How can I use GlobalSearch with Parallelization?
2 Ansichten (letzte 30 Tage)
Ältere Kommentare anzeigen
Hello, I am trying the code below which uses 'parfor' and 'globalsearch' at the same time.
But I got an warning message saying 'The function RUN might make an invalid workspace access inside PARFOR loop'
I am not sure what this means. But when I run the code ignoring such message, something is running but I am a bit worried that this would give me wrong results.
So I was wondering whether this warning is something that I can just ignore or I should not use PARFOR and GLOBALSEARCH(especailly RUN) at the same time.
Thanks in advance.
parfor i_a = 1:Na %Loop over state variable a
for i_d = 1:Nd %Loop over state variable d
for i_y = 1:Ny %Loop over state variable y
for i_t = 1:Nt %Loop over state variable y
noadjamount = (1-delta)*D(i_d);
if noadjamount < D(1)
noadjamount = D(1);
end
problem1 = createOptimProblem('fmincon',...
'objective',@(x)-adjvalue_model1(x,i_a,i_d,i_y,i_t,Utility,A,D,Y,T,R,delta,fixed,Interpol_1,Na,Nd),...
'x0',[x0(i_a,i_d,i_y,i_t); x1(i_a,i_d,i_y,i_t)],'options',...
optimoptions(@fmincon,'Algorithm','sqp','Display','off'));
gs1 = GlobalSearch('Display','off');
rng(14,'twister') % for reproducibility
[adj_sol, adjval] = run(gs1,problem1);
problem2 = createOptimProblem('fmincon',...
'objective',@(x)-noadjvalue_model1(x,i_a,i_d,i_y,i_t,Utility,A,D,Y,T,R,delta,fixed,Interpol_1,Na,Nd),...
'x0',x2(i_a,i_d,i_y,i_t),'options',...
optimoptions(@fmincon,'Algorithm','sqp','Display','off'));
gs2 = GlobalSearch('Display','off');
rng(14,'twister') % for reproducibility
[noadj_sol, noadjval] = run(gs2,problem2);
[Vnew(i_a,i_d,i_y,i_t), ind] = max([-adjval, -noadjval],[],2);
% Store values needed(Policy function, Value, Adjust or not)
pol_ap(i_a,i_d,i_y,i_t) = (ind == 1).* adj_sol(1) + (ind == 2).*noadj_sol;
pol_dp(i_a,i_d,i_y,i_t) = (ind == 1).* adj_sol(2) + (ind == 2).*noadjamount;
indadj(i_a,i_d,i_y,i_t) = (ind == 1).*1;
indnadj(i_a,i_d,i_y,i_t) = (ind == 2).*1;
x0(i_a,i_d,i_y,i_t) = adj_sol(1);
x1(i_a,i_d,i_y,i_t) = adj_sol(2);
x2(i_a,i_d,i_y,i_t) = noadj_sol;
end
end
end
end
0 Kommentare
Antworten (1)
Siehe auch
Kategorien
Mehr zu Surrogate Optimization finden Sie in Help Center und File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!