Positive eigenvector coefficients with matlab

19 visualizaciones (últimos 30 días)
crocusy
crocusy el 12 de Nov. de 2017
Comentada: crocusy el 13 de Nov. de 2017
Hello, I need positive eigenvector coefficients. I am new in Matlab. I've thought to place eig function into fmincon function with the constraint e>=0. However, i couldn't manage it. How can i obtain positive eigenvector coefficients with matlab?
  2 comentarios
Kaushik Lakshminarasimhan
Kaushik Lakshminarasimhan el 12 de Nov. de 2017
What do you mean by positive eigenvector coefficients? Do you just mean positive coefficients? Can you explain what you're trying to fit?
crocusy
crocusy el 12 de Nov. de 2017
Yes, i mean positive coefficients. I need eigenvector coefficients all positive. I've tried to write something as below.
if true
f=@(e,l,g) eig(S);
e0=zeros(N,N);
e1=fmincon(f,e0,[],[],[],[],lb,[],[],opts);
end
There must be lb for eigenvector coefficients, i.e. e>=0 but I couldn't put it anywhere, since there are three outputs. When i write lb, it doesn't directly affect the coefficients. I know it is not rational, however i wonder if there is a way such that to find positive coefficients.

Iniciar sesión para comentar.

Respuesta aceptada

Walter Roberson
Walter Roberson el 12 de Nov. de 2017
Any scalar constant times an eigenvector remains an eigenvector. Therefore if you have an eigenvector that contains all negative values, multiply it by a negative scalar.
If you have an eigenvector that contains a mix of negative and positive values, you need to reject it (for your purpose) and move on to the next eigenvector. You would not do this with fmincon or the like: you would just look through the list of eigenvectors that were output to find one for which the sign() was the same for all of the elements.
Question: do you consider 0 to be positive for this purpose?
  1 comentario
crocusy
crocusy el 13 de Nov. de 2017
Yes, 0 is considered as positive. Indeed, i need nonnegative cofficients. Thank you for your answer. It helped me a lot.

Iniciar sesión para comentar.

Más respuestas (0)

Categorías

Más información sobre Linear Algebra en Help Center y File Exchange.

Etiquetas

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by