diff 9506729..6fbd3ac o4_model_exact_mass_v2.txt
diff --git a/o4_model_exact_mass_v2.txt b/o4_model_exact_mass_v2.txt
index 215f25f..c90e18e 100644
--- a/o4_model_exact_mass_v2.txt
+++ b/o4_model_exact_mass_v2.txt
@@ -15,7 +15,7 @@ denotes the partition function that makes mu_beta a probability measure.
 The periodic infinite-volume limit mu_beta exists and is unique as a
 periodic local weak limit; this is the uniqueness statement proved in
 the sharp lattice bounds for this model, obtained from the Fortuin-
-Kasteleyn-Ginibre (FKG) inequalities of Fortuin, Kasteleyn and Ginibre
+Kasteleyn-Ginibre (FKG) ineualities of Fortuin, Kasteleyn and Ginibre
 (1971) together with the Edwards-Sokal coupling (Edwards and Sokal,
 1988).
 
@@ -153,13 +153,13 @@ under nu_J.  A monomial is a product of individual spin components with
 coefficient one; its site list has one entry per factor, repetitions
 included.
 
-Two classical finite-system inequalities are used.  For a zero-field
+Two classical finite-system ineualities are used.  For a zero-field
 ferromagnetic plane rotator with angles theta_x, every pair of integer
 vectors k, l indexed by V satisfies
 
     E cos(k . theta) >= 0,   Cov( cos(k . theta), cos(l . theta) ) >= 0.
 
-These are the plane-rotator inequalities of Ginibre (1970).  They have a
+These are the plane-rotator ineualities of Ginibre (1970).  They have a
 short replica proof: Fourier expansion gives the first from the
 nonnegative coefficients, and for the second one takes independent
 copies theta, theta' and substitutes theta = phi + psi, theta' = phi -
@@ -172,7 +172,7 @@ whose expansion writes twice the covariance as a sum of squares of real
 integrals with nonnegative coefficients.  Absolute convergence holds on
 a finite graph.  Differentiating a cosine mean in a bond coupling shows
 the mean is nondecreasing in that coupling.  The zero-field Ising GKS
-inequalities of Kelly and Sherman (1968) are also used: sign-monomial
+ineualities of Kelly and Sherman (1968) are also used: sign-monomial
 means and their covariances are nonnegative.
 
 At each vertex decompose
@@ -199,7 +199,7 @@ signs:
 For the plane rotator write the components as (eps cos g, eps' sin g),
 0 <= g <= pi/2; conditional on g the signs are independent
 ferromagnetic Ising systems, the same mixed-derivative computation with
-the Ising GKS inequalities proves association of g, and the same-plane
+the Ising GKS ineualities proves association of g, and the same-plane
 monomial means are monotone in g.  For the three-sphere case the density
 of a relative to its product one-site prior is proportional to
 Z_L(a) Z_R(a), where Z_L, Z_R are the two circle partition functions with
@@ -212,7 +212,7 @@ with O_e the bond observables.  All first coupling derivatives of J^L
 are nonpositive and all of J^R nonnegative; a nonzero mixed derivative
 at distinct vertices occurs only at a bond's endpoints and is
 nonnegative.  The bond means and covariances are nonnegative by the
-Ginibre inequalities, so the logarithmic density has nonnegative mixed
+Ginibre ineualities, so the logarithmic density has nonnegative mixed
 derivatives, satisfies the FKG lattice condition, and is associated by
 the Fortuin-Kasteleyn-Ginibre theorem (1971); continuous arrays follow by
 restriction to compact subintervals of (0,pi/2) and passage to the limit.
@@ -222,7 +222,7 @@ cosine-character means, hence nonnegative and decreasing in a; a
 component-four monomial has a nonnegative increasing conditional mean by
 a fixed rotation of the right circle.  Conditional independence gives the
 nonpositive covariance of unlike components, and the like-component case
-combines the conditional Ginibre inequality with the association of a.
+combines the conditional Ginibre ineuality with the association of a.
 This proves the lemma.
 
     Proposition (finite-graph sector comparison).  For the law nu_J let
@@ -305,7 +305,7 @@ and d_k of P_L^* with |c_k| <= d_k after replacement, because each sine or
 cosine supplies two exponentials of modulus 1/2 and the even total number
 of sines makes the original coefficients real and symmetric under
 k -> -k; collecting equal frequencies preserves the domination.  The
-Ginibre inequality then gives
+Ginibre ineuality then gives
 
     |Cov(P_L, Q_L | a)| <= Cov(P_L^*, Q_L^* | a),
 
@@ -334,7 +334,7 @@ B_{P,L}.  Put Y_P = b_{P,R} b_{P,L}, D_P = B_{P,R} - B_{P,L}; the modulus
 bounds give |Y_P(a') - Y_P(a)| <= D_P(a') - D_P(a), so D_P + Y_P and
 D_P - Y_P are increasing, and association applied to these and to
 D_Q +- Y_Q, together with the two matching-sign and two opposite-sign
-inequalities, proves
+ineualities, proves
 
     |Cov(Y_P, Y_Q)| <= Cov(D_P, D_Q).
 
@@ -416,11 +416,11 @@ expectation of the corresponding row-transfer chain.
     at most tau_{beta,l}^2.  There is a real row function F with |F| <= 1,
     odd under reflection of one spin component, such that
 
-        omega_{beta,l}( F * F circ vartheta_n )
+        omega_{beta,l}( F * F circ theta_n )
             >= ( tau_{beta,l} / (1 + tau_{beta,l}) )
                    * tau_{beta,l}^{n},   n >= 0,
 
-    where vartheta_n translates by n rows.
+    where theta_n translates by n rows.
 
 The finite-graph estimate passes to the infinite-time cylinder through
 the row transfer representation.  Component vectors q_x^i Omega are
@@ -455,7 +455,7 @@ visibility is not lost.
     l >= 4, such that bounded row functions F, G of modulus at most one,
     both odd under reflection of component one, satisfy
 
-        | omega_{beta,l}( F * G circ vartheta_n ) |
+        | omega_{beta,l}( F * G circ theta_n ) |
             <= C_beta sum_{x in R_0, y in R_n}
                     omega_{beta,l}( q_x^1 q_y^1 ),   n >= 1,
 
@@ -533,11 +533,11 @@ bare coupling with given N, h is not assumed.
         N log L = pi beta - (1/2) log beta + O_{L,H}(1).
 
     Hence a is comparable to sqrt(beta) exp(-pi beta), and the error bound
-    and the two inequalities are uniform over these trajectories.  The
+    and the two ineualities are uniform over these trajectories.  The
     periodic plane limit is unique and has a positive full OS gap for
     every fixed beta > 0.
 
-Admission and coverage, calibration, the seed inequality, the upper
+Admission and coverage, calibration, the seed ineuality, the upper
 bound, and the periodic-limit statements are the corresponding
 propositions, calibration estimate, concluding seed estimate, upper
 bound, and uniqueness/gap theorems of those sharp lattice bounds; the
@@ -673,13 +673,13 @@ numerator and denominator bounds its difference from the cylinder
 expectation by 2 ||F||_inf s(w/2,w).  By the basic tails, cylinders and
 square tori therefore have the same local limits, the unique periodic
 plane state.  Taking a subsequence realizing the limsup of E(w) and
-passing the slab inequality to the plane, applied to reflected local
+passing the slab ineuality to the plane, applied to reflected local
 autocorrelations, positivity of their spectral measures and density in
 the OS space prove m(beta)/a >= limsup E(w); an infinite limsup would
 give an infinite gap, contradicting m(beta)/a <= log 8.
 
 Take B = 10 in the truncated-rate convergence.  Its limit is at most
-log 8 < 10, by the preceding inequality and m(beta)/a <= log 8, and
+log 8 < 10, by the preceding ineuality and m(beta)/a <= log 8, and
 uniform convergence makes the truncation inactive for all large widths,
 uniformly over the trajectories.  Thus
 
@@ -687,7 +687,7 @@ uniformly over the trajectories.  Thus
     | E(w) - E_inf | <= C_H w^{-1/4},
     E_inf <= m(beta)/a <= log 8.
 
-For the reverse inequality fix beta, a and keep t comparable to sqrt(w).
+For the reverse ineuality fix beta, a and keep t comparable to sqrt(w).
 Apply the detector and visibility lemmas at n = t/a.  Since E(w) is
 bounded, tau_{beta,w/a} = exp(-a E(w)) and the detector overlap are
 bounded below by positive constants for the fixed trajectory, so some
@@ -884,7 +884,7 @@ scalar resolvent applied to any nonzero nonnegative component, including
 its contact trace, and a nonnegative source supported only on contact
 hyperplanes has nonnegative resolvent by the form maximum principle.
 
-The absolute-value form inequality permits a ground vector to be chosen
+The absolute-value form ineuality permits a ground vector to be chosen
 nonnegative.  In its component equations the off-diagonal contacts give
 nonnegative sources, and moving a sufficiently large positive multiple of
 the component to the right produces a nonzero interior source whenever
@@ -897,7 +897,7 @@ integers, while the exchange (11),(22) <-> (12),(21) changes n_{11} by one
 and connects successive admissible counts; permutations and the positive
 multiplicities from identical-species symmetry do not affect this
 argument.  If a real ground vector changed sign, the absolute-value
-inequality and the variational principle would make its positive and
+ineuality and the variational principle would make its positive and
 negative parts nonnegative ground vectors with disjoint supports,
 contradicting strict positivity.  This proves ground-state simplicity.
 
@@ -1207,7 +1207,7 @@ positive right limit at zero.  If its minimum on the positive half-line
 were negative it would be attained at some X > 0, and averaging against
 K_1 with zero exterior value would give a value strictly larger than that
 minimum, contradicting U = K_1 * U.  Thus U >= 0, and the averaging
-equation with positivity of the kernel makes the inequality strict; the
+equation with positivity of the kernel makes the ineuality strict; the
 formula for V gives V > 0.  This proves the lemma.
 
 The normalization U(X) ~ sqrt(X) matches the leading outer terms.
@@ -1811,7 +1811,7 @@ sum t_{ij}, so Gaussian convolution dominates them and gives spatial
 decay for compactly supported initial data; it also justifies convergence
 to the stated initial condition, first for data supported strictly inside
 the chamber and then by density.  The contact form is bounded below by
-the hyperplane trace inequality, so uniqueness of its L^2 heat evolution
+the hyperplane trace ineuality, so uniqueness of its L^2 heat evolution
 identifies K_s with exp(-s H_c).
 
 For the contour lowering process the momenta from last to first.  After
@@ -2008,9 +2008,9 @@ radii small enough that this uses less than half the original pair-gap
 decay gives the displayed estimate.
 
 Fix a circle length W_0 > 0 and choose finitely many seams and smooth
-nonnegative permutation-invariant cutoffs vartheta_alpha with
-sum_alpha vartheta_alpha^2 = 1, such that all positions in the support of
-vartheta_alpha stay a positive distance from the corresponding seam: at
+nonnegative permutation-invariant cutoffs theta_alpha with
+sum_alpha theta_alpha^2 = 1, such that all positions in the support of
+theta_alpha stay a positive distance from the corresponding seam: at
 fixed particle number, among sufficiently many candidate seams at least
 one stays a fixed distance from every position, and cutting at that seam
 identifies each localized function with a compactly supported function on
@@ -2059,10 +2059,10 @@ wavefunction has an L^2 limit under every pairing coordinate map.
 
 The form localization identity gives
 
-    sum_alpha < vartheta_alpha psi_j,
-            ( H_{c_j} + M c_j^2/2 ) vartheta_alpha psi_j >
+    sum_alpha < theta_alpha psi_j,
+            ( H_{c_j} + M c_j^2/2 ) theta_alpha psi_j >
         = < psi_j, ( H_{c_j} + M c_j^2/2 ) psi_j >
-          + sum_alpha || |grad vartheta_alpha| psi_j ||^2.
+          + sum_alpha || |grad theta_alpha| psi_j ||^2.
 
 The error is bounded independently of j and every localized term is
 nonnegative by the line resolution, so it suffices to prove the assertion
@@ -2137,7 +2137,7 @@ complement retain additional norm.  Finally remove the momentum cutoff
 using the uniform C'/R^2 bound and take a diagonal subsequence over the
 finitely many seam charts and pair assignments and a countable
 exhaustion of the separation and gap cutoffs; the identity
-sum_alpha vartheta_alpha^2 = 1 gives the asserted total norm, and
+sum_alpha theta_alpha^2 = 1 gives the asserted total norm, and
 applying the same cutoffs to two sequences, strong convergence on the
 retained charts gives convergence of their inner products with
 Cauchy-Schwarz controlling all discarded parts.
@@ -2334,7 +2334,7 @@ other pairs, and the next bound controls precisely those products.
 The edge lower bound for rho_{B_c} also bounds rho_ext below by a
 constant times sqrt(1 + B_c) in a fixed interval just outside each
 endpoint, and convolution over these intervals proves the first
-inequality; its final comparison follows from the scale estimate at the
+ineuality; its final comparison follows from the scale estimate at the
 end of the counting-density lemma, for larger W' the left side only
 gaining the needed margin.
 
@@ -2759,7 +2759,7 @@ pair relative to its (0,0) contribution is therefore
         / | ( (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,
+To see the ineuality 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
@@ -2904,7 +2904,7 @@ Consequently
     sum_r epsilon(u_r) <= W int_{-B_c}^{B_c} epsilon_+(x) d(x) dx + C/W
         <= W int_R epsilon_in(x) d(x) dx + C/W,
 
-the last inequality by positivity, since any positive portion outside
+the last ineuality by positivity, since any positive portion outside
 the counting interval is only omitted on the left-hand side.
 
 For the lower bound choose M as in the statement.  The endpoint
@@ -3107,7 +3107,7 @@ edge intertwiner carry the same shape factor, and both alias divisions
 and their gluing apply, the common real bounds giving positivity.  At
 depth zero the specified strip lift is retained without the fine
 inverse-curl correction, whose deletion improves the positive-defect
-inequality, and its second-stack rows are the square root after removing
+ineuality, and its second-stack rows are the square root after removing
 c_0/2 on each side of the edge matrix.  Zero rows are added to give common
 index sets, and the direct sum of the q and w constructions multiplies all
 w edge, observation, and propagated output entries by sqrt(chi).
@@ -3325,7 +3325,7 @@ observation on the cover and preserves both winding integers.
 
 Put A_Y = sum_{x,z in Y} Q_Y(x) Q_Y(z) d(s_x, s_z) with s_x = (q_x, w_x).
 The medoid obeys sum_x Q_Y(x) d(m_Y, s_x) <= A_Y, by averaging the same
-expression over candidate constituents, and the triangle inequality gives
+expression over candidate constituents, and the triangle ineuality gives
 
     d(m_Y, m_Y') <= A_Y + A_Y'
         + sum_{x in Y, z in Y'} Q_Y(x) Q_Y'(z) d(s_x, s_z).
@@ -3605,7 +3605,7 @@ it is already linear.  Thus
 is a principal restriction of the numerical precision
 d^* A_s^* A_s d + Q^* Q, direct-summed with the real coordinate and its
 chi weights.  Every row incident on an exterior column is selected, and
-the block Poincare inequality, including Q, gives
+the block Poincare ineuality, including Q, gives
 
     c_L I <= A_fl <= C_L I
 
@@ -4101,7 +4101,7 @@ chosen with length at most C(1 + |x-o|) after a fixed load enlargement:
 follow the old connected cube record and add a shortest shortcut from the
 anchor whenever the current graph distance exceeds three times the
 ambient distance, each added length being at most a fixed multiple of the
-travel since the preceding shortcut, by the triangle inequality; add
+travel since the preceding shortcut, by the triangle ineuality; add
 connections inside cubes and a full enlarged anchor cube for nearby
 arguments.  The total record length remains O(M_j s), and orbit
 symmetrization uses its common graph.  If the enlargement is no longer
@@ -4198,7 +4198,7 @@ stronger suppression than their cap, the canonical contraction is
 triangular with strict diagonal part, and since |T - 1| = O_L(H_j^{-1})
 the change from input to output normalization also has the permitted
 H_j^{-c} |lambda| size.  These facts prove the two contraction
-inequalities.
+ineualities.
 
 For completeness, the constants close in the following order.  First fix
 the direct coefficients and the deletion-radius constants needed in the
@@ -4223,7 +4223,7 @@ H_j >= H,
 
 Each follows from a positive power of H_j dominating fixed logarithms or
 from taking P_0 above the fixed powers of M_j.  The extra curvature
-inequality C_L M_j^D t_j/P_j << 1 follows uniformly from H_j <= C beta
+ineuality C_L M_j^D t_j/P_j << 1 follows uniformly from H_j <= C beta
 and the same final increase of H, all L-independent constants selecting
 L were identified before this step, and enlarging P_0 does not enlarge
 the fixed exponent D.
@@ -4439,7 +4439,7 @@ deterministic drift changes Brownian measure by the Cameron-Martin
 exponential; differentiating that exponential gives scores bounded by
 fixed polynomials in N and g^{-1}, the preceding transformation estimates
 controlling the differentiated path factors and profiles.  Haar and
-Lebesgue endpoint translations have Jacobian one, and Holder's inequality,
+Lebesgue endpoint translations have Jacobian one, and Holder's ineuality,
 using a small part of the Gaussian reserve for the scores, leaves the tail
 bounds.  Off the path-size profiles the deleted-row factors use the paths
 only through H(s)q(s) and deterministic circle translation, whose
@@ -4748,7 +4748,7 @@ real interaction loss
 
 one half of the squared normalized matrix-spin chord, is at the flagged
 endpoint at least c r_e^2 by the principal vertical branch.  The triangle
-inequality for the matrix-spin chord gives
+ineuality for the matrix-spin chord gives
 
     1 - cos( U(s) ) q(s) . q'(s)
         >= c r_e^2 - C g^2 ( N_1^2 + N_2^2 ),
@@ -4756,7 +4756,7 @@ inequality for the matrix-spin chord gives
 using the path displacement estimate and decreasing c.  Only a
 sufficiently small fixed fraction of the interaction loss is used for
 this payment, its adverse term having an exponential moment controlled
-by the path lemma.  A fixed Holder inequality retains positive fractions
+by the path lemma.  A fixed Holder ineuality retains positive fractions
 of the horizontal and stochastic reserves.  Each row touches only
 boundedly many bonds, both endpoints flagged share the same loss, and
 all other touching interactions have modulus at most one.  Endpoint
@@ -4973,7 +4973,7 @@ stiffness-ratio, and coupling bands, uniformly along the whole trajectory.
         c_x, c_t = beta + O_L(1)
 
     for which both evolutions are admitted through layer zero and, with
-    the distances of the two contraction inequalities,
+    the distances of the two contraction ineualities,
 
         |lambda_j| + U_j <= C_L rho^{N-j},   lambda_0 = 0,
         max(q_*, L^{-1}) < rho < 1.
@@ -5001,7 +5001,7 @@ every such segment, including its first attempted output,
     | z_{h+1} - rho^{-1} z_h | <= C''_L + C''_L H^{-c} D.
 
 Choose D large and then H large so that the right side of the second
-inequality is much smaller than (1 - rho)D; the comparison with the
+ineuality is much smaller than (1 - rho)D; the comparison with the
 reference then puts the whole barrier, and the next coefficient step,
 strictly inside the admitted bands.
 
@@ -5148,14 +5148,14 @@ multiplication by every fixed logarithmic factor needed here.
 
     Lemma (selected kinetic rows).  For a sufficiently large dyadic
     integer B_T comparable to a fixed power of log H, suppose both
-    periods are multiples of 2 B_T.  There is 0 < vartheta < 1 such that
+    periods are multiples of 2 B_T.  There is 0 < theta < 1 such that
     for every set I of unoriented rows and either time seam,
 
         (1/Z_{F,+}^end) int_F exp( X_F^eta ) Xi_F^eta
-            exp( vartheta sum_{e in I} Y_e ) d nu
+            exp( theta sum_{e in I} Y_e ) d nu
             <= exp( C_L |I| B_T^2 log H ),
 
-    and the choices may make 1 - vartheta smaller than any prescribed
+    and the choices may make 1 - theta smaller than any prescribed
     fixed inverse power of log H.
 
 At the fine scale, cut the flat reference into two half cylinders and
@@ -5217,7 +5217,7 @@ Z_{F,+}^end.  Finally average over the B_T^2 translations of the grid:
 each row is internal for a fraction at least 1 - C_L D_1/B_T of them, and
 generalized Holder with Y_e >= 0 gives the selected-row estimate with
 
-    vartheta = (1 - delta)(1 - C_L D_1 / B_T),
+    theta = (1 - delta)(1 - C_L D_1 / B_T),
 
 reflection covariance permitting the twist to be moved separately for
 each grid.  Choosing delta small and then B_T large gives the stated
@@ -5267,7 +5267,7 @@ partition D(z).
 
 At every lift of a point of U^c all squares containing that point have
 average gradient at most alpha, and the concentric-average proof applies
-on the whole cover: the discrete L^1 Poincare inequality compares the
+on the whole cover: the discrete L^1 Poincare ineuality compares the
 point with averages on doubled squares, at scale R at cost C R times the
 average gradient, and comparing two such averages at a scale comparable
 to their separation and summing the geometric scales gives, in the
@@ -5361,9 +5361,9 @@ row read by I_E is bounded by C_L H t_0^2 = C_L p_0^2, and the count
 
 The row estimate and the replacement estimate are now combined, which is
 the step that avoids any pointwise division by a signed restricted sum.
-First choose B_T and vartheta so that
+First choose B_T and theta so that
 
-    (1 - vartheta) C_L d^2 < c / (4 d^2),
+    (1 - theta) C_L d^2 < c / (4 d^2),
     p_0^2 / d^2 >> d^2 B_T^2 log H + log H + 1.
 
 The first is achieved by the selected-row moment lemma and the second by
@@ -5391,8 +5391,8 @@ the edit energy bounds give the exponent comparison
         <= C_L n - c p_0^2 n / d^2 + sum_{e in I_E} Y_e(z').
 
 Retain a fixed fraction of the original direct energy on edges incident
-on E, decreasing c; the first parameter inequality then permits replacing
-the last coefficient by vartheta at a cost of at most a fixed fraction of
+on E, decreasing c; the first parameter ineuality then permits replacing
+the last coefficient by theta at a cost of at most a fixed fraction of
 the negative term.
 
 Apply the uniform bound on the conditional fill density and integrate the
@@ -5413,12 +5413,12 @@ retaining its flatness constraint.  The resulting upper bound is
 
     exp{ C_L n log H - c'' p_0^2 n / d^2 } (1/Z_{F,+}^end)
         int_F exp(X_F^eta) Xi_F^eta
-            exp( vartheta sum_{I_E} Y_e ) d nu.
+            exp( theta sum_{I_E} Y_e ) d nu.
 
 There is no injective-change-of-variables assertion in this argument: the
 original and filled variables are integrated separately using the
 conditional-density bound.  The selected-row moment estimate, the count
-|I_E| <= C d^2 n, and the second parameter inequality then give the
+|I_E| <= C d^2 n, and the second parameter ineuality then give the
 claimed integrated gain.
 
     Proposition (integrated comparison on flat rectangles).  Fix a