Issue with Matrix Dimensions

Hello All,

I am trying to run this code for some calculation but I seem be getting an disnion mismatch error. I have tried revisiting this code again and aagain but not able to find what I am missing. I am coming to the community to understand what is that I am missing Appreciate your help.

M=44;
N=11;
N1=6;
h=0.1;
Re=100;

for i=1:M
    for j=1:N
        u(i,j)=0;
        v(i,j)=0;
        si(i,j)=0;
        om(i,j)=0;
    end
end

for j=1:N1
    y=(j-1)*h;
    u(1,j)=-1.5+6*y*y;
    si(1,j)=-0.5-1.5*y+2*y*y*y;
    om(1,j)=-12*y;
end
  

for j=N1+1:N
    y=(j-1)*h;
    u(1,j)=-18*y*y+12*y-1.5;
    si(1,j)=-6*y*y*y+6*y*y-1.5*y-0.5;
    om(1,j)=36*y-12;
end

for i=(2:M-1)
    si(i,N)=-0.5;
end

for j=(1:N)
    y=(j-1)*h;
    si(M,j)=-0.25+y*y*y-0.75*y;
    u(M,j)=-0.75+3*y*y;
    om(M,j)=-6*y;
end

for ite=(1:500)
    for i=(2:M-1)
        om(i,1)=-u(i:2)/h;
        om(i,N)=u(i:N-1)/h;
    end
for i=(2:M-1)
    for j=(2:N-1)
        si(i,j)=(si(i-1,j)+si(i+1,j)+si(i,j-1)+si(i,j+1)+om(i,j)*h*h)/4;
    end
end
for i=(2:M-1)
    for j=(2:N-1)
        omx=(om(i+1,j)-om(i-1,j))/(2*h); %this line seems to be having an issue.
        omy=(om(i,j+1)-om(i,j-1))/(2*h);
        con=u(i,j)*omx+v(i,j)*omy;
        om(i,j)=(((om(i-1,j)+om(i+1,j)+om(i,j-1)+om(i,j+1))-Re*con*h*h)/4);
    end
end
end

for i=(1:M)
    for j=(1:N)
        u1(i,j)=1.5-6*y*y+u(i,j);
        v1(i,j)=v(i,j);
    end
end 

Answer 1

[William Rose]

@Nilesh Naik,

The error occurs at lines

for ite=(1:500)
    for i=(2:M-1)
        om(i,1)=-u(i:2)/h; %error here
        om(i,N)=u(i:N-1)/h; %error here
    end
    ...
end

Change the lines to

om(i,1)=-u(i,2)/h;
om(i,N)=u(i,N-1)/h;  % changed colon to comma

That works. Is it doing what you want? I don't know. But there is not an error now.

Answer 2

[Voss]

The error I run into first is on this line:

om(i,N)=u(i:N-1)/h;

When i == 2, you have

om(2,11) = u(2:10)/0.1;

Just like the error message says, the left-hand side is a single scalar element and the right-hand side is a 1-by-9 vector. How to resolve that depends on what your intent is there, which I don't know.

Note that if that problem were resolved, then you'll have the same type of error on the next iteration with the previous line:

om(i,1)=-u(i:2)/h;

When i == 3, this is

om(3,1)=-u(3:2)/h;

because u(3:2) is empty and om(3,1) is a scalar.

Are these colon operations really supposed to be comma-separated lists (i.e., row/column indexing operations)? That is, u(i,N-1) and u(i,2)? Let me see if that runs:

M=44;
N=11;
N1=6;
h=0.1;
Re=100;

for i=1:M
    for j=1:N
        u(i,j)=0;
        v(i,j)=0;
        si(i,j)=0;
        om(i,j)=0;
    end
end

for j=1:N1
    y=(j-1)*h;
    u(1,j)=-1.5+6*y*y;
    si(1,j)=-0.5-1.5*y+2*y*y*y;
    om(1,j)=-12*y;
end
  

for j=N1+1:N
    y=(j-1)*h;
    u(1,j)=-18*y*y+12*y-1.5;
    si(1,j)=-6*y*y*y+6*y*y-1.5*y-0.5;
    om(1,j)=36*y-12;
end

for i=(2:M-1)
    si(i,N)=-0.5;
end

for j=(1:N)
    y=(j-1)*h;
    si(M,j)=-0.25+y*y*y-0.75*y;
    u(M,j)=-0.75+3*y*y;
    om(M,j)=-6*y;
end

for ite=(1:500)
    for i=(2:M-1)
        om(i,1)=-u(i,2)/h;
        om(i,N)=u(i,N-1)/h;
    end
for i=(2:M-1)
    for j=(2:N-1)
        si(i,j)=(si(i-1,j)+si(i+1,j)+si(i,j-1)+si(i,j+1)+om(i,j)*h*h)/4;
    end
end
for i=(2:M-1)
    for j=(2:N-1)
        omx=(om(i+1,j)-om(i-1,j))/(2*h); %this line seems to be having an issue.
        omy=(om(i,j+1)-om(i,j-1))/(2*h);
        con=u(i,j)*omx+v(i,j)*omy;
        om(i,j)=(((om(i-1,j)+om(i+1,j)+om(i,j-1)+om(i,j+1))-Re*con*h*h)/4);
    end
end
end

for i=(1:M)
    for j=(1:N)
        u1(i,j)=1.5-6*y*y+u(i,j);
        v1(i,j)=v(i,j);
    end
end 

Runs without error.