Solving Physics' 10³⁵ Discrepancy: Why the Initial State Matters

Paper K: The Constrained Feynman-Vernon Influence Functional from the Quantum-Geometric Duality Series

Two Answers, Thirty-Five Orders of Magnitude Apart

How fast does gravity destroy quantum superpositions?

Two well-established frameworks give radically different answers. Perturbative quantum field theory---the standard tool for computing quantum processes---predicts decoherence rates scaling as $G^2$. For a microgram particle in superposition across a millimeter, this gives a decoherence time of about $10^{26}$ years. The Diosi-Penrose hypothesis predicts rates scaling as $G^1$, giving about 1.6 nanoseconds for the same particle.

That is a factor of $10^{35}$---possibly the largest quantitative discrepancy between competing theoretical predictions in physics. This paper shows that the entire discrepancy traces to a single assumption about the initial quantum state.

The Hidden Assumption

The standard perturbative calculation uses the Feynman-Vernon influence functional, a powerful technique for computing how an environment affects a quantum system. You write down the total system (matter plus gravitational field), trace out the gravitational degrees of freedom, and obtain an effective description of how the matter behaves.

The critical step, usually stated without comment, is the choice of initial state. The standard treatment assumes:

$|\Psi(0)\rangle = |\psi_{matter}\rangle \otimes |0_{grav}\rangle$

Matter in some quantum state, gravitational field in its vacuum. A product state---matter and gravity uncorrelated.

This seems natural. It is how we set up calculations in quantum electrodynamics and every other quantum field theory. But for gravity, it harbors a fundamental inconsistency.

The Wheeler-DeWitt Constraint

General relativity is not like other theories. It has a constraint that other forces lack: the total Hamiltonian must vanish.

$\hat{H}_{total} |\Psi_{phys}\rangle = 0$

This is the Wheeler-DeWitt equation, the central equation of canonical quantum gravity. In linearized gravity, it reduces to a concrete requirement: the gravitational field configuration must be consistent with the matter configuration. Specifically, the Poisson equation

$\nabla^2 \Phi(\mathbf{x}) = 4\pi G \rho(\mathbf{x})$

must hold as an operator equation on the physical Hilbert space. A mass at position $\mathbf{x}_A$ must carry its Newtonian field. You cannot have a mass here and its gravitational field somewhere else. The product state $|\psi_{matter}\rangle \otimes |0_{grav}\rangle$ places a mass in superposition while leaving the gravitational field in its vacuum---violating this constraint.

The Constrained Initial State

Imposing the Wheeler-DeWitt constraint forces the initial state to be entangled:

$|\Psi_{phys}\rangle = \frac{1}{\sqrt{2}}\left(|L\rangle|\Phi_L\rangle + |R\rangle|\Phi_R\rangle\right)$

Each branch of the superposition carries its own coherent gravitational field state $|\Phi_A\rangle$, determined uniquely by the mass configuration in that branch. The left-displaced mass carries the left-displaced Newtonian field; the right-displaced mass carries the right-displaced field.

This is not a choice. It is a consequence of the constraint equations of general relativity applied to quantum states. A mass must dress itself with its own gravitational field.

From $G^2$ to $G^1$

This single change---replacing the product state with the constraint-entangled state---transforms the physics entirely.

In the standard (unconstrained) calculation, decoherence arises through the noise kernel mechanism. The gravitational field fluctuates, and different branches of the superposition experience different fluctuations. Each interaction vertex contributes $\sqrt{G}$ (the gravitational coupling), and two vertices are needed (emission and absorption of a graviton), giving $G^2$.

In the constrained calculation, the mechanism is completely different. Each branch already carries its own gravitational field from the start. Decoherence arises through coherent-state overlap: the two field configurations $|\Phi_L\rangle$ and $|\Phi_R\rangle$ have an overlap that decreases over time. No graviton propagator is needed---the field is locked to the matter by the constraint.

The counting changes:

• Coherent state amplitude: $\alpha \sim GM$ (one power of $\sqrt{G}$ from the coupling, but no propagator)
• Overlap: $|\langle\Phi_L|\Phi_R\rangle|^2 \sim \exp(-GM^2/\hbar c d)$

The result is $G^1$---one power of Newton's constant, not two.

The Decoherence Rate

The constrained influence functional gives a decoherence rate:

$\Gamma = C \times \frac{GM^2}{\hbar d} + O(G^2)$

where the coefficient $C$ lies in the range $[1/2, 2]$, with the best estimate $C = 1$---matching the Diosi master equation exactly.

For concrete numbers:

The linearized approximation used throughout is controlled by the parameter $GM/(c^2 d) \sim 10^{-38}$---extraordinarily well controlled.

Why Gravity Is Special

One might ask: does the same argument apply to electromagnetism? Gauss's law in QED also constrains the electric field to match the charge configuration, producing similar coherent-state dressing.

The answer reveals something deep about gravity. In QED, the constraint produces a Coulomb phase that is perfectly reversible. Time is a background parameter, and unitary evolution preserves the overlap between different field configurations indefinitely. No decoherence occurs.

In gravity, the Wheeler-DeWitt constraint $\hat{H}_{total} = 0$ eliminates background time entirely. Physical time must emerge from internal correlations within the quantum state itself---through what is called the Page-Wootters mechanism. The quantum uncertainty of this gravitational clock converts the coherent-state overlap into irreversible decoherence.

This is why gravity decoheres and electromagnetism does not: gravity is the clock, and a clock in superposition cannot keep coherent time for both branches.

What This Resolves

The result clarifies the relationship between competing predictions:

• Anastopoulos and Hu ($G^2$): Their calculation is correct for the unconstrained product state. If nature allowed arbitrary initial states, their rate would apply.
• Diosi ($G^1$): His noise kernel $1/|\mathbf{x} - \mathbf{y}|$ is not an independent postulate but follows from the Wheeler-DeWitt constraint.
• Penrose ($G^1$): His "incompatible time translations" correspond to distinct modular flows for the two coherent-state branches---a precise mathematical realization of his physical intuition.

The $10^{35}$ discrepancy is not a conflict between different physics. It is the difference between an unphysical initial state and the physical one.

Honest Uncertainties

The $O(1)$ coefficient is uncertain by about a factor of 2. The energy scale ($E_G = GM^2/d$) is established rigorously from the constrained influence functional; converting it to a rate ($\Gamma = E_G/\hbar$) requires additional input from the modular Hamiltonian identification and Page-Wootters mechanism, which is less rigorous.

All higher-order corrections ($G^2$ and beyond) have been checked and are suppressed by factors of at least $10^{-38}$ for laboratory parameters. The result is robust within linearized gravity.

The Experimental Question

Whether nature actually enforces the Wheeler-DeWitt constraint on the initial state is not a theoretical question. It is an experimental one.

If nature uses the constrained state, decoherence occurs at nanosecond timescales for microgram masses---testable by next-generation matter-wave interferometry experiments expected in the late 2020s and 2030s.

If nature somehow permits the unconstrained product state, decoherence is unobservably slow. The absence of gravitational decoherence at the $G^1$ level would imply that the constraint equations of general relativity do not apply to quantum initial states---a profound statement about the foundations of quantum gravity.

Either answer transforms our understanding of how gravity and quantum mechanics fit together.

This is Paper K of the Quantum-Geometric Duality series, resolving the $G^1$ versus $G^2$ discrepancy through the Wheeler-DeWitt constraint on the Feynman-Vernon influence functional.