Error using fmincon Supplied objective function must return a scalar value

C1=log(1+(h(1,1).*P(1,1)/n+x1));

C2=log(1+(h(2,2).*P(2,1)/n+x2));

C3=log(1+(h(3,3).*P(3,1)/n+x3));

C4=log(1+(h(4,4).*P(4,1)/n+x4));

C=[C1,C2,C3,C4];

M=4;

w=linspace(0,1,M);

C_sum= C.*w;

Pimax(1:4)= 10^-4;

x0 = ones(64,1);

c_lin = [H',zeros(1,48);zeros(1,16),H',zeros(1,32);zeros(1,32),H',zeros(1,16);zeros(1,48),H'];

b = Pimax;

X = fmincon(@myfunc,x0,c_lin,b);

function C_sum1=myfunc(C_sum)

C_sum1=C_sum;

end

[SWASTIK SAHOO]

clc;

clear all;

close all;

x1=0;

x2=0;

x3=0;

x4=0;

n=10^-9; %noise

N= 16; % number of channels

h_rayleigh=(1/sqrt(2)).*(randn(N,1)+1i*randn(N,1));

h_abs= abs(h_rayleigh);

H= h_abs;

h=reshape(h_abs,4,4);

syms A [4 16] matrix

P= A*H;

i=1;

for j=1:1:4

if(j~=i)

x1= x1+h(j,i).*P(j,1);

else

x1=x1;

end

end

i=2;

for j=1:1:4

if(j~=i)

x2= x2+h(j,i).*P(j,1);

else

x2=x2;

end

end

i=3;

for j=1:1:4

if(j~=i)

x3= x3+h(j,i).*P(j,1);

else

x3=x3;

end

end

i=4;

for j=1:1:4

if(j~=i)

x4= x4+h(j,i).*P(j,1);

else

x4=x4;

end

end

X=[x1,x2,x3,x4];

size(X);

C1=log(1+(h(1,1).*P(1,1)/n+x1));

C2=log(1+(h(2,2).*P(2,1)/n+x2));

C3=log(1+(h(3,3).*P(3,1)/n+x3));

C4=log(1+(h(4,4).*P(4,1)/n+x4));

C=[C1 C2 C3 C4];

M=4;

w=linspace(0,1,M);

C_sum= C.*w;

Pimax(1:4)= 10^-4;

x0 = ones(64,1);

c_lin = [H',zeros(1,48); zeros(1,16), H',zeros(1,32);zeros(1,32),H',zeros(1,16);zeros(1,48),H'];

b = Pimax;

X = fmincon(@myfunc,x0,c_lin,b);

function C_sum1=myfunc(C_sum)

C_sum1=C_sum;

end

[SWASTIK SAHOO]

The above is the full code.. please help

thanks in advance

Answer 1

[Sam Chak]

HI @SWASTIK SAHOO

The error message says that your objective function "myfunc()" must return a scalar value. Perhaps the trapz() function might work. I'm unfamiliar with the physics of Rayleigh scattering. So, you need to check if the proposed objective function is acceptable.

x1         = 0;
x2         = 0;
x3         = 0;
x4         = 0;
n          = 1e-9;  % noise
N          = 16;    % number of channels
h_rayleigh = (1/sqrt(2)).*(randn(N,1) + 1i*randn(N,1));
h_abs      = abs(h_rayleigh);
H          = h_abs;
h          = reshape(h_abs, 4, 4);

syms A [4 16] matrix
P = A*H;

i = 1;
for j = 1:1:4
    if(j ~= i)
        x1 = x1 + h(j,i).*P(j,1);
    else
        x1 = x1;
    end
end

i = 2;
for j = 1:1:4
    if(j ~= i)
        x2 = x2 + h(j,i).*P(j,1);
    else
        x2 = x2;
    end
end

i = 3;
for j = 1:1:4
    if(j ~= i)
        x3 = x3 + h(j,i).*P(j,1);
    else
        x3 = x3;
    end
end

i = 4;
for j = 1:1:4
    if(j ~= i)
        x4 = x4 + h(j,i).*P(j,1);
    else
        x4 = x4;
    end
end

X          = [x1, x2, x3, x4];
size(X);
C1         = log(1 + (h(1,1).*P(1,1)/n + x1));
C2         = log(1 + (h(2,2).*P(2,1)/n + x2));
C3         = log(1 + (h(3,3).*P(3,1)/n + x3));
C4         = log(1 + (h(4,4).*P(4,1)/n + x4));
C          = [C1 C2 C3 C4];
M          = 4;
w          = linspace(0, 1, M);
C_sum      = C.*w;
Pimax(1:4) = 10^-4;
x0         = ones(64,1);
c_lin      = [H', zeros(1,48); zeros(1,16), H', zeros(1,32); zeros(1,32), H', zeros(1,16); zeros(1,48), H'];
b          = Pimax;

options = optimoptions(@fmincon, 'MaxFunctionEvaluations', 3e6);
X       = fmincon(@myfunc, x0, c_lin, b, [], [], [], [], [], options);

Problem appears unbounded.

fmincon stopped because the objective function value is less than
the value of the objective function limit and constraints 
are satisfied to within the value of the constraint tolerance.

function C_sum1 = myfunc(C_sum)
    C_sum1 = trapz(C_sum);
end

[SWASTIK SAHOO]

Thank you so much....