推 robrob:感謝你 01/15 03:59
※ 引述《robrob (慢慢的學習)》之銘言：
: 我現在有個矩陣
: M=sparse(5000,5000);
: ---------
: 補充一下
: A B n nx 都是定值
: M22=(-(diag(-2*A-n+1-nx^2)+diag(B,1)+diag(B,-1))*nx);
: M12=((diag(C/2,2)+diag(-B-n(1:Npt-1)+1-nx^2,1)+diag(B+n(1:Npt-1)-1+nx^2,-1)
: +diag(-C/2,-2))/i/dL);
: M11=((diag(-2*beta*A-n+1-beta*nx^2)+diag(beta*B,1)+diag(beta*B,-1))*nx);
: M21=((diag(beta*C/2,2)+diag(-beta*B-n(1:Npt-1)+1-beta*nx^2,1)
: +diag(beta*B+n(1:Npt-1)-1+beta*nx^2,-1)+diag(-beta*C/2,-2))/i/dL);
: M=[M11 M12;M21 M22];
: ----------
: 現在要求
: M*X=0
: X=?
You meant M*X = lambda*X?
: 然後 我用eigs 可是跑出來 只有6個值???
Did you use eigs(M) or eigs(M, num), num > 6?
For example,
>> A = rand(8);
>> [V D] = eigs(A,8); diag(D)
Warning: For nonsymmetric and complex problems, must have number of eigenvalues
k < n-1.
Using eig(full(A)) instead.
> In eigs>checkInputs at 819
In eigs at 86
ans =
4.1518
-0.6278
0.4448 - 0.3889i
0.4448 + 0.3889i
-0.3430 - 0.3799i
-0.3430 + 0.3799i
0.3186
0.1559
Well, if you really want to find M*X = 0 then just pick those X
determined from eigs() and without zero eigenvalues.
: 順便徵求 有更好的方法求出X嗎
※ 編輯: saltlake 來自: 152.16.233.74 (01/15 00:54)