Error while adding constraints

1

Attempt to add “cvx_problem” to a static workspace

2

Please post a complete MATLAB session showing all code and (error) messages.

3

This is the code snippet%%%%%%

cvx_begin quiet
cvx_solver sdpt3
cvx_save_prefs
variables alpha_1 beta_1 x(N,1) ;
% variables beta_1;
% variables x(N,1);
variable X_fin(N,1) complex;
minimize quad_over_lin(alpha_1, beta_1);
subject to
for t=1:N
abs(x(t))<=alpha_1;
end
subject to
for s=1:N
real(X_fin(s)'f(s)) <= beta_1;
end
subject to
X_fin(1)==0;
if(mod(N,2)==0)
X_fin(N/2+1)==0;
% % real(w(1))==real(w(M/2+1));
for k=1:(N/2-1)
X_fin(k+1)==X_fin(N-k+1)’;
end
else
for i=1:(N-1)/2
X_fin(i+1)==X_fin(N-i+1)’;
end
end
subject to
X_fin==Sx;

cvx_end

ERROR%%%%%%%%%%%%

Error using variable (line 83)
Invalid variable name: complex
This is a reserved word in CVX.
Trying to declare a structured matrix? Use the VARIABLE keyword
instead.

Error in variables (line 26)
evalin( ‘caller’, [ 'variable ', varargin{k} ] );

Error in test_new>Min_peak (line 81)
variables alpha_1 beta_1 x(N,1) X_fin(N,1) complex;

Error in test_new>Min_Peak_1 (line 49)
X_opt=Min_peak(N,X) ;

Error in test_new (line 8)
[o,x_opt,X_opt,X_in]=Min_Peak_1(N);

4

I would make matters easier if you posted the code which generated the error message, not a different version of your code.

Quite simply, you can’t declare complex or other variable keywords when using variables because CVX does not know that they are not additional variable names. So to declare complex, you need to use variable.

variables alpha_1 beta_1
variable x(N,1) complex
variable X_fin(N,1) complex;

or whatever you want

5

Hi Mark,
Thank you for your response. This is the same code for which I am getting errors.
As suggested, I have made changes in my code. Now I am getting this error.

%%%%%%%%%%Code
cvx_begin quiet
cvx_solver sdpt3
cvx_save_prefs
variables alpha_1 beta_1;
variable x(N,1) complex;
% variables beta_1;
% variables x(N,1);
variable X_fin(N,1) complex;
minimize quad_over_lin(alpha_1, beta_1);
subject to
for t=1:N
abs(x(t))<=alpha_1;
end

subject to
for s=1:N
real(X_fin(s)'exp(jangle(f(s))))<=beta_1;
end

subject to
X_fin(1)==0;
if(mod(N,2)==0)
X_fin(N/2+1)==0;
% % real(w(1))==real(w(M/2+1));
for i=1:(N/2-1)
X_fin(i+1)==X_fin(N-i+1)’;
end
else
for i=1:(N-1)/2
X_fin(i+1)==X_fin(N-i+1)’;
end
end

subject to
X_fin==S*x;

cvx_end

%%%%%Error
Error using .* (line 173)
Disciplined convex programming error:
Cannot perform the operation: {complex affine} .* {invalid}

Error in Min_peak_new (line 29)
real(X_fin(s)'exp(jangle(f(s))))<=beta_1;

6

{invalid} most likely means the input data used there is NaN. Perhaps angle(f(s))) for at least one value of s is 'NaN. If you set this input data from output of a previous optimization, it may be that that optimization did not solve to optimality, so CVX variables might have been NaN. Don't use quiet` mode until you have verified everything is working correctly in your program.