18 Z. Nussinov / Nuclear Physics B 953 (2020) 114948
θ(t) becomes the polar angle of the center of mass of the ring (i.e., θ(t) is the angle between
(i) a vector connecting the origin to the center of the center of the ring (see, e.g., Figs. 4 and
5) and (ii) a vector along the positive S
z
tot
axis). The expectation value S
z
tot
is that of S
z
tot
in
the time evolved state (classically, it is the average of S
z
tot
around the full ring (0 ≤ ϕ
< 2π)
at time t), i.e., S
z
tot
(t) = S
z
tot
cos θ(t). The possible values of S
z
(ϕ
) appear symmetrically twice
in the interval 0 ≤ ϕ
< 2π. We may thus consider only 0 ≤ ϕ
<π. By the normalization of
the probability distribution for ϕ
and the corresponding probability distribution for the energy
density, P(
)d
=
dϕ
π
. Thus,
P(
) =
1
π
dϕ
d
=
N
π
B
z
∂S
z
tot
(ϕ
)
∂ϕ
. (16)
Combining Eq. (9)(which may derived from a geometric analysis of Fig. 5 as we next ex-
plain) with Eq. (16) then provides Eq. (13). We may indeed readily calculate the spread σ
S
z
tot
of S
z
tot
values and rationalize the finite standard deviation σ
of Eq. (9) from simple geo-
metric considerations, when 1/S
tot
is set to zero (the semi-classical limit). From geometry,
σ
S
z
tot
=
R
ring
√
2
| sin θ(t)| ≡R
g
| sin θ(t)| where R
ring
= S
tot
¯
h
√
1 −w
2
. Here, R
g
is the radius of gy-
ration of the ring of Fig. 5 (corresponding to θ = π/2) about an axis parallel to the S
z
tot
axis that
passes through the center of mass of this ring. The finite radius of gyration R
g
= 0 implies a
spread of energy densities σ
=
|B
z
sin θ(t)|R
g
N
= 0at general times. This semi-classical result for
σ
coincides with Eq. (9). We will further comment on the |w| = 1 states below and at the end
of Section 6.1.3. We now first briefly comment on another trivial limiting case. When w =0, the
initial state will correspond, in the description of Fig. 4, to the equator. Applying a transverse
field will then lead to a rotation of the equator around the S
y
tot
axis; this so generated ring (an-
other great circle on the sphere) will, generally, display a non-vanishing spread of S
z
tot
/N values
(leading to σ
=0). However, when the initial state has w = 0, such a rotation will not yield any
change in the energy density, d/dt = 0. This trivial limiting case illustrates that, as a matter of
principle, a finite rate of variation of the energy density is not mandatory in order to a finite ha
ve
σ
. As we demonstrate in the current work, the converse statement holds (a finite d/dt implies
a finite σ
).
Although the Hamiltonian of Eq. (6)is extremely general as are its eigenstates of high to-
tal spin S
tot
= O(N ) (e.g., states of large total spin in typical low temperature ferromagnets),
characteristic equilibrium states of this Hamiltonian will correspond to a special subset having
|w| = 1 (that is, the total spin will be polarized along the externally applied field direction). As
we discussed earlier, such equilibrium states will thus emulate product states (in which all indi-
vidual spins assume the same polarization). Thus, as wa
s indeed evident in Eqs. (9), (15), when
w =±1, the broadening σ
=0. In a related vein, the fully polarized state – a coherent spin state
on a sphere of radius S
tot
– is rotated “en block” without any other change of the wavefunction
under the action of a transverse field. To see the effect for our exactly solvable system, we have
to go away from the limit |w| = 1. Away from this limit, the state of the system evolves non-
trivially. In the parlance of Section 4, when e
volving under the transverse field Hamiltonian of
Eq. (8), the |w| =1spin state is not merely “translated” (rotated on a sphere of radius S
tot
) with
no other accompanying changes. Appendix J discusses a gedanken experiment in which starting
from an equilibrium state, one may apply transverse fields and let the closed system equilibrate
anew so as to generate a state |ψ
0
Spin
of total spin S
tot
=O(N) with w =±1.