What is the easiest and accurate way in Matlab to find the eigenvalues and eigenvectors of a problem that do have some zero diagonal elements?

How to find the eigenvalues and eigenvectors of a problem that have some zero diagonal elements which dont have the usual form of the standard eigenvalue problem?

clc
clear
K=load('Ks.mat').K;
M=load('Ms.mat').M;
%Eig Method
[Qi,D] = eig(K,M);
omega=sort(real(sqrt(real(diag(D)))*(0.7588455916e-5/0.15176911835e-5)));

omega=

2.40257944339103

2.40257944339103

3.70603843883719

4.39253729249290

4.39253729249290

.

.

.

%LV Method
m=2000;
n=800;
M11=M(1:m,1:m);
M12=M(1:m,m+1:m+n);
M21=M(m+1:m+n,1:m);
M22=M(m+1:m+n,m+1:m+n);
K11=K(1:m,1:m);
K12=K(1:m,m+1:m+n);
K21=K(m+1:m+n,1:m);
K22=K(m+1:m+n,m+1:m+n);
Mt=inv(sqrtm(M11));
LV=1000000;
A11=Mt*K11*Mt;
A12 = LV*Mt*K12;
A21 = LV*K21*Mt;
A22 = K22;
AA =[A11, A12;A21, A22];
[Qi, B]=eig((AA));
C=sqrt(diag(B));
D=C*(0.7588455916e-5/0.15176911835e-5);
Omega=sort(real(D))

1.54228989861890

2.33587868130244

2.98460674608965

3.31683033853761

The results obtained from LV method are much close to the results of the problem which are as below:

1.5477

2.2752

2.9762

3.3202

Why eig is unable to give an accurate result??

How I can get these results through eig command or other way you recommend?

[Matt J]

Have you tried eig with the 'qz' input switch?

[Mehdi]

[Qi,D] = eig(K,M,'qz');

omega=sort(real(sqrt(real(diag(D)))*(0.7588455916e-5/0.15176911835e-5)));

2.40257944339103

2.40257944339103

3.70603843883719

4.39253729249290

4.39253729249290

4.43682500952759

.

.

.

Nothing changed!