parallel mean solve

CovidAlertVaccinationModel/abm_timeseries.jl

@@ -3,10 +3,10 @@ using OnlineStats
 using Plots
 
 using CovidAlertVaccinationModel:vaccination_data,ymo_vac,ymo_attack_rate
-const samples = 10
+const samples = 25
 
 function solve_and_plot_parameters()
-    p =  CovidAlertVaccinationModel.get_app_parameters()
+    p =  CovidAlertVaccinationModel.get_parameters()
     display(p)
     out,avg_populations = mean_solve(samples, p,DebugRecorder)
     p = plot_model(nothing,[nothing],[out],p.infection_introduction_day,p.immunization_begin_day)

CovidAlertVaccinationModel/src/ABM/model_setup.jl

@@ -13,7 +13,7 @@ function get_parameters()#(0.0000,0.00048,0.0005,0.16,-1.30,-1.24,-0.8,0.35,0.35
         π_base_y = -1.37,
         π_base_m = -1.46,
         π_base_o = -0.95,
-        η = 0.01,
+        η = 0.0,
         κ = 0.0,
         ω = 0.0055,
         ω_en = 0.0,

CovidAlertVaccinationModel/src/ABM/output.jl

@@ -125,12 +125,15 @@ end
 
 function mean_solve(samples,parameter_tuple,recorder;progmeter = nothing)
     stat_recorder = recorder(Variance(), parameter_tuple.sim_length)
-    output_recorder = recorder(0.0,parameter_tuple.sim_length)
     avg_populations = [0.0,0.0,0.0]
-    for _ in 1:samples
+    sol_list = ThreadsX.map(1:samples) do _
+        output_recorder = recorder(0.0,parameter_tuple.sim_length)
         sol = abm(parameter_tuple,output_recorder)
         isnothing(progmeter) || next!(progmeter)
-        avg_populations .+= length.(sol.index_vectors)
+        return output_recorder,length.(sol.index_vectors)
+    end
+    for (output_recorder,pop) in sol_list
+        avg_populations .+= pop
         fit!(stat_recorder,output_recorder)
     end
     avg_populations ./= samples
@@ -140,11 +143,14 @@ end
 function mean_solve(samples,parameter_tuple,recorder::Type{HeatmapRecorder};progmeter = nothing)
     stat_recorder = recorder(parameter_tuple.sim_length)
     avg_populations = [0.0,0.0,0.0]
-    for _ in 1:samples
+    sol_list = ThreadsX.map(1:samples) do _
         output_recorder = recorder(parameter_tuple.sim_length)
         sol = abm(parameter_tuple,output_recorder)
         isnothing(progmeter) || next!(progmeter)
-        avg_populations .+= length.(sol.index_vectors)
+        return output_recorder,length.(sol.index_vectors)
+    end
+    for (output_recorder,pop) in sol_list
+        avg_populations .+= pop
         fit!(stat_recorder,output_recorder)
     end
     avg_populations ./= samples

CovidAlertVaccinationModel/src/ABM/solve.jl

@@ -113,10 +113,13 @@ Base.@propagate_inbounds @views function update_vaccination_opinion_state!(t,mod
         if !isempty(neighbors(random_soc_network,i))
                 random_neighbour = sample(Random.default_rng(Threads.threadid()), neighbors(random_soc_network.g,i))
                 if u_vac[random_neighbour] == u_vac[i]
-                    vac_payoff = π_base[Int(demographics[i])] + total_infections*ω             
+
+                    app_vac_payoff = 0.0    
                     if app_user[i] && time_of_last_alert[app_user_list[i]]>=0
-                        vac_payoff += Γ^(-1*(t - time_of_last_alert[app_user_list[i]])) * (η + total_infections*ω_en) 
+                        app_vac_payoff = Γ^(-1*(t - time_of_last_alert[app_user_list[i]])) * (η + total_infections*ω_en) 
                     end
+                    vac_payoff = π_base[Int(demographics[i])] + total_infections*ω + app_vac_payoff
+
                     if u_vac[i]
                         # display(1 - Φ(vac_payoff,ξ))
                         if rand(Random.default_rng(Threads.threadid())) < 1 - Φ(vac_payoff,ξ)