DOes the trace determine the Hamiltonian

tpm · Mar 27, 2007

if we define Z as:

[tex] Z(s)=Tr[exp(-sH)] [/tex]

my 2 questions are..

a) is the trace unique and define the Hamiltonian completely? i mean if

we have 2 Hamiltonians H and K then [tex] Tr[exp(-sH)]\ne Tr[exp(-sK) [/tex]

and if we use the 'Semiclassical approach' then [tex] Z(s)=Tr[exp(-sH)]\sim As^{-1/2}\int_{-\infty}^{\infty}dx exp(-sV(x)) [/tex]

then given Z(s) we could calculate approximately V(x) by solving an integral equation.

StatMechGuy · Mar 27, 2007

1 + 2 + 3 = 6
2 + 2 + 2 = 6

Both of these could be the diagonal elements of a hamiltonian. Clearly the trace alone is not sufficient to categorize the entire hamiltonian. You get no information about the off-diagonal elements, too, which is relevant since you have not necessarily taken the trace in a basis which diagonalizes the hamiltonian.

vanesch · Mar 27, 2007

As Statmechguy points out, the trace isn't sufficient to know the hamiltonian. However, the function Z(s), being a function of s, will allow you (by derivatives for instance) to find all individual diagonal elements of H.

tpm · Mar 28, 2007

Well..since the trace is just the sum of 'Eigenvalues' of a Hamiltonian, considering every Hamiltonian has an infinite number of 'energies'.. perhaps the trace is unique (in this case since Statsmechguay has only pointed a finite-dimensional case), could it Happen that both H and K had the same energies but different potential V(x) ??

Log in or Sign up

DOes the trace determine the Hamiltonian

tpm

StatMechGuy

vanesch

tpm

Does every observable commute with the hamiltonian? (Replies: 12)

Does the time dependent hamiltonian have stationary states? (Replies: 5)

When does a wavefunction inherit the symmetries of the hamiltonian? (Replies: 3)

Does hamiltonian/energy eigenstate always exist? (Replies: 4)

Why does "tracing out" work (Replies: 2)