full identification with sign restrictions

[KOBE24]

Dear Tom,

first of all I want to apologize for not answering you earlier. Some problems took me out from econometrics during last months.
I hope to be a little bit more regular now.

Thank you so much for your help with Fry and Pagan criterion, and for your precious guide in programming with RATS.
I'll have a look and try to implement to my case.

In the meanwhile, I am reading a little bit about sign restrictions and I found an (apparent) alternative way to identify shocks: the one in Canova and De Nicolò (JME, 2002) or Peersman (JAE 2005).

This uses a rotation matrix with eigenvalues-eigenvectors decomposition, if I understand.
I am particularly interested in Peersman case, because I want to fully identify a system of 3 shocks: supply, fiscal policy and monetary policy.
And I have just 3 variables. I already did it with Uhlig's code, but I figure there is some difference, especially in terms of variance decomposition.
Could you please clarify me?
Would it be sufficient switch from Choleski initial decomposition to the eigenvalues-eigenvectors, and then taking draws?

Summing up, is this an alternative and more efficient way to fully identify a system k*k (k variables, k shocks) and having variance decomposition?

Thanks in advance, and please accept my apologies.