why does it give this result???

the code works well up to the moment I introduce the while loop. It says: "Warning: Matrix is singular, close to singular or badly scaled. Results may be inaccurate.

RCOND = NaN.

> In progettofinale (line 84)"

then, it always gives as a result BFinale=0 and consequently RAGGIO=inf

why??

nmax=100;

COEFFICIENTI=zeros(26,26);
BUCKLING=zeros(1,nmax);
K=zeros(1,nmax);
FLUSSI=zeros(26,nmax);
FLUSSIN=zeros(26,nmax);

B1=6E-5;

FLUSSO0=zeros(26,1);
for ii=1:26
    FLUSSO0(ii)=1/F(ii);
end

for ii=1:26
    for jj=1:26
        for ii=jj
            COEFFICIENTI(ii,jj)=D(ii).*B1+REMOVAL(ii);
        end
    end
end

for ii=2:26
        COEFFICIENTI(ii,(ii-1))=-ELREM(ii);
end


for ii=1:26
    for jj=1:26
        COEFFICIENTI(ii,jj)=COEFFICIENTI(ii,jj)-INELASTIC(ii,jj);
    end
end

FLUSSI(:,1)=COEFFICIENTI\X;
K1=F'.*FLUSSI(:,1);
FLUSSIN(:,1)=FLUSSI(:,1)\K1;

if K1>1
    B2=B1+(0.5*B1);
else if K1<1
        B2=B1-(0.5*B1);
    end
end

for ii=1:26
    for jj=1:26
        for ii=jj
            COEFFICIENTI(ii,jj)=D(ii).*B2+REMOVAL(ii);
        end
    end
end

FLUSSI(:,2)=COEFFICIENTI\X;
K2=F'.*FLUSSI(:,2);
FLUSSIN(:,2)=FLUSSI(:,2)\K2;


%dalla 3 iterazione in poi si può utilizzare l'interpolazione lineare, servendosi dei
%due precedenti valori di B per calcolare quello successivo:


n=3;

while abs(K2-1)>10^-5 & n<nmax
    
    BUCKLING(n)=BUCKLING(n-1)+(BUCKLING(n-2)-BUCKLING(n-1))*(1-K(n-1))\(K(n-2)-K(n-1));
    
for ff=1:26
    for ll=1:26
        for ff=ll
            COEFFICIENTI(ff,ll)=D(ff).*BUCKLING(n)+REMOVAL(ff);
        end
    end
end

FLUSSI(:,n)=COEFFICIENTI\X;
Kn=F'.*FLUSSI(:,n);
FLUSSIN(:,n)=FLUSSI(:,n)\Kn;

errorenuovo=abs(Kn-1);
iterazioni=n;
n=iterazioni+1;

end


BFinale=BUCKLING(n)
RAGGIO=pi/(sqrt(BFinale))

Answer 1

[Tanmay Das]

Hi,

In the while loop part of the code, mldivide(\) operator is used which has the same precedence as that of matrix multiplication(*). So, (BUCKLING(n-2)-BUCKLING(n-1))*(1-K(n-1)) becomes 0. Also the value of (K(n-2)-K(n-1)) is 0. Hence, it is of the form 0 \ 0 which is returning NaN. That is why there is a warning that "Matrix is singular, close to singular or badly scaled. Results may be inaccurate". If you change the line to:

BUCKLING(n)=BUCKLING(n-1)+(BUCKLING(n-2)-BUCKLING(n-1))*((1-K(n-1))\(K(n-2)-K(n-1)));

then this particular warning issue may get resolved. But still you will get BFinale=0 and RAGGIO=inf as the form of (BUCKLING(n-2)-BUCKLING(n-1))*((1-K(n-1))\(K(n-2)-K(n-1))) is still 1 \ 0 which will always return zero and the value of BUCKLING(n) will never increase for any value of n. Hence, it is recommended to check the code as well as the names of variables once again.