Henrik Sönnerlind
                                                                                                                                                    COMSOL Employee
                                                         
                            
                                                                                                                                                
                         
                                                
    
        Please login with a confirmed email address before reporting spam
     
    
 
                                                Posted:
                            
                                7 years ago                            
                            
                                6 août 2018, 07:47 UTC−4                            
                        
                        
                                                    Hi Sarah,
This is used if your eigenvalue problem is nonlinear in the eigenvalue itself. Think about a simple standared eigenvalue problem like

If the matrix A itself depends on the eigenvalue, that is

then the problem is linearized using

where  is the value you supply as Value of eigenvalue linearization point.
 is the value you supply as Value of eigenvalue linearization point.
This means that you can expect accurate eigenvalues only in the vicinity of  , and may have to do several analyses with different values of the the linearization point.
, and may have to do several analyses with different values of the the linearization point.
Regards,
Henrik
    -------------------
    Henrik Sönnerlind
COMSOL                                                
 
                                                
                            Hi Sarah,  
This is used if your eigenvalue problem is nonlinear in the eigenvalue itself. Think about a simple standared eigenvalue problem like
 ( \mathbf A-\lambda \mathbf I) \mathbf x = 0
If the matrix **A** itself depends on the eigenvalue, that is 
 ( \mathbf A(\lambda)-\lambda \mathbf I) \mathbf x = 0
then the problem is linearized using 
 ( \mathbf A(\lambda_0)-\lambda \mathbf I) \mathbf x = 0
where  \lambda_0 is the value you supply as _Value of eigenvalue linearization point_.
This means that you can expect accurate eigenvalues only in the vicinity of  \lambda_0, and may have to do several analyses with different values of the the linearization point.
Regards,  
Henrik