"Index in position 1 is invalid"

Hello,

running this code always returns the error "Index in position 1 is invalid. Array indices must be positive integers or logical values.".

The error is located in the line expressing obj_1.

It's seems to be independent from the for cycles since the same happens if I try to run the line outside.

obj_1 = @(p) 0;
for jj=1:5 %peaks
    n = 40;
    for ii = (floor(omega_r_idx_1(jj))-n):(floor(omega_r_idx_1(jj))+n) %funzione obiettivo su un intervallo n punti prima e dopo
        obj_1 = @(p) obj_1(p) + ((real (H1(freq(ii),5)) -real((freq(ii)^2*p(jj,1))+p(jj,3)./(p(jj,2).^2-freq(ii)^2+...
            2*1i*p(jj,4)*p(jj,2).*freq(ii))+p(jj,5))))^2+((imag(H1(freq(ii),5)) -imag((freq(ii)^2*p(jj,1))+p(jj,3)./(p(jj,2).^2-...
            freq(ii)^2+2*1i*p(jj,4)*p(jj,2).*freq(ii))+p(jj,5))))^2;
    end
end

I already tried to evaluate all the parameters like H1, freq and omega_r_idx_1 in the specified indexes and it works.

Could you please help me?

[Geoff Hayes]

Federico - is H1 an array? Is freq only integers or does it contain doubles? Can there be negative or zero frequency in freq? Also, what is the initial value for ii - can it be zero or does your code guarantee it to be a positive integer?

[FEDERICO ZUCCARI]

Hello Geoff, thanks for replying.

You made me realize that I wanted to evaluate H1(jj,5) and not H1(freq(jj),5) because freq is not an integer and is the frequency array associated to the same indexes (sampling points) of H1.

I will try to fix it and come back to you with a feedback.

EDIT:

Unfortunately, the same error is still present.

H1 is a bidimentional complex double array, freq is an array of doubles that is always positive and different from zero.

Both jj and ii are cleared just before the lines I reported and the minimum value for ii (as defined above) is about 700; ii cannot be negative or zero.

[Geoff Hayes]

Is there anything more to the error message that you have posted above? It would be nice to know what part of the line is causing the problem. Also, i don't understand

obj_1 = @(p) obj_1(p)

since obj_1 is your anonymous function, why are you assigning it to itself? Or are you creating a new anonymous function but that doesn't seem correct either since this is the only (?) line in an inner loop. can you clarify?

[FEDERICO ZUCCARI]

No Geoff, there is nothing more added to the error message and the line that is causing the problem is the one in which I define obj.

To clarify that part: I have to sum the function you see as obj_1 over the index ii and I thought it was good to do that using a for loop.

So, the part you asked about is to take what is evaluated at the step ii-1 and add it to the evaluation at the step ii.

I hope I was clear enough.

[FEDERICO ZUCCARI]

I made tests and now the error is still present but it's different and located in the following part of the code where I use fmincon to minimize the function obj_1.

I report the complete section for the minimization.

clearvars jj ii
 obj_1 = @(p) 0; %inizializzazione

for jj=1:2 %peaks
    n = 2;
    for ii = (floor(omega_r_idx_1(jj))-n):(floor(omega_r_idx_1(jj))+n) %funzione obiettivo su un intervallo n punti prima e dopo
        obj_1 = @(p) obj_1(p) + ((real(H1(ii,5)) -real((freq(ii)^2*p(jj,1))+p(jj,3)./(p(jj,2).^2-freq(ii)^2+2*1i*p(jj,4)*p(jj,2).*freq(ii))+p(jj,5))))^2+((imag(H1(ii,5)) -imag((freq(ii)^2*p(jj,1))+p(jj,3)./(p(jj,2).^2-freq(ii)^2+2*1i*p(jj,4)*p(jj,2).*freq(ii))+p(jj,5))))^2;
    end

%fmincon setup     
    a=[];
    b=[];
    Aeq=[];
    beq=[];
    ub(jj,:) = [1,(omega_r_1(jj)+10),0,(damping_ratio_1(jj)+0.1),1];  %upper bound
    lb(jj,:)=[1e-10,(omega_r_1(jj)-10),-(2*pi)^2,0.001,1e-10];  %lower bound
    M(jj) = 1e-09;
    K(jj)= 1e-09;

    p0_1(jj,:) = [M(jj),omega_r_1(jj),(A_5_1(jj)),damping_ratio_1(jj),K(jj)]; %first guess matrix


% minimization (MESH 1)

options = optimoptions(@fmincon,'Algorithm','sqp','MaxIterations',10000); %set of the max number of iterations 
options = optimoptions(options,'PlotFcns',{@optimplotfval}); %live plot of the iteration process

[p_opt,fval]=fmincon(obj_1,p0_1(jj,:),a,b,Aeq,beq,lb(jj,:),ub(jj,:),[],options);
obj_1 = @(p) 0
end

The error is found at the line showing obj_1 but it happens in the fmincon evaluation; I report the error message for clarity:

Index in position 1 exceeds array bounds (must not exceed 1).

Error in CODE_allmeshes_TEST (line 728)
        obj_1 = @(p) obj_1(p) + ((real(H1(ii,5))
        -real((freq(ii)^2*p(jj,1))+p(jj,3)./(p(jj,2).^2-freq(ii)^2+2*1i*p(jj,4)*p(jj,2).*freq(ii))+p(jj,5))))^2+((imag(H1(ii,5))
        -imag((freq(ii)^2*p(jj,1))+p(jj,3)./(p(jj,2).^2-freq(ii)^2+2*1i*p(jj,4)*p(jj,2).*freq(ii))+p(jj,5))))^2;

Error in CODE_allmeshes_TEST (line 728)
        obj_1 = @(p) obj_1(p) + ((real(H1(ii,5))
        -real((freq(ii)^2*p(jj,1))+p(jj,3)./(p(jj,2).^2-freq(ii)^2+2*1i*p(jj,4)*p(jj,2).*freq(ii))+p(jj,5))))^2+((imag(H1(ii,5))
        -imag((freq(ii)^2*p(jj,1))+p(jj,3)./(p(jj,2).^2-freq(ii)^2+2*1i*p(jj,4)*p(jj,2).*freq(ii))+p(jj,5))))^2;

Error in CODE_allmeshes_TEST (line 728)
        obj_1 = @(p) obj_1(p) + ((real(H1(ii,5))
        -real((freq(ii)^2*p(jj,1))+p(jj,3)./(p(jj,2).^2-freq(ii)^2+2*1i*p(jj,4)*p(jj,2).*freq(ii))+p(jj,5))))^2+((imag(H1(ii,5))
        -imag((freq(ii)^2*p(jj,1))+p(jj,3)./(p(jj,2).^2-freq(ii)^2+2*1i*p(jj,4)*p(jj,2).*freq(ii))+p(jj,5))))^2;

Error in CODE_allmeshes_TEST (line 728)
        obj_1 = @(p) obj_1(p) + ((real(H1(ii,5))
        -real((freq(ii)^2*p(jj,1))+p(jj,3)./(p(jj,2).^2-freq(ii)^2+2*1i*p(jj,4)*p(jj,2).*freq(ii))+p(jj,5))))^2+((imag(H1(ii,5))
        -imag((freq(ii)^2*p(jj,1))+p(jj,3)./(p(jj,2).^2-freq(ii)^2+2*1i*p(jj,4)*p(jj,2).*freq(ii))+p(jj,5))))^2;

Error in CODE_allmeshes_TEST (line 728)
        obj_1 = @(p) obj_1(p) + ((real(H1(ii,5))
        -real((freq(ii)^2*p(jj,1))+p(jj,3)./(p(jj,2).^2-freq(ii)^2+2*1i*p(jj,4)*p(jj,2).*freq(ii))+p(jj,5))))^2+((imag(H1(ii,5))
        -imag((freq(ii)^2*p(jj,1))+p(jj,3)./(p(jj,2).^2-freq(ii)^2+2*1i*p(jj,4)*p(jj,2).*freq(ii))+p(jj,5))))^2;

Error in fmincon (line 552)
      initVals.f = feval(funfcn{3},X,varargin{:});

Error in CODE_allmeshes_TEST (line 749)
[p_opt,fval]=fmincon(obj_1,p0_1(jj,:),a,b,Aeq,beq,lb(jj,:),ub(jj,:),[],options); % p_opt cloumns:   (M, RESONANCE FREQ, A, DAMPING
RATIOS, K) for acc 5.

Caused by:
    Failure in initial objective function evaluation. FMINCON cannot continue.

[FEDERICO ZUCCARI]

I found the error: since I am addressing the index jj in the parameters p in my function obj_1 I have to provide the full matrix p0_1 and not just the line p0_1(jj,:).

Now it works.

Thanks a lot for your help!