Merge branch 'main' into obc

Project.toml

@@ -68,7 +68,7 @@ RuntimeGeneratedFunctions = "^0.5"
 SpecialFunctions = "^2"
 SpeedMapping = "^0.3"
 Subscripts = "^0.1"
-SymPyPythonCall = "^0.1.1"
+SymPyPythonCall = "^0.2"
 Symbolics = "^5"
 ThreadedSparseArrays = "^0.2"
 julia = "1.8"

docs/src/unfinished_docs/todo.md

@@ -8,6 +8,7 @@
 - [ ] nonlinear conditional forecasts
 - [ ] add balanced growth path handling
 - [ ] feedback: write out RBC equations, provide option for external SS guess, sell the sampler better (ESS vs dynare), more details on algorithm (SS solver)
+- [ ] higher order solutions: some kron matrix mults are later compressed. write custom compressed kron mult; check if sometimes dense mult is faster? (e.g. GNSS2010 seems dense at higher order)
 - [ ] recheck function examples and docs (include output description)
 - [ ] riccati with analytical derivatives (much faster if sparse) instead of implicit diff
 - [ ] add user facing option to choose sylvester solver

src/MacroModelling.jl

@@ -1527,7 +1527,7 @@ function solve_steady_state!(𝓂::ℳ, symbolic_SS, Symbolics::symbolics; verbo
 
     SS_solve_func = []
 
-    atoms_in_equations = Set()
+    atoms_in_equations = Set{Symbol}()
     atoms_in_equations_list = []
     relevant_pars_across = []
     NSSS_solver_cache_init_tmp = []
@@ -1568,16 +1568,17 @@ function solve_steady_state!(𝓂::ℳ, symbolic_SS, Symbolics::symbolics; verbo
             soll = try SPyPyC.solve(eq_to_solve,var_to_solve_for)
             catch
             end
-
+            
             if isnothing(soll)
                 # println("Could not solve single variables case symbolically.")
                 println("Failed finding solution symbolically for: ",var_to_solve_for," in: ",eq_to_solve)
                 # solve numerically
                 continue
             # elseif PythonCall.pyconvert(Bool,soll[1].is_number)
-            elseif soll[1].is_number == SPyPyC.TRUE
-                # ss_equations = ss_equations.subs(var_to_solve_for,soll[1])
-                ss_equations = [eq.subs(var_to_solve_for,soll[1]) for eq in ss_equations]
+
+            elseif soll[1].is_number == true
+                # ss_equations = ss_equations.subs(var_to_solve,soll[1])
+                ss_equations = [eq.subs(var_to_solve,soll[1]) for eq in ss_equations]
 
                 push!(𝓂.solved_vars,Symbol(var_to_solve_for))
                 push!(𝓂.solved_vals,Meta.parse(string(soll[1])))
@@ -1595,8 +1596,10 @@ function solve_steady_state!(𝓂::ℳ, symbolic_SS, Symbolics::symbolics; verbo
                 push!(𝓂.solved_vals,Meta.parse(string(soll[1])))
 
                 # atoms = reduce(union,soll[1].atoms())
-                [push!(atoms_in_equations, a) for a in soll[1].atoms()]
+
+                [push!(atoms_in_equations, Symbol(a)) for a in soll[1].atoms()]
                 push!(atoms_in_equations_list, Set(union(setdiff(get_symbols(parsed_eq_to_solve_for), get_symbols(minmax_fixed_eqs)),Symbol.(soll[1].atoms()))))
+
                 # println(atoms_in_equations)
                 # push!(atoms_in_equations, soll[1].atoms())
 
@@ -1635,7 +1638,9 @@ function solve_steady_state!(𝓂::ℳ, symbolic_SS, Symbolics::symbolics; verbo
                     end
                     numerical_sol = true
                     # continue
-                elseif length(intersect(vars_to_solve,reduce(union,map(x->x.atoms(),collect(soll[1]))))) > 0
+                # elseif length(intersect(vars_to_solve,reduce(union,map(x->x.atoms(),collect(soll[1]))))) > 0
+                elseif length(intersect((union(SPyPyC.free_symbols.(soll[1])...) .|> SPyPyC.:↓),(vars_to_solve .|> SPyPyC.:↓))) > 0
+                # elseif length(intersect(union(SPyPyC.free_symbols.(soll[1])...),vars_to_solve)) > 0
                     if verbose
                         println("Failed finding solution symbolically for: ",vars_to_solve," in: ",eqs_to_solve,". Solving numerically.")
                     end
@@ -1651,7 +1656,7 @@ function solve_steady_state!(𝓂::ℳ, symbolic_SS, Symbolics::symbolics; verbo
                     # relevant_pars = reduce(union,map(x->x.atoms(),collect(soll[1])))
                     atoms = reduce(union,map(x->x.atoms(),collect(soll[1])))
                     # println(atoms)
-                    [push!(atoms_in_equations, a) for a in atoms]
+                    [push!(atoms_in_equations, Symbol(a)) for a in atoms]
 
                     for (k, vars) in enumerate(vars_to_solve)
                         push!(𝓂.solved_vars,Symbol(vars))
@@ -1679,7 +1684,7 @@ function solve_steady_state!(𝓂::ℳ, symbolic_SS, Symbolics::symbolics; verbo
 
                 for i in eqs_to_solve
                     # push!(syms_in_eqs, Symbol.(PythonCall.pystr.(i.atoms()))...)
-                    push!(syms_in_eqs, Symbol.(SPyPyC.unSym.(SPyPyC.free_symbols(i)))...)
+                    push!(syms_in_eqs, Symbol.(SPyPyC.:↓(SPyPyC.free_symbols(i)))...)
                 end
 
                 # println(syms_in_eqs)
@@ -1943,7 +1948,7 @@ function solve_steady_state!(𝓂::ℳ, symbolic_SS, Symbolics::symbolics; verbo
     parameters_in_equations = []
 
     for (i, parss) in enumerate(𝓂.parameters) 
-        if parss ∈ union(Symbol.(atoms_in_equations),relevant_pars_across)
+        if parss ∈ union(atoms_in_equations, relevant_pars_across)
             push!(parameters_in_equations,:($parss = params[$i]))
         end
     end
@@ -1995,7 +2000,7 @@ function solve_steady_state!(𝓂::ℳ, symbolic_SS, Symbolics::symbolics; verbo
     # fix parameter bounds
     par_bounds = []
 
-    for varpar in intersect(𝓂.bounded_vars, intersect(𝓂.parameters,union(Symbol.(atoms_in_equations),relevant_pars_across)))
+    for varpar in intersect(𝓂.bounded_vars, intersect(𝓂.parameters,union(atoms_in_equations, relevant_pars_across)))
         i = indexin([varpar],𝓂.bounded_vars)
         push!(par_bounds, :($varpar = min(max($varpar,$(𝓂.lower_bounds[i...])),$(𝓂.upper_bounds[i...]))))
     end
@@ -2096,7 +2101,7 @@ function solve_steady_state!(𝓂::ℳ; verbose::Bool = false)
 
     SS_solve_func = []
 
-    atoms_in_equations = Set()
+    atoms_in_equations = Set{Symbol}()
     atoms_in_equations_list = []
     relevant_pars_across = []
     NSSS_solver_cache_init_tmp = []
@@ -2309,7 +2314,7 @@ function solve_steady_state!(𝓂::ℳ; verbose::Bool = false)
     parameters_in_equations = []
 
     for (i, parss) in enumerate(𝓂.parameters) 
-        if parss ∈ union(Symbol.(atoms_in_equations),relevant_pars_across)
+        if parss ∈ union(atoms_in_equations, relevant_pars_across)
             push!(parameters_in_equations,:($parss = params[$i]))
         end
     end
@@ -2351,7 +2356,7 @@ function solve_steady_state!(𝓂::ℳ; verbose::Bool = false)
     # fix parameter bounds
     par_bounds = []
 
-    for varpar in intersect(𝓂.bounded_vars, intersect(𝓂.parameters,union(Symbol.(atoms_in_equations),relevant_pars_across)))
+    for varpar in intersect(𝓂.bounded_vars, intersect(𝓂.parameters,union(atoms_in_equations, relevant_pars_across)))
         i = indexin([varpar],𝓂.bounded_vars)
         push!(par_bounds, :($varpar = min(max($varpar,$(𝓂.lower_bounds[i...])),$(𝓂.upper_bounds[i...]))))
     end

test/test_standalone_function.jl

@@ -1,5 +1,6 @@
 using SparseArrays
-using MacroModelling: timings
+using MacroModelling
+import MacroModelling: timings
 using ForwardDiff, FiniteDifferences, Zygote
 import Optim, LineSearches
 import LinearAlgebra as ℒ