Keep MC running even when solver fails.

src/monte_carlo.jl

@@ -83,8 +83,14 @@ function mc(
     params_i = NamedTuple(zip(first.(parameters_variation), values_i))
     # take simulated values from solution
     sv = sol.prob.kwargs[:callback].discrete_callbacks[1].affect!.saved_values
-    simulation = Simulation(sv, params_i, sol.retcode)
-
+    #simulation = Simulation(sv, params_i, sol.retcode)
+    if SciMLBase.successful_retcode(sol)
+      #println("Iteration is successful, val $(size(sv.u))") 
+      simulation = Simulation(sv, params_i, sol.retcode)
+    else
+      #println("Iteration is failed with code $(sol.retcode), val $(size(sv.u))")
+      simulation = Simulation( params_i, sol.retcode)
+    end
     return (output_func(simulation, i), false)
   end
 
@@ -254,7 +260,14 @@ function mc(
     params_i = parameters_pregenerated[iter_i[1]]
     # take simulated values from solution
     sv = sol.prob.kwargs[:callback].discrete_callbacks[1].affect!.saved_values
-    simulation = Simulation(sv, params_i, sol.retcode)
+    #simulation = Simulation(sv, params_i, sol.retcode)
+    if SciMLBase.successful_retcode(sol)
+      #println("Iteration is successful, val $(size(sv.u))") 
+      simulation = Simulation(sv, params_i, sol.retcode)
+    else
+      #println("Iteration is failed with code $(sol.retcode), val $(size(sv.u))")
+      simulation = Simulation( params_i, sol.retcode)
+    end
 
     return (output_func(simulation, i), false)
   end

src/simulate.jl

@@ -65,7 +65,13 @@ function sim(
   #parameters_names = Symbol[first(x) for x in parameters]
   #return build_results(sol, scenario, parameters_names)
   sv = sol.prob.kwargs[:callback].discrete_callbacks[1].affect!.saved_values
-  simulation = Simulation(sv, parameters_nt, sol.retcode)
+  if SciMLBase.successful_retcode(sol)
+    println("Iteration is successful, val $(size(sv.u))") 
+    simulation = Simulation(sv, parameters_nt, sol.retcode)
+  else
+    println("Iteration is failed with code $(sol.retcode), val $(size(sv.u))")
+    simulation = Simulation( parameters_nt, sol.retcode)
+end
 
   return SimResult(simulation, scenario)
 end

src/solution_interface.jl

@@ -93,12 +93,18 @@ end
 
 function DataFrame(mcr::MCResult; itercol=true, kwargs...)
   # df performance
-  df = DataFrame(mcr[1]; iter=(itercol ? 1 : nothing), kwargs...)
+  j = 1
   lmcr = length(mcr)
-
+  while mcr[j].status != Symbol(SciMLBase.ReturnCode.Success)
+    j+=1
+  end
+  df = DataFrame(mcr[j]; iter=(itercol ? 1 : nothing), kwargs...)
+  j+=1
   if lmcr > 1
-    for i in 2:lmcr
-      append!(df, DataFrame(mcr[i]; iter=(itercol ? i : nothing), kwargs...))
+    for i in j:lmcr
+      if mcr[i].status == Symbol(SciMLBase.ReturnCode.Success)
+        append!(df, DataFrame(mcr[i]; iter=(itercol ? i : nothing), kwargs...))
+      end
     end
   end
   return df

src/types.jl

@@ -237,6 +237,13 @@ Simulation(sv::SavedValues, params, status) = Simulation(
   Symbol(status)
 ) 
 
+# copy fix is tmp needed not to rewrite SavedValues with new simulation
+Simulation( params, status) = Simulation(
+  DiffEqArray([0],[0]),
+  Symbol("error"),
+  params,
+  Symbol(status)
+) 
 status(s::Simulation) = s.status
 times(s::Simulation) = s.vals.t
 vals(s::Simulation) = s.vals.u