Why ignore the potential term in the quantum Hamiltonian?

[andrewkirk]

I am reading about the recovery of some classical rules from quantum mechanics.

My text (Shankar) considers a Hamiltonian operator in a one-dimensional space
H = P^2 / 2m + V(X)
where P and X are the momentum and position operators respectively.

It then asserts that [X,H] = [X,P^2/2m]

That is, it has discarded the potential term of the Hamiltonian without comment or explanation.

How is that justified? I would have thought that if, as indicated, V is a function of X, the hamiltonian operator should be expressed in terms of that function.
For example, if V is a gravitational potential V(X) = -k/X, I would expect the above commutator to be

[X,H] = [X,P^2/2m-k/X] = [X,P] - k[X,1/X]

Why does Shankar discard the second term?
If one didn't discard it, what would [X,1/X] mean? Is there any way to handle the reciprocal of an operator?

Thanks for any help.

 

[Matterwave]

The commutator of an operator with any function of itself is 0. That's why it's discarded. [x,f(x)]=0 for any f.

 

[tom.stoer]

You can see that by letting act V(x) and x act on a test function:

[x, V(x)] f(x) = xVf - Vxf = 0

(simply b/c multiplication of functions is commutative)

This does no longer hold for p which acts as a derivative; using the stanbdard rules for derivatives w.r.t. x one finds

[p, V(x)] f(x) = pVf - Vpf = (pV)f + V(pf) - V (pf) ~ V'f + Vf' - Vf' = V'f

 

[cmos]

If one didn't discard it, what would [X,1/X] mean? Is there any way to handle the reciprocal of an operator?

Think of it in terms of linear algebra. Parameter x -> operator X; parameter 1/x -> operator X^-1. Then you have XX^-1=X^-1X=I, (the identity operator), because inverses commute (otherwise you wouldn't get I regardless).