editors: nobody
Coefficient extraction also for the infinite rotor sum is justified in
the ordered complex tube 0 < Re Y_1 < ... < Re Y_{n_p} < W. At fixed
finite n_p intermediate labels differ from the through labels by bounded
steps, so their energies differ from E_{qj} by O_{n_p}(1 + |q| + d_j),
and on each compact subset of this tube the endpoint terms are bounded
by a fixed polynomial in the labels times
exp( - W E_{qj} + C (1 + |q| + d_j) ).
The quadratic energy gives normal convergence of the analytic exponential
sums, including after any finite imaginary translation. Modulo
even-integer shifts their exponent vectors have only finitely many types,
and the commuting translations Y_a -> Y_a + i pi multiply an exponential
of rate gamma_a by exp(i pi gamma_a), these multipliers distinguishing the
types, so polynomial interpolation in the translations projects onto each
type. In the type containing the required plane divide by
exp(i sum_a k_a Y_a); the resulting locally convergent Laurent series in
w_a = exp(2 Y_a), on 1 < |w_1| < ... < |w_{n_p}| < exp(2W), has the
desired plane coefficient as its constant coefficient, and uniqueness of
Laurent coefficients proves that no other term of the infinite sum
changes the extraction.
Finally
i ( z - 2 eta ) = ( - i q - k ) - 2 i S_j . sigma = L_j(-i q - k),
so the ordered product in each matrix index is the monodromy whose
twisted trace has eigenvalue t_j(-i q), the two indices giving its
square. Multiplying by lambda^{n_p} and summing the seam weights yields
the transfer formula. At zero twist this is the top transfer eigenvalue
by the ground branch and the ground-state lemma, and exceptional ring
lengths follow by continuity.
The nontrivial rotor sectors must now be estimated uniformly in the band
of the parity proposition. The (q,j) = (0,0) term is
Z_{n_p}(zeta, chi) = prod_{t=1}^{n_p} 2 lambda / ( 4 + k_t(zeta, chi)^2 ),
and the issue is to compare every other term with this product without
accumulating a fixed loss for each of the O(W b^2) particles.
Lemma (uniform transfer asymptotics). Fix A, D > 0 and positive
constants c_-, c_+. Suppose b -> infinity,
c_- exp(pi b) b^{-1/2} (log b)^D <= W
<= c_+ exp(pi b) b^{-1/2} (log b)^D,
| n_p/W - b^2 | <= A b.
There is c' > 0, depending only on these fixed constants, such that
at zero twist the sum over (q,j) != (0,0), divided by Z_{n_p}(0,0),
is O(exp(-c' W)), uniformly over these particle numbers. If u_r are
the real centres of the centre equations at W' = W, then
log ||T_{n_p}|| = sum_{r=1}^{M} p(u_r) + l f(0) + o(W^{-1}),
with the error uniform over the same band.
Set the spin twist to zero, initially retaining continuation from
arbitrarily small nonsingular spin twists where needed. In the auxiliary
trace evaluated at -i q every shifted argument of Q_0 has an odd integral
imaginary part relative to a real formal root: since 2m = 2j = q mod 2,
q + 2m +- 1 is odd, and the same is true of q +- (2j + 1), so these
factors are uniformly separated from zero.
At the formal pair spacing k_{r+-} = u_r +- i the contribution of this
pair to F and its accompanying Q_0 factors in each summand cancels to
(-i q - u_r)^2 + d^2 with d = 2j + 1, independently of m. In each of the
d^2 summands obtained by squaring t_j the absolute contribution of this
pair relative to its (0,0) contribution is therefore
(u_r^2 + 1)(u_r^2 + 9)
/ | ( (i u_r - q)^2 - (d + 2)^2 )
( (i u_r - q)^2 - (d - 2)^2 ) | <= 1.
To see the inequality factor the denominator into four linear factors,
whose absolute values are sqrt(u_r^2 + w^2) with widths
|q +- (d + 2)| and |q +- (d - 2)|; all are positive odd integers and at
least two are at least three, so their product is at least
(u_r^2 + 1)(u_r^2 + 9).
This comparison remains uniform for the true roots. The linear factors
just used, and the individual factors before cancellation for the two
pair constituents, stay a fixed distance from zero; the root estimates
make the corrections smaller than every power of W^{-1}; and there are
only polynomially many factors in W in the particle band, so the total
multiplicative change is bounded and in fact differs from one by less
than every prescribed power of W^{-1}. This is where an estimate per
particle would have been insufficient without the sign-free bound of one
at the formal spacing.
If l = 1 the denominators contributed by the single momentum can vanish
at zero common twist before cancellation, and they are controlled on one
complex zeta circle instead. Choose a sufficiently large fixed C and
put |zeta| = W^{-C}. The common twist translates every momentum and
auxiliary root by zeta/W; at zeta = 0 the single momentum is zero up to
an error smaller than every power of W^{-1}, so on this circle every
possibly vanishing single factor has modulus bounded below by a fixed
multiple of W^{-C-1}, and all single-root contributions are bounded by a
fixed power of 1 + W + d + |q|. The change in all pair factors is bounded
as well, since each argument moves by W^{-C-1} and the number of factors
is polynomial in W, and choosing C once sufficiently large controls
their product.
The maximum-modulus principle is applied on this circle; this requires
analyticity on the whole interior, with no appeal to a lower bound for
the gap between different total-momentum sectors. Let psi_0 be the
normalized untwisted ground vector, with total momentum zero, and define
the change of gauge
(U_zeta psi)(X) = exp( (i zeta / W) sum_{t=1}^{n_p} X_t ) psi(X).
After this change of gauge the Hamiltonian on periodic functions is
H_2 + 2 zeta P / W + n_p zeta^2 / W^2, with P = -i sum_t d/dX_t, and in
the P = 0 subspace its ground line is therefore the same simple line for
every zeta. The twisted transfer T_{n_p}(zeta, 0) has an analytic rotor
expansion, and so does the fixed-gauge scalar
< psi_0, U_zeta^{-1} T_{n_p}(zeta, 0) U_zeta psi_0 >.
Near zero this scalar equals Lambda_{n_p}(zeta, 0) by commutation, so it
continues Lambda_{n_p} across any crossings with other momentum sectors.
The coefficient formula may first be used with arbitrarily small spin
twists and then continued to zero spin twist, equivalently taking the
removable analytic limits in the transfer formula.
On the zeta circle the ratio of the nontrivial-sector sum to Z_{n_p} is
therefore bounded by
C sum_{(q,j) != (0,0)} d^2 (1 + W + d + |q|)^K
exp( - W E_{qj} + W^{-C} |q| ) = O( exp(-c' W) )
for fixed K and some c' > 0. Every nontrivial admissible sector has
E_{qj} >= 2 and the energies grow quadratically in q, d, so the
polynomial factors and the number of summands are absorbed by the
exponential. The ratio is analytic on the enclosed disk, since Z_{n_p}
is analytic and nonzero there and the whole nontrivial-sector sum is
Lambda_{n_p} - Z_{n_p}, so the maximum-modulus principle gives the same
bound at zero twist; for l = 0 the separated pair factors already give
the bound directly and the same circle argument also applies.
It remains to take the logarithm of the leading product. Since
2 lambda = b^2 a formal pair contributes
b^4 / ( (4 + (u_r + i)^2)(4 + (u_r - i)^2) )
= b^4 / ( (u_r^2 + 1)(u_r^2 + 9) ) = exp( p(u_r) ),
and the possible zero single contributes b^2/4 = exp(f(0)). The
cumulative error from replacing the true roots by the centres is smaller
than every power of W^{-1}, and the nontrivial-sector error is
exponentially smaller, so the positive zero-twist eigenvalue has the
stated logarithm.
The transfer asymptotics have now reduced the problem to a sum over real
pair centres. The dressed energy of the integral equations, with
endpoint B = B_*, expresses this sum as a common bulk term, an explicit
single-particle cost, and a quadrature error.