🔬 physics

Finite Local Information Capacity in Continuum Dynamics

This paper establishes a regulator-resolved framework that distinguishes finite local information capacity from spatial lattice spacing and coarse-graining scales, deriving operational error bounds and an effective sigma-model action, and validating the approach through numerical tests on a Klein–Gordon chain to provide a controlled foundation for finite-information models without asserting a unique theory of quantum gravity.

Original authors: Florian Neukart, Eike Marx

Published 2026-10-26

📖 7 min read🧠 Deep dive

CC BY 4.0

Original authors: Florian Neukart, Eike Marx

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Nature, as we understand it through our most precise theories, is often described as a smooth, continuous fabric. Space and time flow without interruption, and the fields that fill the universe can vibrate at any frequency, from the slowest rumble to the highest, most energetic hum. This picture of a seamless continuum works incredibly well for predicting how particles move and how stars shine. However, when scientists try to build a theory that unifies gravity with quantum mechanics, or when they attempt to simulate these complex systems on a computer, they run into a fundamental problem: the idea of infinite smoothness requires an infinite amount of information to describe. A truly continuous space would need an infinite number of tiny bits to define every possible point and vibration. Since the universe we observe is finite, and our computers are certainly finite, many researchers have wondered if space itself might be made of a finite number of information-carrying units, like a vast grid of tiny cells, each holding a limited amount of data.

The question of whether the universe has a finite information capacity is not just a mathematical curiosity; it touches on the deepest mysteries of black holes and the origin of the cosmos. If space is made of these finite cells, it would mean there is a limit to how much information can be packed into any region. But a major difficulty has always been distinguishing between three very different things that look similar on paper. First, there is the idea of a physical grid, where space is literally chopped into small squares with a specific size. Second, there is the idea of a limit on how many states a single point can hold, like a switch that can only be in a few positions rather than a dial that can turn to any angle. Third, there is the idea of a "blur" we apply when we look at the big picture, where we ignore tiny details to see the general flow. For a long time, these three concepts were often mixed together, leading to confusion about what a theory of finite information actually predicts.

In a new study, researchers Florian Neukart and Eike Marx have untangled these three ideas to create a clear, operational framework for testing finite information capacity. They argue that a finite limit on local information is not the same thing as a grid spacing, nor is it the same as simply ignoring small details. To prove this, they built a precise mathematical model that treats these three factors as independent variables. They did not just propose a theory; they constructed a specific, solvable quantum system—a chain of four connected points that behaves like a vibrating string—and ran exact simulations to see how the system behaves when you change the local information limit while keeping the grid size fixed, and vice versa. Their work shows that you can increase the amount of information a single point can hold without changing the size of the grid, and you can change the grid size without changing the information limit. Crucially, they found that the errors caused by having too little information at a single point are completely different from the errors caused by the grid itself.

The researchers defined a "finite-information deviation" not by subtracting one equation from another, but by comparing two different ways of evolving a system. Imagine you have a complex, detailed state of a system. One way to predict its future is to let the detailed system evolve step-by-step and then simplify the result. The other way is to simplify the system first and then let that simplified version evolve. If the universe were perfectly continuous and our simplified models were perfect, both methods would yield the exact same result. The difference between the two outcomes is the "deviation." Neukart and Marx showed that this difference can be broken down into three distinct parts: the error from not having enough local information, the error from the grid spacing, and the error from how we choose to simplify the view. By separating these, they could measure exactly how much each factor contributed to the total error.

Using their four-point chain model, the team tested how well a system with a limited number of internal states could mimic a continuous system. They started with a very small limit on the number of states each point could hold and gradually increased it. They found that as they increased the local information capacity, the errors dropped very rapidly. For example, when they increased the number of allowed states from four to twelve, the error in predicting the system's energy and behavior shrank by a factor of nearly a billion. This rapid convergence proved that the local information limit was no longer the bottleneck; the system was now limited only by the grid spacing itself. Even with a huge amount of local information, the system still showed a distinct difference from a truly continuous space because the grid spacing remained fixed. This confirmed that the "pixelation" of space and the "memory limit" of a single point are two separate physical effects that must be studied independently.

The study also clarified a common misunderstanding about entropy, which is a measure of disorder or information content. In many theories, scientists try to describe the universe using a single field that represents the entropy at every point. However, the researchers demonstrated that the total entropy of a group of connected points is not simply the sum of the entropy of each individual point. The difference between the whole and the sum of its parts is a measure of the correlations, or connections, between the points. If you ignore these connections and just add up the local entropies, you lose a significant amount of physical information. In their simulations, they found that even in a simple, smooth system, the connections between neighboring points carried a substantial amount of information that could not be captured by looking at the points in isolation. This means that a theory of finite information cannot rely on a single "entropy field" to describe the universe; it must account for the complex web of relationships between different regions.

Furthermore, the researchers showed that when you try to build a theory of how these finite information cells move and interact, the result is generally a complex, multi-dimensional description rather than a simple, one-dimensional wave. The geometry of the possible states is curved and intricate, much like a landscape with many hills and valleys, rather than a flat plain. Only under very specific and restrictive conditions does this complex landscape collapse into a single, simple field. This finding challenges the idea that a theory of finite information will naturally simplify into a single, easy-to-handle equation. Instead, it suggests that the fundamental description of space-time, if it is finite, is inherently complex and requires a multi-faceted approach to describe the collective behavior of its parts.

The work provides a rigorous, step-by-step method for testing any future theory that claims space is made of finite information. It establishes that such a theory must clearly define its local information limits, its grid spacing, and its method of simplification, and it must show how these three elements scale together to recover the smooth physics we observe. The researchers did not claim to have found the final theory of space-time, nor did they prove that the universe is definitely a grid of finite cells. Instead, they provided the tools and the rules to test such a hypothesis properly. They showed that if a theory is to be valid, it must be able to reproduce the correct behavior of particles and fields without confusing the errors of its own construction with the actual physics of the universe. By separating the different sources of error and proving that they behave differently, this study offers a clear path forward for anyone trying to understand the fundamental, information-based structure of reality.

Technical Summary: Finite Local Information Capacity in Continuum Dynamics

Problem Statement

Continuum quantum field theory (QFT) provides high-precision descriptions of nature but relies on infinite local degrees of freedom. Practical calculations and quantum simulations require regulators, such as spatial lattices, finite volumes, or local basis truncations. A significant conceptual difficulty arises when proposals suggest that finite information capacity is a fundamental property of space-time (e.g., via the Quantum Memory Matrix program or holographic bounds).

Current frameworks often conflate three logically distinct concepts:

Local Hilbert-space dimension ( $d$ ): Limits the number of locally distinguishable states.
Spatial lattice spacing ( $a$ ): Limits spatial resolution and modifies the momentum spectrum.
Coarse-graining scale ( $\ell$ ): Determines which microscopic distinctions are discarded in macroscopic observables.

Furthermore, operational comparisons between microscopic and effective theories are often ill-defined, attempting to subtract generators acting on different domains. Additionally, the relationship between block entropy, local entropy averages, and correlation measures is frequently misunderstood, leading to the erroneous assumption that a single entropy field can universally parameterize local physics.

Methodology

The authors propose a regulator-resolved, operational framework that strictly separates these three regulators ( $a, d, \ell$ ) and defines finite-information deviations without subtracting operators from different spaces.

1. Operational Definition of Deviation

The framework defines the finite-information deviation as the failure of a commutative diagram involving:

Microscopic evolution ( $\Phi^t_{a,d}$ )
A coarse-graining channel ( $C_{a,d,\ell}$ )
Effective evolution ( $\Phi^t_{\text{eff}}$ )

The deviation is quantified as the trace distance (or diamond norm) between two states in a common effective state space:
$\delta_{a,d,\ell}(t; \rho) = \frac{1}{2} \| C_{a,d,\ell} \circ \Phi^t_{a,d}(\rho) - \Phi^t_{\text{eff}} \circ C_{a,d,\ell}(\rho) \|_1$
This measures the difference between "evolve microscopically then coarse-grain" versus "coarse-grain then evolve effectively."

2. Regulator Decomposition

Using the triangle inequality, the total discrepancy is decomposed into three independent contributions:

Capacity Error ( $\delta_{\text{cap}}$ ): Difference between finite- $d$ and infinite- $d$ (but fixed $a$ ) lattice states.
Lattice Error ( $\delta_{\text{lat}}$ ): Difference between infinite- $d$ lattice and continuum states.
Coarse-graining Error ( $\delta_{\text{cg}}$ ): Difference between the continuum state and the effective theory prediction.

3. Theoretical Derivations

Finite-Time Capacity Bounds: The authors derive a bound for capacity error controlled by the discarded state weight ( $\varepsilon_d$ ) and Hamiltonian leakage ( $\|Q_d H P_d\|$ ) across the cutoff boundary. This demonstrates that error scaling is state-dependent and not a universal function of $1/d$ .
Entropy and Correlations: The paper establishes that the difference between block entropy and the average of single-site entropies is exactly the total correlation (measured by relative entropy). Thus, replacing block entropy with local averages discards physical correlations.
Information-Geometric Effective Action: By expanding the relative entropy between neighboring reduced states, the authors derive an effective action. The generic result is a nonlinear sigma model on a manifold of collective coordinates. A single scalar entropy field emerges only under a restrictive one-dimensional closure condition.

4. Numerical Realization

To validate the framework, the authors implement an exactly evolved periodic Klein–Gordon chain with $N=4$ sites.

Variables: Lattice spacing $a=1$ , mass $m=0.7$ , and local number-state cutoff $d \in \{4, \dots, 14\}$ .
Protocol: Ground-state energies, excitation frequencies (extracted from spectral weights), and real-time reduced density matrices are computed.
Reference: The $d=14$ calculation serves as a high-precision numerical reference, validated against analytic lattice dispersion relations.

Key Results

1. Separation of Regulators

Numerical results confirm that local capacity errors and spatial discretization errors are distinct.

As $d$ increases, ground-state energy and frequency errors converge rapidly (e.g., relative error drops from $3 \times 10^{-3}$ at $d=4$ to $10^{-11}$ at $d=12$ ).
However, even at converged $d$ , a significant discrepancy remains between the lattice dispersion and the continuum dispersion at high momenta ( $k \approx \pi/a$ ). This confirms that increasing local dimension does not remove lattice artifacts; the limits $d \to \infty$ and $a \to 0$ do not commute.

2. State-Dependent Convergence

The capacity error is shown to be highly dependent on the state sector.

For coherent states with higher amplitude ( $\alpha=1.00$ ), the discarded weight and subsequent leakage are significantly larger than for lower amplitudes ( $\alpha=0.55$ ).
The trace distance deviation is non-monotonic in time but its envelope decreases systematically with $d$ .
The Audenaert continuity bound successfully certifies that convergence in trace distance implies convergence in von Neumann entropy.

3. Entropy vs. Correlation

In the converged ground state, the mutual information between adjacent sites is non-zero ( $I \approx 0.154$ ). The block entropy is strictly less than the sum of local entropies. This numerically validates the proposition that averaging local entropies erases correlation physics.

4. Effective Action Structure

The derivation shows that the leading-order effective action for coarse-grained states is a nonlinear sigma model with a target-space metric derived from the Bogoliubov–Kubo–Mori (BKM) metric. A single entropy field is not an automatic consequence of finite capacity but requires a specific closure condition where the accessible state manifold is one-dimensional.

Significance and Claims

The paper claims to provide a controlled mathematical and numerical foundation for finite-information models without asserting a unique microscopic theory of quantum gravity. Its primary contributions are:

Operational Clarity: It replaces phenomenological ansätze with a rigorous framework where finite-information deviation is defined as a non-commutativity in a common state space, separating capacity, lattice, and coarse-graining errors.
Rejection of Universal Scaling: It demonstrates that there is no universal inverse- $d$ expansion for error; convergence depends on the state sector, the Hamiltonian, and the encoding.
Correction of Entropy Assumptions: It clarifies that block entropy and local entropy averages are not interchangeable and that a single entropy field is a special case, not a generic feature of finite-dimensional systems.
Falsifiability: It outlines a development path for finite-information space-time theories where every step (microscopic model, matching map, scaling trajectory, effective coefficients) must be specified and can fail independently.

The authors emphasize that their results do not prove space-time is a lattice of finite-dimensional cells. Instead, they establish the consistency requirements any such proposal must meet: identifying the local Hilbert space, the matching map, the scaling trajectory, and the collective variables that close the low-energy dynamics. The work serves as a benchmark for distinguishing genuine continuum limits from mere basis convergence tests.