PDA

View Full Version : Decomposition of linear operator


dingo_d
Oct11-10, 07:00 AM
1. The problem statement, all variables and given/known data

Show that any linear operator \hat{O} can be decomposed as \hat{O}=\hat{O}'+i\hat{O}'', where \hat{O}' and \hat{O}'' are Hermitian operators.

2. Relevant equations

Operator is Hermitian if:

T=T^{\dagger}


3. The attempt at a solution

I don't know where to start :\ Should I try to see for some arbitrary vector |\psi\rangle, that I can write it in some basis, and see what it would do to write eigenvalue equation with those operators?

arkajad
Oct11-10, 07:28 AM
Play with \hat{O}^\dagger+\hat{O} and i(\hat{O}^\dagger-\hat{O}).

dingo_d
Oct11-10, 07:46 AM
Hmmm.

So \hat{O}^{\dagger}+\hat{O}=\hat{O}'^{\dagger}-i\hat{O}''^{\dagger}+\hat{O}'+i\hat{O}'' and since those with ' and '' are hermitian it follows:

\hat{O}^{\dagger}+\hat{O}=\hat{O}'-i\hat{O}''+\hat{O}'+i\hat{O}''=2\hat{O}'

Am I on the right track?

arkajad
Oct11-10, 08:08 AM
Not too bad ....

dingo_d
Oct11-10, 08:15 AM
XD Too bad? XD

For the i(\hat{O}^\dagger-\hat{O}) part I got 2\hat{O}''

So I should somehow prove linear superposition or?

arkajad
Oct11-10, 08:38 AM
Well, so what is your guess for O' and O''? Guess what they should be, then check carefully if your guess has all the required properties.

This is how we often solve problems: we play, we guess, and the we check carefully, in all detail, if our guess really constitutes a solution of the original problem. The way towards a solution does not have to be logical. What needs to be logical is the verification.