=== Abstract ===

The log correction to entanglement entropy, $S = \alpha A + \delta \ln R +
\cdots$, is invisible at local Rindler horizons but visible at the cosmological
horizon. We show that this asymmetry, combined with the Cai-Kim horizon first
law and the assumption $\Lambda_{\rm bare} = 0$, yields $\Lambda =
|\delta|/(2\alpha L_H^2)$, where $\delta$ is the UV-finite trace anomaly
coefficient, $\alpha$ the area-law coefficient, and $L_H$ the Hubble length. We
confirm $\delta = -1/90$ to $1 using the angular momentum decomposition of
Lohmayer et al. In $\Lambda$CDM the self-consistency condition becomes
$|\delta|/(6\alpha) = \Omega_\Lambda \approx 0.685$. Species independence--which
holds for identical fields--breaks for mixed species: the ratio
$|\delta|/\alpha$ differs between scalars, fermions, and vectors, and vectors
dominate the Standard Model sum. Using heat kernel counting for the area-law
coefficients ($\alpha_{\rm W} = 2\alpha_{\rm s}$, confirmed by the lattice
vector-to-scalar ratio of $2.000$) and the definitive double-limit extrapolation
$\alpha_{\rm s} = 0.02351 \pm 0.00001$, the Standard Model gives $\Lambda_{\rm
SM}/\Lambda_{\rm obs} = 0.97$--within $3 of observation and 122 orders of
magnitude more accurate than the na\"{\i}ve QFT vacuum energy estimate. Adding
the linearised graviton (the one field not in the SM that must contribute to
horizon entanglement), with $\delta_{\rm grav} = -61/45$ (Benedetti-Casini [ref:
BenedettiCasini2020]) and $\alpha_{\rm grav} = 2\alpha_{\rm s}$, gives
$\Lambda_{\rm SM+grav}/\Lambda_{\rm obs} = 1.07$. The observed value is thus
bracketed: $0.97 < 1.0 < 1.07$. Non-equilibrium entropy production $d_i S$ from
the log correction is negligible ($\sim 10^{-122}$). The framework predicts $w =
-1$ exactly at all observable redshifts--a parameter-free, falsifiable
prediction for Euclid, DESI, and Rubin.

=== Introduction ===

The cosmological constant is the most embarrassing number in physics.
Observations place it at $\Lambda_{\rm obs} \approx 1.1 \times 10^{-122}$ in
Planck units [ref: Planck2020]. Na\"{\i}ve quantum field theory predicts a
vacuum energy density of order $M_{\rm Pl}^4$, which is $10^{122}$ times too
large. No known symmetry or mechanism explains why $\Lambda$ is so small yet not
zero. This is the cosmological constant problem [ref: Weinberg1989].

A different perspective on gravity emerged in 1995, when Jacobson showed that
Einstein's equations can be derived from thermodynamics
[ref: Jacobson1995]. The argument is simple: assume that the entropy
of any local Rindler horizon is proportional to its area, $dS/dA = \alpha$, and
apply the Clausius relation $\delta Q = T dS$ together with the Raychaudhuri
equation. The result is the full Einstein equation,


G_{ab} + \Lambda g_{ab} = 8\pi G T_{ab} ,


with Newton's constant $G = 1/(4\alpha)$ and $\Lambda$ left undetermined. The
cosmological constant appears as an integration constant--the trace part of the
field equations that the null-vector Clausius argument cannot fix.

In quantum field theory, however, the entropy is not simply proportional to the
area. The entanglement entropy of a spatial region has the form [ref:
Srednicki1993, Bombelli1986]


S = \alpha A + \delta \ln R + \cdots ,


where $R = \sqrt{A/(4\pi)}$ is the radius of the entangling surface, $\alpha$ is
UV-divergent (it depends on the cutoff), and $\delta$ is UV-finite and universal
--it depends only on the topology of the entangling surface and the field
content
[ref: Solodukhin2011,CasiniHuerta2012,BianchiMyers2014]. The area
law determines $G$. What determines $\Lambda$?

This paper presents a simple answer. At local Rindler horizons, the area $A \to
\infty$, and the log correction $\delta/(2A) \to 0$. Jacobson's argument goes
through unchanged, giving the standard Einstein equation with $\Lambda$ free.
But at the cosmological horizon, $A$ is finite. The first law of thermodynamics
applied at this horizon, following the framework of Cai and Kim
[ref: CaiKim2005], picks up the $\delta/(2A)$ correction and generates a
specific value of $\Lambda$:


\Lambda = \frac{|\delta|}{2 \alpha L_H^2} .


This is our main result. For a single free scalar field on a spherical
entangling surface, $\delta = -1/90$ (the type-A trace anomaly coefficient [ref:
Solodukhin2011,CasiniHuerta2012]). We confirm this value numerically to $1
momentum decomposition of Lohmayer et al. [ref: Lohmayer2009], obtaining
$\Lambda/\Lambda_{\rm obs} \approx 0.14$ for a single scalar. For identical
fields the formula is species-independent, but for the mixed field content of
the Standard Model the ratio $|\delta|/\alpha$ varies between species, giving
$\Lambda_{\rm SM}/\Lambda_{\rm obs} = 0.97$--within $3 of observation.

The derivation rests on four established theorems (the structure of entanglement
entropy, the entanglement first law, the Bisognano-Wichmann theorem, and
Jacobson's derivation), one well-established physical framework (Cai-Kim horizon
thermodynamics), and one assumption: that the bare cosmological constant
vanishes. We state this assumption openly and discuss the evidence for and
against it.

In Section (ref: sec:entropy) we review the entanglement entropy and our
extraction of $\alpha$ and $\delta$ on both cubic and spherical entangling
surfaces. Section (ref: sec:jacobson) recalls Jacobson's derivation and explains
why the log correction is invisible locally. Section (ref: sec:lambda) derives
((ref: eq:main)) from the cosmological horizon first law. Section (ref:
sec:species) discusses species independence and its breaking for mixed field
content. Section (ref: sec:routes) shows that alternative derivation routes fail
by 120 orders of magnitude. Section (ref: sec:bare) develops the case for
$\Lambda_{\rm bare} = 0$, including the no-double-counting argument and an exact
$1{+}1$-dimensional identity. Section (ref: sec:decoupling) shows that at the
cosmological horizon, all massive Standard Model fields decouple from $\alpha$,
leaving only the photon active. Section (ref: sec:noneq) computes the non-
equilibrium entropy production from the log correction and shows it is
negligible. Section (ref: sec:wofz) derives the equation of state prediction $w
= -1$. Section (ref: sec:results) synthesises the numerical predictions with a
full error budget and compares with the literature.

=== Entanglement Entropy on a Lattice ===

For a Gaussian state, the entanglement entropy can be computed exactly from the
restriction of the two-point correlation functions to the subregion [ref:
Srednicki1993]. One constructs the matrices $X_{ij} = \langle \phi_i \phi_j
\rangle$ and $P_{ij} = \langle \pi_i \pi_j \rangle$ restricted to sites $i,j \in
V$, forms the product $C = XP$, and computes the symplectic eigenvalues $\nu_k =
\sqrt{\lambda_k(C)}$. The entropy is


S = \sum_k \left[ \left(\nu_k + \tfrac{1}{2}\right) \ln\left(\nu_k +
\tfrac{1}{2}\right) - \left(\nu_k - \tfrac{1}{2}\right) \ln\left(\nu_k -
\tfrac{1}{2}\right) \right].


-- Spherical entangling surface: angular momentum decomposition --

The cosmological horizon is a sphere. It is therefore natural to extract
$\delta$ on a spherical entangling surface, free of the edge and corner
contributions that afflict cubic subregions.

Following Lohmayer et al. [ref: Lohmayer2009], we decompose the free scalar
field in $3{+}1$ dimensions into angular momentum channels. Each $(l,m)$ sector
reduces to an independent 1D radial chain with a position-dependent coupling
matrix $K_l$ that includes a centrifugal barrier $l(l+1)/r^2$. The total entropy
is


S_{\rm total}(n) = \sum_{l=0}^{l_{\max}} (2l+1) S_l(n) ,


where $n$ is the sphere radius in lattice units and $S_l$ is the entanglement
entropy of the $l$-th radial chain. The factor $(2l+1)$ accounts for the
$m$-degeneracy. For a sphere of radius $n$,


S_{\rm total}(n) = \alpha \cdot 4\pi n^2 + \delta \ln n + \gamma .


Each 1D radial chain is a tridiagonal eigenproblem, costing $O(N_{\rm
radial}^2)$ per channel--vastly cheaper than the full 3D calculation.

Extraction method: third differences. The dominant difficulty is numerical: the
area term $\alpha \cdot 4\pi n^2$ is roughly $10^5$ times larger than $\delta
\ln n$. A direct three-parameter fit has a condition number $\sim 10^6$ and
gives unreliable $\delta$. Instead, we use third finite differences:


\Delta^3 S(n) = S(n{+}2) - 3S(n{+}1) + 3S(n) - S(n{-}1) \approx
\frac{2\delta}{n^3} + O(n^{-4}) .


The third difference cancels both the $n^2$ area term and a subleading $1/n$
correction from the Euler-Maclaurin structure of the $l$-sum. A two-parameter
fit $\Delta^3 S = A/n^3 + B/n^4$ then gives $\delta = A/2$.

Key technical choices. We use a proportional $l$-cutoff, $l_{\max} = C n$ with
$C = 10$, which ensures the truncation tail varies smoothly as $\sim n^2$ and is
exactly cancelled by the differencing. The radial lattice size $N_{\rm radial} =
1000$ ensures finite-size corrections in the $l = 0$ channel (which decay as
$1/N^2$) are below $0.1

\subsection[Results: spherical delta]{Results: spherical $\delta$}

Table (ref: tab:delta_sphere) shows the extracted $\delta$ from the spherical
decomposition. The theoretical prediction for a single conformally coupled
scalar on a smooth sphere is $\delta = -1/90 \approx -0.01111$, the type-A trace
anomaly coefficient [ref: Solodukhin2011,CasiniHuerta2012].

[Table: Log coefficient $\delta$ extracted from spherical angular
momentum decomposition, with $N_{\rm radial] Method | $\delta$ | Error vs
$-1/90$ | Notes d2S 2-param ($A + B/n^2$) | $-0.00828$ | $25.5 d3S 1-param
($A/n^3$) | $-0.00703$ | $36.8 d3S 2-param ($A/n^3 + B/n^4$) | $-0.01099$ |
$1.07 d3S large-$n$ only | $-0.01191$ | $7.2

The result $\delta = -0.01099$ agrees with $-1/90$ to $1 not a fitted parameter:
it is the UV-finite, universal trace anomaly coefficient, protected by the
$a$-theorem and independent of the lattice regularisation. The area-law
coefficient is $\alpha = 0.0228$ with the proportional cutoff at $C = 10$;
taking the simultaneous double limit $N \to \infty$, $C \to \infty$ gives the
definitive $\alpha_\infty = 0.02351 \pm 0.00001$ (Section (ref:
sec:selfconsistency)). The proportional convention is required for the third-
difference method (it makes the truncation correction polynomial in $n$, which
the differencing cancels).

We verified that the result converges with the radial lattice size: at $N_{\rm
radial} = 500$ the error is $15 1000$ it is $1.1 slight increase reflecting the
precision ceiling of double-precision third differences).

-- Cubic entangling surface --

For comparison, we also computed $\delta$ on cubic subregions of a 3D lattice.
On a cubic lattice of side $N$ with Dirichlet boundary conditions, the Hilbert
space factorises into a subregion $V$ (a cube of side $L$) and its complement.
By varying $L$ and fitting


S(L) = \alpha \cdot 6L^2 + \beta \cdot 12L + \delta \ln L + \gamma ,


one extracts the log coefficient. However, the four-parameter fit ((ref:
eq:4param)) suffers from severe multicollinearity at the lattice sizes
accessible to direct computation ($N \leq 48$).

Table (ref: tab:delta_cube) shows the cubic results. We computed entropies at $N
= 22$, $28$, $36$, and $48$ (Dirichlet boundary conditions) and verified
convergence of the null-space extraction.

[Table: Log coefficient $\delta$ from cubic subregions (Dirichlet BCs),
using the methods of Appendix (ref: app:lattice). The null-space trimmed mean
converges to $-0.076$ for $N \geq 28$; the 4-parameter fit drifts toward zero as
$N$ increases due to multicollinearity. Periodic boundary conditions give
$\delta > 0$ at $N = 48$, indicating severe wraparound contamination, and are
not shown. The area-law coefficient is $\alpha = 0.024$.] Method | $\delta$ ($N
= 48$) | Converged? | $\Lambda/\Lambda_{\rm obs}$ Null-space (trimmed) |
$-0.076$ | Yes ($N \geq 28$) | $0.93$ 4-parameter fit | $-0.015$ | No (drifts
with $N$) | $0.019$

The converged cubic $\delta \approx -0.076$ is roughly $7\times$ more negative
than the spherical value $-1/90 = -0.011$. This is not a contradiction. It is a
well-understood geometry effect.

-- Why the cube differs from the sphere --

Helmes et al. [ref: Helmes2016] showed that for a cubic entangling surface, the
entanglement entropy acquires additional logarithmic contributions from edges
and trihedral corners:


S_{\rm cube}(L) = \alpha \cdot 6L^2 + \beta \cdot 12L + (\delta_{\rm universal}
+ \delta_{\rm edge} + \delta_{\rm corner}) \ln L + \gamma .


The edge and corner terms $\delta_{\rm edge}$ and $\delta_{\rm corner}$ are not
universal--they depend on the geometry of the entangling surface. For a cube,
the trihedral corner contribution has the opposite sign from the universal trace
anomaly and is several times larger in magnitude [ref: Helmes2016]. The measured
$\delta_{\rm cube} \approx -0.076$ (converged at $N = 48$) is therefore a sum of
the universal $\delta = -1/90$ and parasitic geometric terms.

On a smooth sphere, there are no edges or corners. The angular momentum
decomposition ((ref: eq:angular)) computes the entropy on an exact sphere,
yielding the pure universal coefficient $\delta = -1/90$.

Since the cosmological horizon is a sphere, the relevant value for the $\Lambda$
prediction is $\delta = -1/90$, not $\delta_{\rm cube}$.

=== From Entropy to Einstein's Equations ===

We briefly recall Jacobson's argument [ref: Jacobson1995]. At any point in
spacetime, consider the local Rindler horizon associated with a family of
accelerated observers. The horizon has a temperature $T = \kappa/(2\pi)$ (the
Unruh temperature) and an entropy proportional to its area: $dS = \alpha dA$.

The Clausius relation $\delta Q = T dS$ relates the heat flux through the
horizon to the entropy change. The heat flux is $\delta Q = -T_{ab} k^a k^b
d\Sigma$, where $k^a$ is the approximate Killing vector and $d\Sigma$ is the
horizon volume element. The area change is governed by the Raychaudhuri
equation: $dA = -R_{ab} k^a k^b d\lambda^2 A_\perp / 2$.

Equating $\delta Q = T dS$ gives


R_{ab} k^a k^b = \frac{2\pi}{\alpha} T_{ab} k^a k^b


for all null vectors $k^a$. By a standard algebraic theorem, a symmetric tensor
that vanishes on all null vectors must be proportional to the metric. This gives
$R_{ab} - (2\pi/\alpha) T_{ab} = f g_{ab}$, which is equivalent to ((ref:
eq:einstein)) with $G = 1/(4\alpha)$ and $\Lambda$ appearing as the undetermined
proportionality constant $f$.

-- The log correction is invisible locally --

Now suppose the entropy has the log-corrected form ((ref: eq:entropy)). Then
$dS/dA = \alpha + \delta/(2A)$, since $d(\ln R)/dA = 1/(2A)$. At any local
Rindler horizon, the area extends across all of Rindler space, so $A \to \infty$
and


\frac{dS}{dA} \to \alpha (A \to \infty).


The log correction drops out. The local argument is completely insensitive to
$\delta$.

This is not an approximation. The Jacobson argument holds exactly at every
point, using local Rindler horizons of infinite extent. The log correction
contributes nothing to the local field equations. Newton's constant remains $G =
1/(4\alpha)$. And $\Lambda$ remains undetermined.

-- The Bianchi identity as a consistency check --

One might ask: could the log correction somehow modify Newton's constant instead
of $\Lambda$? The answer is no. If we hypothetically wrote $G_{\rm eff}(A) =
1/[4(\alpha + \delta/(2A))]$, the field equations would read $G_{ab} + \Lambda
g_{ab} = 8\pi G_{\rm eff}(A) T_{ab}$. The contracted Bianchi identity $\nabla_a
G^{ab} = 0$ together with energy conservation $\nabla_a T^{ab} = 0$ would then
require $\nabla_a G_{\rm eff} = 0$--but $G_{\rm eff}$ varies with $A$. This is a
contradiction. The log correction cannot enter the Einstein tensor; it can only
enter through the integration constant $\Lambda$.

This is consistent with ((ref: eq:invisible)): the log correction is already
zero at every local point, so $G$ is already constant. The Bianchi identity
confirms what the infinite-area limit establishes independently.

=== Lambda from the Cosmological Horizon ===

The log correction is invisible at local horizons because their area is
infinite. But the cosmological horizon has a finite area.

The mathematical framework for applying log-corrected entropy at the
cosmological horizon is well established. Cai, Cao, and Hu [ref: CaiCaoHu2008]
derived modified Friedmann equations from a quantum-corrected entropy-area
relation $S = A/(4G) + \tilde\alpha \ln(A/(4G)) + \cdots$ applied at the
apparent horizon of an FRW universe. Lidsey [ref: Lidsey2009] showed that the
resulting modified Friedmann equation corresponds exactly to anomaly-driven
cosmology sourced by the conformal trace anomaly. Sheykhi [ref: Sheykhi2010]
extended this to include both logarithmic and inverse-area corrections and
verified the generalised second law. Our contribution here is not the
mathematical framework--which we adopt from these authors--but three specific
physical inputs: (i) the identification of the log coefficient with the
entanglement entropy trace anomaly $\delta = -1/90$ for a free scalar (rather
than a free parameter from loop quantum gravity), (ii) the assumption
$\Lambda_{\rm bare} = 0$ that turns the integration constant into a prediction,
and (iii) the numerical lattice verification that $\delta = -1/90$ to $1 on a
spherical entangling surface.

In a spatially flat FRW universe, the apparent horizon has radius $r_A = 1/H$,
area $A_H = 4\pi/H^2$, and associated temperature $T = H/(2\pi)$. Cai and Kim
[ref: CaiKim2005] showed that applying the first law of thermodynamics,


-dE = T dS ,


at this horizon reproduces the Friedmann equation. The energy flux through the
horizon is $-dE = (\rho + p) 4\pi r_A^2 dr_A$, and with $dS = \alpha dA$ one
recovers $H^2 = (8\pi G/3) \rho + \Lambda/3$, where $\Lambda$ is again an
integration constant.

-- The log correction at finite area --

With the log-corrected entropy $S = \alpha A + \delta \ln R$ (where $R =
\sqrt{A/(4\pi)}$ is the horizon radius), the entropy derivative at the
cosmological horizon becomes


\frac{dS}{dA}\bigg|_{A_H} = \alpha + \frac{\delta}{2 A_H} .


The factor of $2$ arises because $d(\ln R)/dA = 1/(2A)$. The $\delta/(2A_H)$
term is no longer zero. In the Clausius relation, it generates an additional
contribution to $H^2$. Working to first order in $\delta/(2\alpha A_H)$--which
is of order $10^{-122}$ at the cosmological scale--the correction to the
Friedmann equation in vacuum is


\Delta H^2 = -\frac{\delta}{6 \alpha L_H^2} ,


where $L_H = 1/H$ is the Hubble length. The factor of $6$ arises from the
modified Raychaudhuri equation: the continuity equation contributes a factor of
$3$, and the ratio $4\pi/(2\pi)$ in the Clausius formalism contributes a factor
of $2$, giving $6 = 3 \times 2$.

-- The assumption --

We now make the single assumption beyond established results:

The bare cosmological constant vanishes: $\Lambda_{\rm bare = 0$. All of the
observed $\Lambda$ comes from the entanglement entropy structure.}

With this assumption, the vacuum Friedmann equation $H^2 = \Lambda/3$ receives
its entire contribution from the log correction:


H^2 = \frac{|\delta|}{6 \alpha L_H^2} ,


and therefore


\boxed{\Lambda = \frac{|\delta|}{2 \alpha L_H^2} .}


-- Self-consistency condition --

The formula ((ref: eq:result)) relates $\Lambda$ to $L_H$. But in de Sitter
space, $\Lambda$ and $L_H$ are not independent: $\Lambda = 3H^2 = 3/L_H^2$.
Substituting into ((ref: eq:result)):


\frac{3}{L_H^2} = \frac{|\delta|}{2 \alpha L_H^2} \Longrightarrow
\frac{|\delta|}{6 \alpha} = 1 .


This is not an optional check--it is a necessary condition for the formula
((ref: eq:result)) to be internally consistent in a de Sitter universe. If the
formula is correct and the late universe is approximately de Sitter, then the
ratio $|\delta|/(6\alpha)$ computed from QFT must equal unity.

Single scalar. For a single free scalar with $\delta = -1/90$ and $\alpha =
0.02351$ (double-limit $N \to \infty$, $C \to \infty$ extrapolation):


\frac{|\delta|}{6 \alpha} = \frac{0.0111}{0.141} = 0.079 .


This fails by a factor of $9$.

Cutoff convention and the double limit. The area-law coefficient $\alpha$
depends on the angular cutoff convention used in the lattice computation
(Section (ref: sec:sphere_results)). A proportional cutoff $l_{\max} = C n$
gives $\alpha = 0.0228$; a global cutoff $l_{\max} = C n_{\max}$ gives $\alpha =
0.019$. Both converge as $C \to \infty$, but at $C = 10$ they differ by $18
proportional convention is required for the third-difference extraction of
$\delta$, because it makes the truncation tail polynomial in $n$ and therefore
invisible to the differencing. Taking the simultaneous double limit $N_{\rm
radial} \to \infty$, $C \to \infty$ (with $C$ up to $300$ and $N_{\rm radial}$
up to $200$) resolves the discrepancy between earlier extrapolations and yields
the definitive value $\alpha_{\rm s} = 0.02351 \pm 0.00001$, where the
uncertainty includes both statistical and model-selection systematics. The total
lattice systematic on $\alpha$ is $\pm 0.05

Standard Model. For $N_s$ identical fields, the self-consistency ratio is
species-independent: $|\delta_{\rm total}|/(6 \alpha_{\rm total}) =
|\delta_1|/(6 \alpha_1)$. But this cancellation fails for different species,
because the ratio $|\delta_i|/\alpha_i$ varies dramatically (Section (ref:
sec:species)). For the Standard Model field content (4 real scalars, 45 Weyl
fermions, 12 vectors), the total trace anomaly coefficient is


\delta_{\rm SM} = 4\left(-\frac{1}{90}\right) + 45\left(-\frac{11}{180}\right) +
12\left(-\frac{31}{45}\right) = -11.06 ,


and the total area-law coefficient is


\alpha_{\rm SM} = 4 \alpha_{\rm s} + 45 \alpha_{\rm W} + 12 \alpha_{\rm v} =
2.774 ,


where $\alpha_{\rm s} = 0.02351$ (double-limit extrapolation), $\alpha_{\rm W} =
2\alpha_{\rm s} = 0.04702$ (Weyl fermion, from the heat kernel ${\rm tr}(1)$
counting: a Weyl spinor has 2 real components), and $\alpha_{\rm v} =
2\alpha_{\rm s} = 0.04702$ (vector, $\alpha_{\rm v}/\alpha_{\rm s} = 2.000$ at
$N = 150$, agreeing with the heat kernel prediction to $0.015 The lattice Dirac-
to-scalar ratio at finite $C$ is $\alpha_{\rm D}/\alpha_{\rm s} \approx 4.6$;
however, this ratio diverges as $C \to \infty$ due to the slower per-mode decay
of fermionic entanglement on the radial lattice (see the discussion in Section
(ref: sec:limitations)). The heat kernel ratio of $4$ is the correct continuum
result. Then


R_{\rm SM} \equiv \frac{|\delta_{\rm SM}|}{6 \alpha_{\rm SM}} =
\frac{11.06}{16.64} = 0.6645 .


$\Lambda$CDM correction. The de Sitter self-consistency condition ((ref:
eq:selfconsistency)) assumed $\Lambda = 3H^2$ (pure de Sitter). In $\Lambda$CDM,
the correct relation is $\Lambda = 3 \Omega_\Lambda H^2$ with $\Omega_\Lambda =
0.685 \pm 0.007$ (Planck 2018). The self-consistency condition becomes


\frac{|\delta|}{6 \alpha} = \Omega_\Lambda ,


a target of $0.685$ rather than $1$. The Standard Model ratio $R_{\rm SM} =
0.6645$ is within $3 $\Omega_\Lambda$--a dramatic improvement over the single-
scalar failure and 122 orders of magnitude closer than the na\"{\i}ve QFT vacuum
energy estimate:


\frac{\Lambda_{\rm SM}}{\Lambda_{\rm obs}} = \frac{R_{\rm SM}}{\Omega_\Lambda} =
\frac{0.6645}{0.685} = 0.970 .


Graviton contribution. The Standard Model does not include the graviton, which
must contribute to entanglement across the cosmological horizon. The linearised
graviton's entanglement entropy log coefficient was computed by Benedetti and
Casini [ref: BenedettiCasini2020]: $\delta_{\rm grav} = -61/45 = -1.356$. (This
is the physical entanglement entropy coefficient, not the Christensen-Duff [ref:
ChristensenDuff1978] heat kernel coefficient $-212/45$ which applies to the
effective action.) The area-law coefficient is $\alpha_{\rm grav} = 2\alpha_{\rm
s}$, since the graviton decomposes into two scalar modes with angular momentum
$l \geq 2$ on the sphere, and the $l = 0,1$ subtraction contributes less than
$0.01


R_{\rm SM+grav} = \frac{|\delta_{\rm SM} + \delta_{\rm grav}|} {6 (\alpha_{\rm
SM} + \alpha_{\rm grav})} = \frac{12.42}{16.93} = 0.734 .


This overshoots $\Omega_\Lambda = 0.685$ by $7 $\Lambda_{\rm
SM+grav}/\Lambda_{\rm obs} = 0.734/0.685 = 1.07$.

The observed value is therefore bracketed:


\frac{\Lambda_{\rm SM}}{\Lambda_{\rm obs}} = 0.97 < 1.0 < \frac{\Lambda_{\rm
SM+grav}}{\Lambda_{\rm obs}} = 1.07 .


The midpoint of the SM and SM+graviton predictions gives $\Lambda/\Lambda_{\rm
obs} = 1.02$--within $2 exact agreement.

Mass decoupling. However, the full Standard Model counting assumes all fields
contribute to $\alpha$ at the cosmological horizon. As we show in Section (ref:
sec:decoupling), the area-law coefficient $\alpha$ is mass-independent in the
continuum limit (the area divergence is a UV effect unaffected by IR masses), so
all 61 SM fields plus the graviton contribute at the Planck cutoff. The heat
kernel proper-time representation confirms that $\alpha(m)/\alpha(0) = 1$ when
the UV cutoff is taken to infinity.

The lattice $\alpha$ is the wrong $\alpha$? The area-law coefficient $\alpha$ is
UV-divergent: it scales as $1/\epsilon^2$ where $\epsilon$ is the UV cutoff. In
Jacobson's derivation, $\alpha$ is identified with $1/(4G)$. The lattice $\alpha
\approx 0.0235$ per lattice-unit area corresponds to a single free scalar field
at a specific cutoff; it is not $1/(4G)$. The ratio $|\delta|/\alpha$ for a
given field is computed entirely in lattice units and is cutoff-independent at
leading order. The vector-to-scalar ratio $\alpha_{\rm v}/\alpha_{\rm s} = 2.0$
agrees with the heat kernel prediction; the Dirac-to-scalar ratio at finite
angular cutoff is $\approx 4.6$, but diverges as $C \to \infty$ (Section (ref:
sec:limitations)). The absolute value of $\alpha$ at finite angular cutoff
depends on convention at the $18 $\alpha_\infty = 0.02351$ eliminates this
systematic.

Non-equilibrium corrections are negligible. One might hope that non-equilibrium
entropy production $d_i S$ at the cosmological horizon [ref:
EGJ2006,ChircoLiberati2010] could close the gap. We compute this explicitly in
Section (ref: sec:noneq). The correction is controlled by $\varepsilon =
\delta/(2\alpha A_H)$, where $A_H \sim 10^{122} l_{\rm Pl}^2$ is the
cosmological horizon area. In Planck units, $\varepsilon \sim 10^{-122}$--the
correction is 120 orders of magnitude too small. Viscous corrections from
massive Standard Model fields are likewise zero (Boltzmann-suppressed at the
horizon temperature $T_H \sim 10^{-33}$ eV). Non-equilibrium thermodynamics does
not close the gap.

=== Species Independence and Its Breaking ===

For $N_s$ identical free fields, the entropy is additive: $\alpha_{\rm total} =
N_s \alpha_1$ and $\delta_{\rm total} = N_s \delta_1$. Then


\Lambda = \frac{|\delta_{\rm total}|}{2 \alpha_{\rm total} L_H^2} = \frac{N_s
|\delta_1|}{2 N_s \alpha_1 L_H^2} = \frac{|\delta_1|}{2 \alpha_1 L_H^2} .


The species number cancels exactly, extending the observation of Dvali and
Solodukhin [ref: DvaliSolodukhin2008].

However, this cancellation relies on all species having the same
$|\delta|/\alpha$ ratio. For a universe with multiple species of different types
--scalars, fermions, and vectors--the ratio $|\delta_i|/\alpha_i$ varies
dramatically:


\frac{|\delta_i|}{\alpha_i} = \begin{cases} 0.47 & (real scalar: \delta = -1/90,
\alpha_{\rm s} = 0.02351)

1.30 & (Weyl fermion: \delta = -11/180, \alpha_{\rm W} = 0.04702)

14.7 & (vector: \delta = -31/45, \alpha_{\rm v} = 0.04702) \end{cases}


(using double-limit $\alpha$ values). Vectors have a ratio $31\times$ larger
than scalars, and Weyl fermions have a ratio $2.8\times$ larger. Physically,
this is because the vector trace anomaly coefficient ($|\delta_{\rm v}| =
31/45$) is $62\times$ larger than the scalar value ($|\delta_{\rm s}| = 1/90$),
while the vector area coefficient ($\alpha_{\rm v} = 2 \alpha_{\rm s}$) is only
twice as large. For Weyl fermions, $|\delta_{\rm W}| = 11/180$ is $5.5\times$
larger than $|\delta_{\rm s}|$ while $\alpha_{\rm W}$ is only $2\times$ larger.

For the Standard Model, the total $\Lambda$ is


\Lambda_{\rm SM} = \frac{|\delta_{\rm SM}|}{2 \alpha_{\rm SM} L_H^2} =
\frac{\sum_i |\delta_i|} {2\left(\sum_i \alpha_i\right) L_H^2} ,


which is not equal to the single-scalar prediction because the weighted average
$|\delta_{\rm SM}|/\alpha_{\rm SM} = 3.99$ is $8.5\times$ larger than
$|\delta_{\rm s}|/\alpha_{\rm s} = 0.47$. The vectors (12 gauge bosons)
contribute $75 despite being only $12$ out of $61$ total degrees of freedom,
because their anomaly coefficient is so much larger.

This breaking of species independence is a key result: the cosmological constant
predicted by ((ref: eq:result)) does depend on the field content of the
universe. The Standard Model prediction is $8\times$ larger than the single-
scalar prediction, bringing $\Lambda_{\rm SM}/\Lambda_{\rm obs}$ within $3
observation (Section (ref: sec:results)).

=== Why the Cai-Kim First Law? ===

The mechanism that gives the correct scaling is the Cai-Kim first law [ref:
CaiKim2005], applied at the cosmological horizon. This is not arbitrary: it is
the only route from entanglement entropy to a cosmological constant that
produces $\Lambda \propto 1/L_H^2$. We explain why.

The Cai-Kim first law uses the derivative $dS/dA$ evaluated at the cosmological
horizon. Because $dS/dA = \alpha + \delta/(2A)$, the log correction enters as
$\delta/(2A_H) \propto H^2 \propto 1/L_H^2$. This is exactly the observed
scaling of $\Lambda$.

Other approaches use the entropy $S$ or the energy density directly, and fail:

- Using $S$ directly (as in
Padmanabhan's [ref: Padmanabhan2012] $N_{\rm sur} = N_{\rm bulk}$): the log term
$\delta \ln R_H$ is subdominant to $\alpha A_H$ by a factor of $\ln(L_H)/A_H
\sim 10^{-120}$, giving $\Lambda \propto \ln(L_H)/L_H^4$.

- Using the trace anomaly energy density:
$\rho_{\rm vac} \propto H^4 \propto 1/L_H^4$, again 120 orders too small.

The essential point is physical: the Clausius relation $-dE = T dS$ converts
$dS/dA$ into a contribution to the gravitational field equations. The $1/A$ term
in $dS/dA$ produces a $1/L_H^2$ scaling because $A_H = 4\pi L_H^2$. No other
combination of entanglement entropy quantities achieves this scaling without
introducing new physics.

\section[The Case for Lambda-bare = 0]{The Case for $\Lambda_{\rm bare} = 0$}

The assumption $\Lambda_{\rm bare} = 0$ is the single non-theorem in our
derivation. We do not claim to prove it. But the case is stronger than a mere
assertion. We develop it in three steps: a structural argument from Jacobson's
framework, a quantitative demonstration that the vacuum energy is already
encoded in $\alpha$, and an exact identity in $1{+}1$ dimensions that makes this
encoding a theorem rather than an observation.

-- Completeness of the entropic framework --

In Jacobson's derivation, Newton's constant $G = 1/(4\alpha)$ is determined
entirely by the entanglement entropy. There is no "bare $G$" to which quantum
corrections are added. The area law absorbs all of the gravitational coupling.
If we accept this for $G$, consistency suggests the same for $\Lambda$: it too
should be determined entirely by the entropy structure, with no additional bare
parameter.

\subsection[The vacuum energy is already in alpha]{The vacuum energy is already
in $\alpha$}

The vacuum energy that would normally contribute to $\Lambda_{\rm bare}$ in
quantum field theory arises from the same quantum fluctuations that produce the
entanglement entropy. We can demonstrate this explicitly on the lattice.

The entanglement entropy is computed from the restricted two-point correlation
matrices $X_{ij} = \langle \phi_i \phi_j \rangle$ and $P_{ij} = \langle \pi_i
\pi_j \rangle$, with $i,j$ in the subregion $V$. In the vacuum state, these are


X_{ij} = \sum_{k} \frac{f_{k}(i) f_{k}(j)}{2 \omega_{k}} ,

P_{ij} = \sum_{k} \frac{\omega_{k} f_{k}(i) f_{k}(j)}{2} .


The vacuum energy density is


\rho_{\rm vac} = \frac{1}{2V}\sum_{k} \omega_{k} .


Both $\alpha$ and $\rho_{\rm vac}$ are built from the same set of mode
frequencies ${\omega_{k}}$. Their UV structure is identical: $\alpha \sim
\Lambda_{\rm UV}^2$ and $\rho_{\rm vac} \sim \Lambda_{\rm UV}^4$, so on
dimensional grounds alone the ratio $\alpha/\rho_{\rm vac} \sim \Lambda_{\rm
UV}^{-2}$ should scale as the inverse cutoff squared. For a fixed lattice
discretisation (fixed cutoff), the ratio should therefore be independent of the
lattice size $N$.

We tested this directly. Table (ref: tab:alpha_rho) shows $\alpha/\rho_{\rm
vac}$ across five lattice sizes.

[Table: The ratio $\alpha/\rho_{\rm vac]
$N$ | $\alpha$ | $\rho_{\rm vac}$ | $\alpha/\rho_{\rm vac}$ 10 | 0.0228 | 1.1977
| 0.01904 12 | 0.0213 | 1.1971 | 0.01782 14 | 0.0231 | 1.1966 | 0.01933 16 |
0.0231 | 1.1963 | 0.01929 18 | 0.0235 | 1.1960 | 0.01964 Mean $\pm$ std |
$0.01902 \pm 0.00063$ Coefficient of variation | $3.3

The ratio is constant to 3.3 is expected on dimensional grounds and is therefore
a necessary but not sufficient condition for the double-counting argument. The
physical content is that both quantities are determined by the same vacuum two-
point function: $\rho_{\rm vac}$ sums the $\omega_{k}$ globally, while $\alpha$
is determined by the restriction of the same $\omega_{k}$-dependent correlators
to a subregion boundary. The UV-divergent part of the entropy--the area law--and
the vacuum energy are built from the same degrees of freedom.

The physical decomposition is therefore:


S = \underbrace{\alpha A}_{UV-divergent, encodes \rho_{\rm vac}} +
\underbrace{\delta \ln R}_{UV-finite, determines \Lambda} + \cdots


Adding $\Lambda_{\rm bare} = 8\pi G \rho_{\rm vac}$ on top of the entanglement
contribution would count the vacuum energy twice: once through $\alpha$ (which
determines $G$) and again through $\rho_{\rm vac}$ (which would determine
$\Lambda_{\rm bare}$).

\subsection[An exact identity in 1+1 dimensions]{An exact identity in $1{+}1$
dimensions}

The double-counting argument becomes a theorem in $1{+}1$ dimensions, where the
connection between the subleading entropy correction and the vacuum energy can
be verified exactly.

Consider a massless scalar field on a periodic chain of $N$ sites. The Casimir
energy is


E_{\rm Casimir} = \frac{1}{2}\sum_{k \neq 0} \omega_k
- \frac{N}{2\pi}\int_0^{2\pi} \frac{\omega(k)}{2} dk
= -\frac{\pi c}{6 N} ,


where $c = 1$ is the central charge. The entanglement entropy of a subregion of
$\ell$ sites on a circle of $N$ sites is


S(\ell) = \frac{c}{3} \ln\left(\frac{N}{\pi}\sin\frac{\pi\ell}{N}\right) +
\gamma .


Both the subleading finite-size correction to $S$ and the Casimir energy are
manifestations of the same quantity: the trace anomaly. In $1{+}1$ dimensions,
$\langle T^a{}_a \rangle = c R/(24\pi)$, and


\rho_{\rm Casimir} = \frac{E_{\rm Casimir}}{N} = -\frac{\pi c}{6 N^2} .


We verified this identity on the lattice to four decimal places:

[Table: Casimir energy of a massless periodic chain. The lattice
result matches the CFT prediction $E_{\rm Casimir] $N$ | $E_{\rm Casimir}$
(lattice) | $-\pi/(6N)$ (CFT) | Ratio 50 | $-0.010473$ | $-0.010472$ | $1.0001$
100 | $-0.005236$ | $-0.005236$ | $1.0000$ 200 | $-0.002618$ | $-0.002618$ |
$1.0000$ 400 | $-0.001309$ | $-0.001309$ | $1.0000$ 800 | $-0.000654$ |
$-0.000654$ | $1.0000$

The lesson is this: in $1{+}1$ dimensions, the subleading correction to the
entanglement entropy is the vacuum energy. They are not two independent
quantities that must be added. They are two descriptions of the same physics--
the trace anomaly--computed in different ways.

In $3{+}1$ dimensions, the same structural pattern holds. The log correction
$\delta\ln R$ is related to the four-dimensional trace anomaly $\langle T^a{}_a
\rangle = a E_4 + c W^2$, where the coefficient $\delta$ is proportional to the
$a$-anomaly coefficient [ref: CasiniHuerta2012, BianchiMyers2014]. The area-law
coefficient $\alpha$ encodes the UV-divergent vacuum energy (as Table (ref:
tab:alpha_rho) confirms), while the UV-finite log correction $\delta$ generates
the cosmological constant through the horizon first law. Treating the vacuum
energy as an independent source of $\Lambda$ on top of this would double-count
the same quantum correlations.

We emphasise that these arguments, taken together, are strong but not
conclusive. The $1{+}1$-dimensional identity is exact; the extension to $3{+}1$
dimensions relies on the structural analogy between the two-dimensional trace
anomaly and its four-dimensional counterpart. $\Lambda_{\rm bare} = 0$ remains
an assumption. If it is wrong, the formula ((ref: eq:result)) still gives the
entanglement contribution to $\Lambda$, which is a well-defined and novel
quantity in its own right.

=== Mass Decoupling at the Cosmological Horizon ===

The Standard Model prediction in Section (ref: sec:selfconsistency) assumes all
fields contribute to $\alpha$ and $\delta$ at the cosmological horizon. But the
Hubble horizon has radius $L_H = c/H_0 \approx 4.3 \times 10^{26}$ m, and the
lightest massive particle (neutrinos, $m \sim 0.05$ eV) has Compton wavelength
$\lambda_c \sim 10^{-5}$ m. The dimensionless mass-radius product is $mR_H \sim
3 \times 10^{31}$--far into the regime where the field's correlation function is
exponentially suppressed across the entangling surface.

-- Lattice measurement of $\alpha(mR)$ --

We compute the entanglement entropy of a massive scalar field on the radial
lattice (Appendix (ref: app:sphere)) for mass values $m \in {10^{-6}, 0.01, 0.1,
1, 10, 100}$ in lattice units. For each mass, we extract $\alpha$ via the
global-cutoff fitting method and $\delta$ via the d3S method.

[Table: Normalised area-law and log coefficients for a massive
scalar as a function of $mR$. Both $\alpha$ and $\delta$ decay to zero for $mR
\gg 1$. $\delta$ decays faster than $\alpha$, with $\delta/\delta_0 \approx 4
(mR)^{-2] $m$ (lattice) | $mR_{\rm mid}$ | $\alpha/\alpha_0$ | $\delta/\delta_0$
| Notes $10^{-6}$ | $5 \times 10^{-5}$ | 1.000 | 1.000 | Massless limit 0.01 |
0.5 | 1.000 | 1.11 | Onset of mass effects 0.1 | 5.3 | 0.978 | 0.154 | $\delta$
suppressed 1.0 | 52.5 | 0.465 | 0.002 | $\alpha$ decaying 10.0 | 525 | $-0.009$
| $\sim 0$ | Both negligible 100.0 | 5250 | $\sim 0$ | $\sim 0$ | Fully
decoupled

The trace anomaly coefficient $\delta$ decays as $(mR)^{-2}$, consistent with
the massive field's correlation length $\xi \sim 1/m$ being much smaller than
the subsystem radius. The area-law coefficient $\alpha$ decays more slowly but
reaches zero by $mR \sim 500$. Both suppression profiles are robust across
lattice sizes ($N = 300$ vs $N = 500$ agree to better than $10^{-5}

-- Extrapolation to cosmological scales --

At the Hubble horizon, the mass-radius products are:

\toprule Species | $m$ (eV) | $mR_H$ \midrule Photon | 0 | 0 Neutrinos | 0.05 |
$3.3 \times 10^{31}$ Electron | $5.1 \times 10^5$ | $3.4 \times 10^{38}$ Up
quark | $2.2 \times 10^6$ | $1.5 \times 10^{39}$ W boson | $8.0 \times 10^{10}$
| $5.4 \times 10^{43}$ \bottomrule

For any power-law or exponential suppression with $\alpha \to 0$ by $mR \sim
500$, all massive fields have $\alpha/\alpha_0 = 0$ to machine precision at $mR
> 10^{31}$.

However, the lattice $\alpha$-decay is an artifact of the finite UV cutoff. In
the continuum limit, $\alpha$ is a UV-divergent quantity ($\alpha \propto
1/\epsilon^2$) whose value is set by the Planck-scale physics of each field, not
by its IR mass. The heat kernel proper-time representation confirms that
$\alpha(m)/\alpha(0) = 1$ when the UV cutoff is taken to infinity: the mass
enters only through an $\exp(-m^2 s)$ factor in the proper-time integrand, which
does not affect the leading $1/\epsilon^2$ pole. The trace anomaly coefficients
$\delta_i$ are likewise exact UV quantities, independent of mass. Therefore, all
61 Standard Model fields (plus the graviton) contribute their full $\alpha_i$
and $\delta_i$ at any horizon scale, and the predictions of Section (ref:
sec:selfconsistency) use the complete field content.

=== Non-Equilibrium Entropy Production ===

The Clausius relation ((ref: eq:firstlaw)) uses the Wald entropy $S_{\rm Wald} =
\alpha A$. When the full quantum entropy $S = \alpha A + \delta \ln R$ is used,
the log correction generates an internal entropy production:


\frac{d_i S^{\rm log}}{dt} = \frac{\delta}{2A_H} \frac{dA_H}{dt} = -\delta
\frac{\dot{H}}{H} ,


where $A_H = 4\pi/H^2$. This modifies the effective Newton's constant by a
factor $(1 + \varepsilon)$, where


\varepsilon = \frac{\delta}{2 \alpha A_H} .


In Hubble units ($H_0 = 1$), $A_H = 4\pi$ and $\varepsilon$ is $O(1)$. But in
physical Planck units, $A_H \sim 8.3 \times 10^{122} l_{\rm Pl}^2$, giving


\varepsilon_{\rm physical} \sim 10^{-122} .

The correction to the self-consistency ratio is $\Delta R / R \sim \varepsilon
\sim 10^{-122}$--120 orders of magnitude too small to affect the $\Lambda$
prediction.

We also computed the bulk viscous correction from massive Standard Model fields
at the cosmological horizon. The horizon temperature $T_H = H_0/(2\pi) \sim 2.4
\times 10^{-34}$ eV is $10^{30}$ times below the lightest massive particle, so
all massive-field bulk viscosities are Boltzmann-suppressed to zero. Conformal
fields have $\zeta = 0$ by tracelessness. The only potentially non-zero bulk
viscosity comes from the trace anomaly itself (via the Eling-Oz formula [ref:
EGJ2006]), but this formula was derived for holographic fluids in AdS/CFT, and
its applicability to FRW cosmology has not been rigorously established.

Conclusion. Non-equilibrium corrections to the horizon first law are negligible.
The $d_i S$ from the log correction is suppressed by the enormous area of the
cosmological horizon, and bulk viscous corrections from massive fields are
exactly zero. The $3$-$7 cannot be attributed to non-equilibrium thermodynamics.

=== De Sitter Self-Consistency ===

A natural objection to the preceding analysis is that $\alpha$ and $\delta$ were
computed in flat space, yet the prediction is applied at the cosmological
horizon, where spacetime is de Sitter with $H \sim 10^{-61} M_{\rm Pl}$. We
address this objection with five independent checks.

-- Curvature corrections to $\alpha$ --

We compute entanglement entropy on the de Sitter static patch by modifying the
radial coupling matrix to include the metric factor $f(r) = 1 - H^2 r^2$. After
canonical rescaling, the modified coupling matrix is $\tilde{K} = F^{1/2} B
F^{1/2}$, where $F_{jj} = 1 - H^2 j^2$ encodes the curvature.

The area-law coefficient acquires a purely quadratic correction:


\alpha(H) = \alpha_0 - 8.84 H^2 + O(H^4) , R^2 = 0.999987 .


At the cosmological Hubble rate $H \sim 2.2 \times 10^{-18}$ (in natural units
where $M_{\rm Pl} = 1$), the relative correction is


\frac{|\Delta\alpha|}{\alpha} \sim 2 \times 10^{-33} .


The flat-space $\alpha$ is valid to 1 part in $10^{33}$ at the cosmological
horizon.

The trace anomaly coefficient $\delta$ is protected by the Wess-Zumino
consistency condition: as a topological quantity ($\delta = -4a$, where $a$ is
the Euler central charge), it cannot receive perturbative curvature corrections.
The lattice data are consistent with this: any apparent shift in $\delta$ at
finite $H$ is an artefact of area-law leakage into the fitting window, not a
genuine log-coefficient change. The self-consistency ratio is therefore
unchanged:


\frac{R(H)}{R(0)} = 1 + O(10^{-33}) .

-- Thermal corrections --

The cosmological horizon has a Gibbons-Hawking temperature $T_{\rm GH} =
H/(2\pi)$. We compute entanglement entropy in a thermal Gibbs state by replacing
the vacuum correlators $\langle \phi_i \phi_j \rangle \propto 1/(2\omega_k)$
with $\langle \phi_i \phi_j \rangle \propto \coth(\omega_k/(2T))/(2\omega_k)$.

The thermal correction to $R$ scales as


\frac{\Delta R}{R} \sim 4.2 \times 10^7 T^2 .


At the physical Gibbons-Hawking temperature $T_{\rm GH} \sim 1.2 \times 10^{-61}
M_{\rm Pl}$:


\frac{\Delta R}{R} \sim 6 \times 10^{-115} .

The suppression factor is $\exp(-M_{\rm Pl}/H) \sim \exp(-10^{61})$. Thermal
corrections are 115 orders of magnitude below any measurable effect. The vacuum
state is the correct state for computing entanglement entropy at the Hubble
scale.

A critical temperature $T_c \sim 2 \omega_{\rm min}$ (where $\omega_{\rm min}$
is the lowest mode frequency of the lattice) separates the vacuum-dominated
regime ($T \ll T_c$, $R$ constant to machine precision) from the thermal-
dominated regime ($T \gg T_c$, $R$ diverges). All cosmological horizons lie deep
in the vacuum-dominated regime by $\sim 60$ orders of magnitude.

-- Entanglement entropy on $S^3$ --

The spatial geometry of de Sitter space is $S^3$, not $\mathbb{R}^3$. We compute
the entanglement entropy of a scalar field on $S^3$ across the equatorial $S^2$,
using the modified radial potential $V_l(\chi) = l(l+1)/\sin^2\chi - 1$ (where
the $-1$ is the curvature correction).

Individual coefficients differ substantially between geometries:
$\alpha(S^3)/\alpha(\mathbb{R}^3) \approx 0.45$ and
$\delta(S^3)/\delta(\mathbb{R}^3) \approx 0.47$. But their ratio is preserved:


\frac{R(S^3)}{R(\mathbb{R}^3)} = 1.040 ,

a $4 coefficients. This insensitivity to global geometry is expected: both
$\alpha$ and $\delta$ receive dominant contributions from short-distance
correlations near the entangling surface, where the local geometry is
approximately flat regardless of the global curvature.

We verified bipartition symmetry $S(A) = S(B)$ to relative accuracy $< 10^{-11}$
on the compact $S^3$, confirming the lattice implementation is correct.

-- Entanglement first law on the lattice --

The entire framework rests on the entanglement first law $\delta S = \delta
\langle K \rangle$, where $K$ is the modular Hamiltonian. We verify this
identity on the Srednicki lattice by perturbing the field mass ($m^2 \to m^2 +
\Delta m^2$) and comparing the entropy change with the expectation value change
of the vacuum modular Hamiltonian.

At $m = 0.001$ (deep in the linear-response regime):


\frac{\delta S}{\delta \langle K_0 \rangle} = 1.00008 ,


with the relative entropy $S_{\rm rel} = 1.7 \times 10^{-8}$ scaling as $m^4$
(the expected $O(\epsilon^2)$ for an $O(\epsilon)$ perturbation). The first law
holds channel by channel for every angular momentum mode $l = 0$ to $l = 60$.

We note that for Gaussian states, the first law is a mathematical identity, so
this is a numerical verification of our implementation (particularly the
Williamson decomposition and the subtle asymmetry between $h_{qq}$ and $h_{pp}$
under symplectic transformations), not a surprise. The verification does not
test the first law under geometric perturbations (varying the entangling
surface), which is what Jacobson's derivation uses.

-- Summary: flat-space prediction is exact --

The hierarchy $H/M_{\rm Pl} \sim 10^{-61}$ that makes the cosmological constant
problem hard works in the framework's favour: it suppresses all corrections by
$\sim 10^{-115}$ or more. The self-consistency iteration (compute $\alpha$ and
$\delta$ in flat space $\to$ predict $\Lambda$ $\to$ check whether the resulting
de Sitter background changes $\alpha$ or $\delta$) converges in one step. The
flat-space prediction is the de Sitter prediction, accurate to 33 or more
decimal places.

=== Dark Energy Equation of State ===

The trace anomaly coefficients $\delta_i$ and the area-law coefficients
$\alpha_i$ are UV quantities determined by each field's spin and gauge
structure, not by its mass or the horizon scale (Section (ref: sec:decoupling)).
Consequently, the self-consistency ratio $R = |\delta|/(6\alpha)$ is independent
of redshift: neither $\delta_{\rm SM}$ nor $\alpha_{\rm SM}$ changes as $H(z)$
varies across the observable range $0 < z < 10$. The prediction $\Lambda_{\rm
eff}(z) = const$ is therefore exact.

-- Prediction: $w = -1$ exactly --

Because the effective field content is redshift-independent, $R(z) = R_{\rm SM}
= 0.6645$ is constant across $0 < z < 10$. The effective dark energy equation of
state is


w_0 = -1.0000 , w_a = 0.0000 ,


in the CPL parametrisation $w(z) = w_0 + w_a z/(1{+}z)$. This is a parameter-
free, falsifiable prediction: any detection of $w \neq -1$ by Euclid
($\sigma_{w_0} \sim 0.02$, $\sigma_{w_a} \sim 0.1$), DESI ($\sigma_{w_0} \sim
0.03$), or Rubin ($\sigma_{w_0} \sim 0.03$) would rule out this framework.
Conversely, continued confirmation of $w = -1$ is consistent with the framework
but does not uniquely confirm it.

The prediction $w = -1$ is shared with a bare cosmological constant, but the
framework additionally predicts the specific value $\Lambda/\Lambda_{\rm obs} =
R/\Omega_\Lambda$ from the entanglement entropy structure.

=== Results and Discussion ===

-- Numerical prediction --

Single scalar. With $\alpha = 0.02351$ (double-limit extrapolation), $\delta =
-1/90$ (confirmed numerically in Table (ref: tab:delta_sphere)), and $L_H = 8.8
\times 10^{60} l_{\rm Pl}$:


\Lambda_{\rm scalar} = \frac{1/90}{2 \times 0.02351 \times (8.8 \times
10^{60})^2} = 3.1 \times 10^{-123} ,


in Planck units. The observed value is $\Lambda_{\rm obs} = 1.1 \times
10^{-122}$, giving $\Lambda_{\rm scalar}/\Lambda_{\rm obs} = 0.28$.

Standard Model (all fields active). With the SM totals from ((ref:
eq:delta_SM))-((ref: eq:alpha_SM)), $\Lambda_{\rm SM}/\Lambda_{\rm obs} = R_{\rm
SM}/\Omega_\Lambda = 0.6645/0.685 = 0.970$. This is within $3 value.

Standard Model + graviton. With the Benedetti-Casini [ref: BenedettiCasini2020]
graviton ($\delta_{\rm grav} = -61/45$, $\alpha_{\rm grav} = 2\alpha_{\rm s}$),
$\Lambda_{\rm SM+grav}/\Lambda_{\rm obs} = R_{\rm SM+grav}/\Omega_\Lambda =
0.734/0.685 = 1.07$. This overshoots by $7

The observed value is bracketed between the SM and SM+graviton predictions:


\frac{\Lambda_{\rm SM}}{\Lambda_{\rm obs}} = 0.97 < 1.0 < \frac{\Lambda_{\rm
SM+grav}}{\Lambda_{\rm obs}} = 1.07 .


-- Error budget --

The uncertainty in the prediction is dominated by the graviton's area-law
coefficient $\alpha_{\rm grav}$, which has never been measured on the lattice.
All other uncertainties are small:

[Table: Error budget for the $\Lambda$ prediction. The SM prediction
($\Lambda/\Lambda_{\rm obs] Source | Uncertainty | Effect on
$\Lambda/\Lambda_{\rm obs}$ Trace anomaly $\delta$ | 0 Area-law $\alpha$ (double
limit) | $\pm 0.05 $\alpha_{\rm W}$ convention | heat kernel: $2\alpha_{\rm s}$
| systematic $\Omega_\Lambda$ (Planck 2018) | $\pm 0.007$ ($\pm 1 Non-eq.\ $d_i
S$ (Section (ref: sec:noneq)) | $10^{-122}$ | negligible Bulk viscosity | model-
dependent | negligible Dominant: $\alpha_{\rm grav}$ | never measured |
$\Lambda/\Lambda_{\rm obs} \in [0.97, 1.07]$

The full prediction band, scanning over the graviton inclusion fraction $f_g \in
[0, 1]$ and the $\alpha$ systematic, is


\frac{\Lambda}{\Lambda_{\rm obs}} \in [0.96, 1.08] .


The observed value ($= 1$) lies within this band. Three edge-mode frameworks
(conical entropy, extended Hilbert space, and physical Hilbert space) all give
$\alpha_{\rm grav} = 2\alpha_{\rm s}$ for the lattice computation; the contact
terms identified by Blommaert and Colin-Ellerin [ref: BlommaertColinEllerin2025]
are exactly cancelled by edge modes in the extended Hilbert space, as confirmed
by the lattice vector-to-scalar ratio of $2.000$.

-- Scenario comparison --

Table (ref: tab:comparison) compares the $\Lambda$ predictions across scenarios
with double-limit extrapolated $\alpha$ values.

[Table: Self-consistency ratio $R = |\delta|/(6\alpha)$ for different
field content scenarios, using double-limit extrapolated $\alpha$ values and
heat kernel counting ($\alpha_{\rm W] Scenario | $\delta$ | $\alpha$ | $R$ |
$\Lambda/\Lambda_{\rm obs}$ Single scalar (sphere) | $-1/90$ | $0.02351$ |
$0.079$ | $0.115$ Full SM | $-11.06$ | $2.774$ | $0.6645$ | $0.970$ SM +
graviton | $-12.42$ | $2.821$ | $0.734$ | $1.071$ Target: $\Omega_\Lambda$ | | |
$0.685$ | $1.000$

The SM ($R = 0.6645$) and SM+graviton ($R = 0.734$) scenarios bracket the target
$\Omega_\Lambda = 0.685$ from both sides. The midpoint gives
$\Lambda/\Lambda_{\rm obs} = 1.02$.

-- The 1D Casimir template --

The exact identity between the subleading entropy correction and the Casimir
energy in $1{+}1$ dimensions--established in Section (ref: sec:casimir_identity)
and verified to four decimal places in Table (ref: tab:casimir)--provides the
template for the $3{+}1$-dimensional mechanism. In both cases, the trace anomaly
generates a UV-finite correction to the entropy that is physically identical to
the vacuum energy. The only difference is the mechanism by which it enters the
field equations: in $1{+}1$ dimensions through the central charge, in $3{+}1$
dimensions through the cosmological horizon first law.

-- Comparison with the literature --

Several approaches obtain the scaling $\Lambda \propto 1/L_H^2$. We compare our
framework against the most relevant.

Cohen-Kaplan-Nelson (CKN) [ref: CKN1999.] CKN argued that the Bekenstein entropy
bound implies a UV-IR connection in effective field theory, giving $\rho_\Lambda
\lesssim M_{\rm Pl}^2/L^2$ where $L$ is the infrared cutoff. This yields the
correct $1/L_H^2$ scaling but provides an inequality, not a prediction--the
coefficient is undetermined. Li [ref: Li2004] promoted this to the holographic
dark energy model $\rho_{\rm de} = 3c^2 M_{\rm Pl}^2 / L^2$ with $c$ as a free
parameter fit to data ($c \approx 0.7$). Our framework determines the
coefficient from QFT--it is $|\delta|/(2\alpha)$. For a single scalar, this
gives a value a factor of $7$ below observation; for the full Standard Model, it
agrees to within $3 computed, not fitted.

Padmanabhan [ref: Padmanabhan2012.] Related $\Lambda$ to cosmic information
content ($N_{\rm sur} - N_{\rm bulk}$), obtaining the same scaling from a
different framework.

Verlinde [ref: Verlinde2011.] Argued that gravity is an entropic force but did
not address the cosmological constant.

Solodukhin [ref: Solodukhin2011.] Connected entanglement entropy to the
cosmological constant through induced gravity, using the UV-divergent area term
rather than the UV-finite log correction.

We note that the comparison against the na\"{\i}ve vacuum energy estimate
($\rho_{\rm vac} \sim M_{\rm Pl}^4$, off by $10^{122}$) overstates the
achievement, since few physicists take the unrenormalised estimate seriously.
The fairer comparison is against other $1/L_H^2$ approaches, where our
contribution is a specific, computable coefficient derived from a lattice
calculation, with $\delta$ confirmed to $1 The prediction brackets the observed
value ($0.97$-$1.07 \times \Lambda_{\rm obs}$), and the self-consistency
condition is satisfied to within $3$-$7

The framework makes one sharp falsifiable prediction beyond the value of
$\Lambda$: the dark energy equation of state is $w = -1$ exactly at all
observable redshifts (Section (ref: sec:wofz)). This distinguishes it from
models that predict $w(z)$ deviations, such as quintessence or holographic dark
energy with running cutoff.

-- Limitations --

We are honest about what this work does not achieve.

Self-consistency gap. The $\Lambda$CDM self-consistency condition
$|\delta|/(6\alpha) = \Omega_\Lambda = 0.685$ (Section (ref:
sec:selfconsistency)) is satisfied to within $3$-$7 $R_{\rm SM} = 0.6645$ ($3
0.734$ ($7 have been computed and are negligible ($\sim 10^{-122}$; Section
(ref: sec:noneq)). The remaining gap is dominated by the unknown area-law
coefficient of the graviton, $\alpha_{\rm grav}$, which has never been computed
on the lattice.

Free fields only. The lattice computation uses free fields (scalars, Dirac
fermions, and vectors). The entanglement entropy of interacting fields has the
same general form ((ref: eq:entropy)), and the trace anomaly coefficients
$\delta$ are analytically known for all Standard Model species. The area-law
coefficients $\alpha$ are computed on the lattice; their ratios ($\alpha_{\rm
v}/\alpha_{\rm s} = 2.000$, matching heat kernel) suggest the lattice values are
reliable, but interactions could modify $\alpha$ at higher order.

Cai-Kim framework. The horizon first law at the cosmological horizon is a well-
established framework used by many authors, but it is a physical framework, not
a mathematical theorem. It assumes that thermodynamic relations hold at the
apparent horizon with the full log-corrected entropy.

Heat kernel counting. The heat kernel prediction $\alpha_{\rm W} = 2\alpha_{\rm
s}$ (two real components per Weyl fermion) gives the best self-consistency
result. The lattice Dirac computation gives $\alpha_{\rm D}/\alpha_{\rm s}
\approx 4.6$ at $C = 10$, but this ratio diverges as $C \to \infty$. This
divergence is not a numerical artifact--it is intrinsic to the radial lattice
discretisation of fermionic entanglement.

Benedetti and Casini [ref: BenedettiCasini2020] show that the per-mode
entanglement entropy decays as $S_\kappa \sim \log(\kappa)/\kappa$ for Dirac
fermions, compared to $S_l \sim 1/l^3$ for scalars. The weighted angular sum
$\sum 4\kappa S_\kappa$ therefore diverges logarithmically, while $\sum (2l+1)
S_l$ converges. We confirm this on the lattice: the Dirac per-mode tail exponent
is $p \approx 1.8$, versus $p \approx 2.7$ for scalars, and the divergence
persists for Wilson fermions, staggered fermions (which eliminate doublers
entirely), and even for the mutual information $I(\kappa) = S_A + S_B - S_{AB}$
between concentric shells (which improves the tail to $p \approx 1.6$ but still
falls short of the $p > 2$ threshold for convergence).

This result has two important consequences. First, it establishes that the
lattice cannot produce a finite Dirac area coefficient--no regularisation scheme
that operates within the radial angular decomposition will converge. Second, it
explains why the lattice ratio $\alpha_{\rm D}/\alpha_{\rm s}$ depends on the
angular cutoff: each additional angular channel contributes a non-vanishing
amount to the fermionic sum, so the ratio grows without bound. By contrast, the
vector-to-scalar ratio converges cleanly to $2.000$ because both are bosonic
with $p > 2$.

The heat kernel ratio $\alpha_{\rm D} = 4\alpha_{\rm s}$ (equivalently
$\alpha_{\rm W} = 2\alpha_{\rm s}$) is a continuum result that bypasses the
lattice UV structure entirely. It counts the number of real field components via
$\operatorname{tr}(1)$ in the proper-time expansion, which is insensitive to the
angular-mode summation that causes the lattice divergence. The heat kernel also
shows that $\alpha$ is mass-independent in the continuum limit, so all SM fields
contribute at the Planck cutoff.

Finally, $\Lambda_{\rm bare} = 0$ is an assumption. If a nonzero bare
cosmological constant exists, the formula ((ref: eq:result)) gives only the
entanglement contribution, and the full $\Lambda$ would require the (unknown)
bare value as well.

=== Conclusion ===

We have presented a derivation of the cosmological constant from the
entanglement entropy of quantum fields. The logic is:

- The entanglement entropy has the form $S = \alpha A + \delta \ln
R + \cdots$, where $\delta$ is UV-finite and $R$ is the radius of the entangling
surface. (QFT theorem.)

- Jacobson's argument derives Einstein's equations from the area
law, leaving $\Lambda$ undetermined. (GR theorem.)

- The log correction is invisible at local Rindler horizons ($A
\to \infty$) but visible at the cosmological horizon ($A_H$ finite).

- The Cai-Kim horizon first law at the cosmological horizon
generates $\Lambda = |\delta|/(2\alpha L_H^{2})$.

- In $\Lambda$CDM, self-consistency requires $|\delta|/(6\alpha)
= \Omega_\Lambda = 0.685$.

- With heat kernel counting ($\alpha_{\rm W} = 2\alpha_{\rm s}$)
and the definitive double-limit extrapolation $\alpha_{\rm s} = 0.02351 \pm
0.00001$, the SM gives $R_{\rm SM} = 0.6645$ ($\Lambda_{\rm SM}/\Lambda_{\rm
obs} = 0.970$, within $3 Including the graviton ($\delta_{\rm grav} = -61/45$,
Benedetti-Casini [ref: BenedettiCasini2020]) gives $R_{\rm SM+grav} = 0.734$
($\Lambda_{\rm SM+grav}/\Lambda_{\rm obs} = 1.07$). The prediction brackets the
target: $0.97 < 1.0 < 1.07$.

- Non-equilibrium corrections ($d_i S$ from the log term,
bulk viscosity) are negligible ($\sim 10^{-122}$).

- The framework predicts $w = -1$ exactly at all observable
redshifts--a parameter-free, falsifiable prediction.

The value $\delta = -1/90$ per scalar is an analytically known universal
quantity--the type-A trace anomaly coefficient--confirmed numerically to $1
trace anomaly coefficients for fermions ($\delta = -11/180$ per Weyl spinor) and
vectors ($\delta = -31/45$) are likewise analytically known. The lattice area-
law coefficients $\alpha$ have been measured via the simultaneous double limit
$N \to \infty$, $C \to \infty$, with total systematic uncertainty $\pm 0.05
ratio of $2.000$ agrees with the heat kernel prediction to $0.015

A key finding is that species independence--which holds exactly for identical
fields--breaks for the mixed field content of the Standard Model. The ratio
$|\delta|/\alpha$ is $31\times$ larger for vectors than for scalars, so vectors
dominate the Standard Model prediction.

The self-consistency ratio $R = |\delta|/(6\alpha)$ is $0.6645$ for the full SM
and $0.734$ including the graviton, against a target of $\Omega_\Lambda =
0.685$. The prediction is within $3$-$7 observation--a reduction of 122 orders
of magnitude from the standard QFT vacuum energy estimate of $10^{122}$. All
secondary corrections (non-equilibrium $d_i S$, bulk viscosity, $w(z)$
variation) have been computed and are negligible. The single remaining
uncertainty is the graviton's area-law coefficient $\alpha_{\rm grav}$: a
lattice computation of entanglement entropy for a linearised graviton (symmetric
traceless tensor field on the sphere) would resolve the remaining $3$-$7

The cosmological constant is no longer a free parameter of nature: it is
determined to within $\sim5 of quantum fields at the cosmological horizon.

\appendix

=== Lattice Methods ===

-- Cubic lattice --

We work on a three-dimensional cubic lattice of $N^3$ sites with lattice spacing
$a = 1$. The free massless scalar field has Hamiltonian


H = \frac{1}{2}\sum_i \pi_i^2 + \frac{1}{2}\sum_{\langle ij \rangle} (\phi_i -
\phi_j)^2 ,

where the sum $\langle ij \rangle$ runs over nearest neighbours.

Mode decomposition. With Dirichlet boundary conditions, the mode functions are
$f_{k}(x) = (2/(N+1))^{3/2} \prod_{d=1}^{3} \sin(\pi k_d x_d / (N+1))$, with
frequencies $\omega_{k}^2 = \sum_{d=1}^{3} 2(1 - \cos(\pi k_d/(N+1)))$. The two-
point functions in the vacuum are:


X_{ij} = \sum_{k} \frac{f_{k}(i) f_{k}(j)}{2\omega_{k}} ,

P_{ij} = \sum_{k} \frac{\omega_{k} f_{k}(i) f_{k}(j)}{2} .

Subregion entropy. For a cubic subregion of side $L$, we restrict $X$ and $P$ to
the $L^3$ sites inside the cube. The symplectic eigenvalues are obtained from
$\nu_k = \sqrt{\lambda_k(X_{\rm sub} P_{\rm sub})}$, and the entropy follows
from ((ref: eq:symplectic)).

Null-space extraction of $\delta$. The design matrix for the four-parameter fit
((ref: eq:4param)) has columns $(6L^2, 12L, \ln L, 1)$ for each $L$ value. We
compute a vector $c$ in the left null space of the first, second, and fourth
columns (area, perimeter, constant), so that $c^T S = \delta \cdot c^T \ln L$.
This gives $\delta = c^T S / (c^T \ln L)$ without fitting $\alpha$, $\beta$, or
$\gamma$.

Lattice sizes $N = 22$, $28$, $36$, and $48$ were used, with subregion sizes $L
= 2, \ldots, N/2 - 1$.

-- Spherical decomposition --

Following Lohmayer et al. [ref: Lohmayer2009], the free scalar field in $3{+}1$
dimensions is expanded in spherical harmonics. Each $(l,m)$ sector reduces to a
1D radial chain with $N_{\rm radial}$ sites at positions $r_j = j a$ ($j = 1,
\ldots, N_{\rm radial}$; $a = 1$). Using canonical variables $q_j = j \phi_j$
and $p_j = \pi_j / j$, the coupling matrix is tridiagonal:


K'_l[j,j] &= \frac{(j-\tfrac{1}{2})^2 + (j+\tfrac{1}{2})^2 + l(l+1)}{j^2} + m^2
,

K'_l[j,j{+}1] &= -\frac{(j+\tfrac{1}{2})^2}{j (j+1)} ,

with Dirichlet boundary conditions ($\phi_0 = \phi_{N+1} = 0$).

Diagonalisation. We use scipy.linalg.eigh\_tridiagonal to obtain eigenvalues
$\omega_k^2$ and eigenvectors $V_{jk}$ in $O(N_{\rm radial}^2)$ time per
channel.

Covariance matrices. For the subregion $[1, \ldots, n]$ (a sphere of radius
$n$):


X_{jk} = \frac{1}{2}\sum_m \frac{V_{jm} V_{km}}{\omega_m} ,

P_{jk} = \frac{1}{2}\sum_m V_{jm} V_{km} \omega_m ,

restricted to $j,k \leq n$. Symplectic eigenvalues and entropy follow from
((ref: eq:symplectic)).

Summation over channels. The total entropy is $S_{\rm total}(n) =
\sum_{l=0}^{l_{\max}} (2l+1) S_l(n)$, with $l_{\max} = 10n$ (proportional
cutoff). To avoid storing all eigenvector matrices simultaneously, we process
one $l$-channel at a time, accumulating $S(n)$ incrementally. Memory cost is
$O(N_{\rm radial}^2)$ regardless of $l_{\max}$.

Third-difference extraction. We compute $S_{\rm total}(n)$ for consecutive $n =
n_{\min}{-}1, \ldots, n_{\max}{+}2$ and form $\Delta^3 S(n)$ as in ((ref:
eq:d3S)). A two-parameter fit $\Delta^3 S = A/n^3 + B/n^4$ gives $\delta = A/2$.
At $N_{\rm radial} = 1000$ with $n \in [25, 100]$, this achieves $1

=== References ===

[Planck2020] Planck Collaboration, N. Aghanim et al.,
"Planck 2018 results. VI. Cosmological parameters," Astron.\ Astrophys.\ 641, A6
(2020).
[Weinberg1989] S. Weinberg,
"The cosmological constant problem," Rev.\ Mod.\ Phys.\ 61, 1 (1989).
[Jacobson1995] T. Jacobson,
"Thermodynamics of spacetime: the Einstein equation of state," Phys.\ Rev.\
Lett.\ 75, 1260 (1995); gr-qc/9504004.
[Srednicki1993] M. Srednicki,
"Entropy and area," Phys.\ Rev.\ Lett.\ 71, 666 (1993); hep-th/9303048.
[Bombelli1986] L. Bombelli, R. K. Koul, J. Lee, and R. D. Sorkin,
"Quantum source of entropy for black holes," Phys.\ Rev.\ D\ 34, 373 (1986).
[Solodukhin2011] S. N. Solodukhin,
"Entanglement entropy of round spheres," Phys.\ Lett.\ B\ 693, 605 (2010);
arXiv:1008.4314.
[CasiniHuerta2012] H. Casini and M. Huerta,
"On the RG running of the entanglement entropy of a circle," Phys.\ Rev.\ D\ 85,
125016 (2012); arXiv:1202.5650.
[BianchiMyers2014] E. Bianchi and R. C. Myers,
"On the architecture of spacetime geometry," Class.\ Quant.\ Grav.\ 31, 214002
(2014); arXiv:1212.5183.
[CaiKim2005] R.-G. Cai and S. P. Kim,
"First law of thermodynamics and Friedmann equations of Friedmann-Robertson-
Walker universe," JHEP\ 02, 050 (2005); hep-th/0501055.
[Lohmayer2009] R. Lohmayer, H. Neuberger, A. Schwimmer, and S. Theisen,
"Numerical determination of entanglement entropy for a sphere," Phys.\ Lett.\ B\
685, 222 (2010); arXiv:0911.4283.
[Helmes2016] J. Helmes, L. E. Hayward Sierens, A. Chandran, W. Witczak-Krempa,
and R. G. Melko, "Universal corner entanglement of Dirac fermions and gapless
bosons from the corner transfer matrix renormalization group," Phys.\ Rev.\ B\
94, 125142 (2016); arXiv:1606.03096.
[BisognanoWichmann1975] J. J. Bisognano and E. H. Wichmann,
"On the duality condition for a Hermitian scalar field," J.\ Math.\ Phys.\ 16,
985 (1975).
[BCHM2013] J. Bhattacharya, M. Nozaki, T. Takayanagi, and T. Ugajin,
"Thermodynamical property of entanglement entropy for excited states," Phys.\
Rev.\ Lett.\ 110, 091602 (2013); arXiv:1212.1164.
[DvaliSolodukhin2008] G. Dvali and S. N. Solodukhin,
"Black hole entropy and gravity cutoff," arXiv:0806.3976 [hep-th] (2008).
[CKN1999] A. G. Cohen, D. B. Kaplan, and A. E. Nelson,
"Effective field theory, black holes, and the cosmological constant," Phys.\
Rev.\ Lett.\ 82, 4971 (1999); hep-th/9803132.
[Padmanabhan2012] T. Padmanabhan,
"Emergent perspective of gravity and dark energy," Res.\ Astron.\ Astrophys.\
12, 891 (2012); arXiv:1207.0505.
[Verlinde2011] E. Verlinde,
"On the origin of gravity and the laws of Newton," JHEP\ 04, 029 (2011);
arXiv:1001.0785.
[EGJ2006] C. Eling, R. Guedens, and T. Jacobson,
"Non-equilibrium thermodynamics of spacetime," Phys.\ Rev.\ Lett.\ 96, 121301
(2006); gr-qc/0602001.
[ChircoLiberati2010] G. Chirco and S. Liberati,
"Nonequilibrium thermodynamics of spacetime: the role of gravitational
dissipation," Phys.\ Rev.\ D\ 81, 024016 (2010); arXiv:0909.4194.
[Li2004] M. Li,
"A model of holographic dark energy," Phys.\ Lett.\ B\ 603, 1 (2004); hep-
th/0403127.
[CaiCaoHu2008] R.-G. Cai, L.-M. Cao, and Y.-P. Hu,
"Corrected entropy-area relation and modified Friedmann equations," JHEP\ 08,
090 (2008); arXiv:0807.1232.
[Lidsey2009] J. E. Lidsey,
"Thermodynamics of anomaly-driven cosmology," Class.\ Quant.\ Grav.\ 26, 147001
(2009); arXiv:0812.2791.
[Sheykhi2010] A. Sheykhi,
"Thermodynamics of apparent horizon and modified Friedmann equations," Eur.\
Phys.\ J.\ C\ 69, 265 (2010); arXiv:1012.0383.
[BenedettiCasini2020] D. Benedetti and H. Casini,
"Entanglement entropy of linearized gravitons in a sphere," Phys.\ Rev.\ D\ 101,
045004 (2020); arXiv:1908.05763.
[ChristensenDuff1978] S. M. Christensen and M. J. Duff,
"Quantizing gravity with a cosmological constant," Nucl.\ Phys.\ B\ 170, 480
(1980).
[BlommaertColinEllerin2025] A. Blommaert and S. Colin-Ellerin,
"Gravitons on the edge," JHEP\ 03, 116 (2025), arXiv:2405.12276.