_solution(sol::OptimalControlSolution; filename_prefix="solution", format="JLD2")
+    if format == "JLD2"
+        save_object(filename_prefix * ".jld2", sol)
+    elseif format == "JSON"
+        open(filename_prefix * ".json", "w") do io
+            JSON3.pretty(io, OCP_Solution_discrete(sol))
+        end
+    else
+        println("ERROR: save_OCP_solution: format should be JLD2 or JSON, received ", format)
+    end
+    return nothing
+end
+
+"""
+$(TYPEDSIGNATURES)
+ 
+Load OCP solution in JLD2/JSON format
+"""
+function load_OCP_solution(filename_prefix="solution"; format="JLD2")
+    if format == "JLD2"
+        return load_object(filename_prefix * ".jld2")
+    elseif format == "JSON"
+        json_string = read(filename_prefix * ".json", String)
+        return JSON3.read(json_string)
+    else
+        println("ERROR: save_OCP_solution: format should be JLD2 or JSON, received ", format)
+        return nothing
+    end
+end

test/suite/misc.jl

@@ -1,5 +1,4 @@
 using CTDirect
-using JLD2
 
 println("Test: misc")
 
@@ -45,7 +44,14 @@ sol_raw = OCPSolutionFromDOCP_raw(docp, dsol.solution)
 
 # test save / load solution in JLD2 format
 @testset verbose = true showtiming = true ":save_load :JLD2" begin
-    save_object("sol.jld2", sol0)
-    sol_reloaded = load_object("sol.jld2")
+    save_OCP_solution(sol0, filename_prefix="solution_test")
+    sol_reloaded = load_OCP_solution("solution_test")
     @test sol0.objective == sol_reloaded.objective
 end
+
+# test save / load solution in JSON format
+@testset verbose = true showtiming = true ":save_load :JSON" begin
+    save_OCP_solution(sol0, filename_prefix="solution_test", format="JSON")
+    sol_reloaded = load_OCP_solution("solution_test", format="JSON")
+    @test sol0.objective == sol_reloaded.objective
+end

test/test_misc.jl

@@ -1,5 +1,4 @@
 using CTDirect
-using JLD2
 
 println("Test: misc")
 
@@ -42,11 +41,16 @@ dsol2 = solve(docp2, print_level=5, tol=1e-12)
 println("\nRebuild OCP solution from raw vector")
 sol3 = OCPSolutionFromDOCP_raw(docp2, dsol2.solution)
 
-# save / load solution in JLD2 format (solution includes complex data such as interpolated functions which are less suitable for more generic formats such as JSON)
-save_object("sol.jld2", sol)
-sol4 = load_object("sol.jld2")
+# save / load solution in JLD2 format
+save_OCP_solution(sol, filename_prefix="solution_test")
+sol4 = load_OCP_solution("solution_test")
 plot(sol4, show=true)
 println(sol.objective == sol4.objective)
 
+# save / load discrete solution in JSON format
+# NB. we recover here a JSON Object...
+save_OCP_solution(sol, filename_prefix="solution_test", format="JSON")
+sol_disc_reloaded = load_OCP_solution("solution_test", format="JSON")
+println(sol.objective == sol_disc_reloaded.objective)
+
 println("")
-#