Unable to perform assignment because the size of the left side is 1-by-1 and the size of the right side is 0-by-1

I am stack to this error anyone can help me please:

"Unable to perform assignment because the size of the left side is 1-by-1 and the size of the right side is 0-by-1.

Error in Achievement (line 89)

[fi(j,i),W(j,i),f(j,i),F(j,i),beta(j,i),CL(j,i),CD(j,i),a1(j,i),Re_new(j,i),chord_dim(j,i),sigma(j,i)]

= ..."

The problems I think is in the definition of this function:

function [fi,W,f,F,beta,CL,CD,a1,Re_new,chord_dim,sigma] = iteration(a,a1,V0,omega,r,r_dim,Re,rho,mu,Nb,R,root,primary,tip,profile_rt,profile_p,profile_t,beta_c_opt)

fi=atan(((1-a)*V0)./((1+a1).*omega.*r_dim));
W=((1-a).*V0)./sin(fi);
f=Nb*(R-r_dim)./(2*r_dim.*sin(fi));
F=(2*acos(exp(-f))./pi);
    
    if r<=root
            [beta,CL,CD,~]=interpReynolds(Re,profile_rt);
            opt=find(beta==fi-beta_c_opt);
            beta=deg2rad(beta(opt)); CL=CL(opt); CD=CD(opt);
    elseif r>=root && r<=primary
            [beta_rt,CL_rt,CD_rt,~]=interpReynolds(Re,profile_rt);
            opt_rt=find(beta_rt==fi-beta_c_opt); beta_rt=beta_rt(opt_rt); CL_rt=CL_rt(opt_rt); CD_rt=CD_rt(opt_rt);
            [beta_p,CL_p,CD_p,~]=interpReynolds(Re,profile_p);
            opt_p=find(beta_p==fi-beta_c_opt); beta_p=beta_p(opt_p); CL_p=CL_p(opt_p); CD_p=CD_p(opt_p);
            [beta,~,CL,CD]=interpRadius(root,primary,beta_rt,beta_p,CL_rt,CL_p,CD_rt,CD_p,r,fi);
    elseif r>=primary && r<=tip
            [beta_p,CL_p,CD_p,~]=interpReynolds(Re,profile_p);
             opt_p=find(beta_p==fi-beta_c_opt); beta_p=beta_p(opt_p); CL_p=CL_p(opt_p); CD_p=CD_p(opt_p);
            [beta_t,CL_t,CD_t,~]=interpReynolds(Re,profile_t);
            opt_t=find(beta_t==fi-beta_c_opt); beta_t=beta_t(opt_t); CL_t=CL_t(opt_t); CD_t=CD_t(opt_t);
            [beta,~,CL,CD]=interpRadius(primary,tip,beta_p,beta_t,CL_p,CL_t,CD_p,CD_t,r,fi);
    elseif r>=tip
            [beta,CL,CD,~]=interpReynolds(Re,profile_t);
            opt=find(beta==fi-beta_c_opt);
            beta=deg2rad(beta(opt)); CL=CL(opt); CD=CD(opt);
    end

sigma=4*V0*a*F/(((CL/tan(fi))+CD)*W);
chord_dim=2*pi*r_dim*sigma/Nb;
a1=sigma*W*(CL-CD/tan(fi))/(4*omega*r_dim*F);        
Re_new=rho*chord_dim*W/mu;
end

This function is recalled by this part of the script:

i=r_opt;
    z(i)=0;
    
    while abs(Re_guess(j,i)-Re(j,i))>1e4
                z(i)=z(i)+1;
                [fi(j,i),W(j,i),f(j,i),F(j,i),beta(j,i),beta_c(j,i),CL(j,i),CD(j,i),a1(j,i),Re_new(j,i),chord_dim(j,i),sigma(j,i)] = ...
                    iteration2(a(j,i),a1(j,i),V0,omega(j),r(i),r_dim(i),Re(j,i),rho,mu,Nb,R,root,primary,tip,...
                    profile_rt,profile_p,profile_t);
                Re_guess(j,i)=Re(j,i);
                Re(j,i)=RF(j)*Re_new(j,i)+(1-RF(j))*Re(j,i);
    end
    
    beta_opt=beta(j,i);
    beta_c_opt=beta_c(j,i)*ones(1,length(r));
    
   
    for i=1:length(r)
            i
            z(i)=0;
            while abs(Re_guess(j,i)-Re(j,i))>1e4
                z(i)=z(i)+1;
                [fi(j,i),W(j,i),f(j,i),F(j,i),beta(j,i),CL(j,i),CD(j,i),a1(j,i),Re_new(j,i),chord_dim(j,i),sigma(j,i)] = ...
                    iteration(a(j,i),a1(j,i),V0,omega(j),r(i),r_dim(i),Re(j,i),rho,mu,Nb,R,root,primary,tip,...
                    profile_rt,profile_p,profile_t,beta_c_opt(i));
                Re_guess(j,i)=Re(j,i);
                Re(j,i)=RF(j)*Re_new(j,i)+(1-RF(j))*Re(j,i);
            end
    end
    
    chord(j,:)=chord_dim(j,:)/R;

Answer 1

[Walter Roberson]

You are using == to compare floating point numbers that are being computed different ways. Most of the time, because of floating point round-off, the values you want to compare will not be bit-for-bit identical, and == always only compares bit-for-bit identical when the same datatype is being compared (exception: -0 and 0 compare equal, and all the different kinds of nans are treated the same.)

You should use ismembertol() or check that abs() of the difference between the numbers is less than some tolerance. And even then you should be writing the code in a way that is robust for the case where there simply is no reasonable match.

You can reliably use == to compare floating point numbers that are integers in the range -2^53 to +2^53, but only if you can guarantee that they are integers. Not, for example, 29/7*7 which is not an integer .

Beyond that... things start to get a bit strange, and you should be careful about using ==

The major exception to this is that if you extract a value from an array by indexing or max() or min(), then you can be certain that it is a bit-for-bit identical copy of what is in the array, and you can == the extracted value to the array. It is safe, for example, to use

x = randi([3 9], 1, 5)

x = 1×5

3 7 3 5 7

minx = min(x)

minx = 3

x == minx

ans = 1×5 logical array
   1   0   1   0   0

[Emilio Pulli]

Thank you so much Walter for the answer.

Does the function ismembertol() returns the index of the value of the vector which is similar to the one I am looking for according to the defined tolerance?

Because I need to find the index of the desired value contained in the vector beta in order to indexing also the vector CL and CD.

[Walter Roberson]

look at the second output of ismembertol for the index

[Emilio Pulli]

How do I have to write to obtain the index?

For instance, I wrote this but I do not obtain the index of the beta value equal (according to the tolerance 0.1) to fi-beta_c_opt.

I obtain, ideed, this error

" Array indices must be positive integers or logical values.

Error in iteration (line 11)

beta=deg2rad(beta(opt)); CL=CL(opt); CD=CD(opt);

Error in Achievement (line 90)

iteration(a(j,i),a1(j,i),V0,omega(j),r(i),r_dim(i),Re(j,i),rho,mu,Nb,R,root,primary,tip,..."

if r<=root
            [beta,CL,CD,~]=interpReynolds(Re,profile_rt);
            [~,opt]=ismembertol(beta,fi-beta_c_opt,0.1)
            beta=deg2rad(beta(opt)); CL=CL(opt); CD=CD(opt);

[Walter Roberson]

you forgot to check the first output which tells you whether a match exists at all.

idx = find(A==B)

translates to

[found, idx] = ismembertol(A, B)
idx = idx(found)

until you get into non-scalar A

[Emilio Pulli]

if a match does not exist, and I want to use the nearest value, what do I have to write?

[Walter Roberson]

dists = abs(beta(:) - reshape(fi-beta_c_opt),1,[]));
opt = min(dists, 2);

opt will now be a vector with the same number of elements that beta has, and each entry will be the linear index into fi-beta_c_opt of the closest value (even if the value is not very close on an absolute scale.)