QuantumPhysics.dvi
wang
(Wang)
#1
operatorT is simpler thanH, it will be advantageous to diagonalize it first. Since T is
unitary, its eigenvalues are pure phaseseiφm. We shall denote the corresponding eigenstates
by|km;T〉; we include the labelTinside the state ket, because these kets will form a basis
in whichTis diagonal and this basis is different from the basis|n〉. The labelTis included
to make this distinction clear. Thus we have,
T|km;T〉=e−iφm|km;T〉 (5.8)
FromTN=I, we haveNφm≡0 (mod 2π). The eigenvalues are all distinct, and we have
φm= 2πm/N m= 0, 1 ,···,N− 1 (5.9)
This leads toNorthogonal eigenstates (sinceTis unitary), and gives an explicit construction
of all the normalized eigenstates ofT,
|km;T〉=
1
√
N
∑N
n=1
e+inφm|n〉 m= 0, 1 ,···,N− 1 (5.10)
These states are usually referred to asBloch states.
5.1.2 Position and translation operator algebra
We may define aposition operatorXon this finite lattice by introducing a physical lattice
spacingainto the problem. The origin of position is arbitrary, which allows us to leave an
arbitrary additive constantx 0 in the eigenvalues,
X|n〉= (an+x 0 )|n〉 (5.11)
Clearly, this operator is self-adjoint and indeed corresponds to anobservable. The physical
length of the lattice is then L= aN. The translation operatorT now has the physical
interpretation of shifting the system forward by one lattice spacinga. Since the action ofT
andXis known on all states, we can compute the action ofTonX,
T†XT|n〉 = T†X|n+ 1〉=T†(a(n+ 1) +x 0 )|n+ 1〉
= (a(n+ 1) +x 0 )|n〉= (X+a)|n〉 (5.12)
This equation being valid on all states, we conclude that
T†XT=X+aI (5.13)
which makes full intuitive sense: the action of the translation generator indeed shifts the
position eigenvalues all bya. Of course, what is not well-accounted for here is the fact that
Xis really a periodic variable, defined only modaN. To account for this periodicity, a more
appropriate operator would bee^2 πiX/(Na).