I am getting this error while solving "fsolve stopped because it exceeded the function evaluation limit,

3 次查看（过去 30 天）
Pankaj Khatiwada
Pankaj Khatiwada 2017-8-18
评论： Rena Berman 2017-8-30
file 1
% model 2 main part, cal & plot
tic;
N = 10; % N = 20
lambda = 0.1:1;
S = [1:50];
PL = 1024;
for i = 1:50
x0 = [1; 1; 1; 0; 1; 0; 1346; 0.0027]; % wired start point
options = optimoptions('fsolve');
[x,fval] = fsolve(@(x)m2_numeric_solver2(x,N,lambda(i)),x0,options);
Data = x(5) * x(6) * PL;
Time = x(7);
S(i) = (Data / Time)/10^6;
formatSpec = 'iteration %d complete\n';
fprintf(formatSpec,i)
end
plot(S);
title('Unsaturation throughput vs arrival rate');
xlabel('arrival rate');
ylabel('saturation throughput');
grid on;
toc;
file 2
% model 2 numberic solver with capture effct and error rate
% x(1) -- tao
% x(2) -- Pcol
% x(3) -- Peq
% x(4) -- Pcap
% x(5) -- Pt
% x(6) -- Ps
% x(7) -- E[T]
% x(8) -- q
% notice: warning i, confuse with complex 'i'
% notice: double for symsum
function F = m2_numeric_solver2(x,N,lambda)
syms k;
z = 4; % << capture threshold
W = 32;
m = 5;
Pe = 0.01; % << channel error rate
tao = 1/(10^6);
Ts = 1140/(10^6);
Tc = 1364/(10^6);
Te = 1364/(10^6);
F = [
x(1) - 2*(1-2*x(3))*x(8) / ( x(8)*( (W+1)*(1-2*x(3)) + W*x(3)*(1 - (2*x(3))^m)) ...
+ 2 * (1-x(8)) * (1-x(3)) * (1-2*x(3)) );
x(2) - ( 1 - (1-x(1)) ^ (N-1) - x(4) );
x(3) - ( x(2) + Pe - Pe*x(2) );
x(4) - double( symsum( nchoosek(N,k+1) * x(1)^(k+1)*(1-x(1))^(N-k-1) / (1+z)^k, 1, N-1) );
x(5) - (1 - (1 - x(1))^N);
x(6) - (N*x(1)*(1-x(1))^(N-1) + x(4)) / x(5);
x(7) - ( (1-x(5))*tao + x(5)*(1-x(6))*Tc + x(5)*x(6)*Pe*Te + + x(5)*x(6)*(1-Pe)*Ts );
x(8) - (1 - exp(-lambda*x(7)) );
];
end

请先登录，再回答此问题。

采纳的回答

Walter Roberson
Walter Roberson 2017-8-18
options = optimoptions('fsolve');
options.MaxIterations = 1000;
options.MaxFunctionEvaluations = 5000;

更多回答（1 个）

Alan Weiss
Alan Weiss 2017-8-18
You wrote
lambda = 0.1:1;
This gives the same value as
lambda = 0.1;
Later you call lambda(i) for i from 1 through 50. This makes no sense. Maybe you meant
lambda = linspace(0.1,1,50);
I would try to not use symbolic variables inside the objective function. If you really need them, perhaps use matlabFunction to convert any symbolic function into a numeric function before you use it. See Using Symbolic Mathematics with Optimization Toolbox™ Solvers.
One more thing: don't set options inside the for loop. Set them at the beginning, before the loop, because setting options is slow.
Alan Weiss
MATLAB mathematical toolbox documentation

类别

在 Help Center 和 File Exchange 中查找有关 Symbolic Math Toolbox 的更多信息

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by