⚛️ quantum physics

Algorithmic Randomness and Physical Typicality

This paper proposes using algorithmic randomness to precisely define physical typicality, demonstrating how this framework allows the distribution postulate in Bohmian mechanics to be formulated as a statistical law that guarantees standard Born statistics for computable experimental protocols.

Original authors: Jeffrey A. Barrett, Eddy Keming Chen, Josiah Lopez-Wild

Published 2026-09-09

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Jeffrey A. Barrett, Eddy Keming Chen, Josiah Lopez-Wild

Original paper licensed under CC BY 4.0 (http://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

In the grand architecture of physics, there is a quiet but persistent question about how the universe begins. Many theories describe how things move and change once they are already in motion, but they often stumble when asked to explain the very first state of the system. To bridge this gap, physicists frequently rely on a concept called typicality. Imagine a vast landscape of every possible way a system could start. Most of these starting points look very similar to one another, while a tiny, scattered few look strange or special. The idea of typicality suggests that our actual universe is one of the common, ordinary starting points, not a rare oddity. This assumption allows scientists to predict that experiments will yield standard results, like the familiar patterns seen in quantum mechanics. However, a nagging problem remains: in the strict mathematics used to define these landscapes, almost every single point is technically unique. If you zoom in close enough, every starting position is an outlier in some way, making the definition of "typical" feel vague and mathematically slippery. Without a precise definition, it is hard to say exactly what physical law is doing the work of selecting our universe's starting point.

A new paper by Jeffrey Barrett, Eddy Keming Chen, and Josiah Lopez-Wild tackles this vagueness by borrowing a tool from computer science known as algorithmic randomness. Instead of asking if a starting point is statistically common in a broad sense, they ask if it is "special" in a way a computer could describe. They propose that a physical state is truly typical only if it passes every possible test that a computer could run to find a pattern. If a computer cannot find a rule that singles out a specific starting position as unusual, then that position is considered random and typical. The authors apply this idea to Bohmian mechanics, a deterministic version of quantum theory where particles have definite paths. In this theory, there is a rule called the distribution postulate, which states that particles start out in a specific statistical arrangement. The authors argue that this rule should not be seen as a vague suggestion or a statement about our ignorance, but as a hard law of nature. This law simply states that the universe's initial configuration is one that no computer could ever describe as an exception.

To test this idea, the researchers built a simplified model involving a sequence of particles and measurements. They imagined a scenario where particles are prepared and their spins are measured one after another. In standard physics, we expect the results to follow a specific pattern, with roughly half the measurements showing one result and half showing the other, just like flipping a fair coin. The authors showed that if the particles start in a position that is algorithmically random, the sequence of measurement results will inevitably follow this pattern. Crucially, this is not just a matter of high probability. In their framework, if the starting point is algorithmically random, the correct pattern is guaranteed to happen. There is no tiny chance that the universe could have started in a weird spot that breaks the rules, because the law of nature explicitly forbids those weird spots from existing.

The paper also addresses a clever counter-argument. One might imagine a mischievous observer who knows the exact position of every particle and could choose measurement directions to force a specific outcome, breaking the expected pattern. The authors show that such an observer would need to know information that is fundamentally uncomputable. Since any real experiment we can actually perform must be describable by a finite set of instructions, it is limited to computable procedures. For every experiment that can actually be carried out by a human or a machine, the theory guarantees the standard results. The only way to break the pattern would be to use a measurement sequence that cannot be described or repeated, which falls outside the realm of physical science.

By reformulating the starting conditions of the universe in this way, the authors turn a fuzzy statistical guess into a sharp, objective constraint. They argue that the initial state of the universe is not just "likely" to be a certain way, but is physically impossible to be any other way if it is to be considered a valid physical reality. This approach separates the objective facts of the universe from our subjective uncertainty. The law of nature dictates the starting point, and the deterministic laws of motion take over from there. The result is a clearer picture of how a universe governed by strict rules can still produce the random-looking statistics we see in our laboratories. The work suggests that the randomness we observe in nature is not a sign of chaos or a lack of knowledge, but a precise signature of a universe that is typical in the deepest, most computable sense possible.

Technical Summary: Algorithmic Randomness and Physical Typicality

Problem Statement
The paper addresses a foundational ambiguity in physics regarding the concept of "typicality." While physical theories often assert that the actual world is typical relative to a probability measure (e.g., the Born rule in quantum mechanics), standard measure theory characterizes the size of sets rather than the properties of individual points. In continuous spaces, every specific world lies within some measure-zero set; thus, defining a typical world as one outside every measure-zero set renders no world typical. Consequently, if a physical law merely asserts that the actual world is "typical" in the standard measure-theoretic sense, it fails to provide a precise constraint on the actual state of the world. This is particularly problematic for the Bohmian distribution postulate, which stipulates that the initial particle configuration $Q_0$ satisfies $\rho(Q, t_0) = |\Psi(Q, t_0)|^2$ . Without a precise definition of typicality, the status of this postulate remains unclear: is it a definition, a norm of rational credence, or an independent law of nature?

Methodology
The authors propose characterizing physical typicality using the theory of algorithmic randomness, specifically Martin-Löf randomness.

Conceptual Framework: A state is considered typical relative to a computable measure if it is not "special" in any effectively describable way. Formally, a point is Martin-Löf random relative to a measure $\mu$ if it passes all uniformly effective statistical tests (i.e., it belongs to no effectively describable $\mu$ -null set).
Toy Model Construction: The authors first illustrate the concept using a simplified model involving infinite sequences of spin-1/2 particles. They demonstrate how a Martin-Löf random binary sequence can determine initial particle positions (top vs. bottom of a wave packet) to guarantee Born statistics for specific measurement directions.
Generalization to Configuration Space: The model is extended to the full configuration space of Bohmian mechanics ( $X = \mathbb{R}^{3N}$ ). The authors assume the universal wave function $\Psi$ is computable and that the configuration space equipped with the Born measure $\mu_\Psi$ forms a computable probability space.
Handling Measurement Constraints: To address the possibility of "sneaky" measurement sequences (analogous to Maxwell's demon) that could exploit exact initial positions to violate randomness, the authors restrict the scope of guaranteed statistics to uniformly layerwise computable sequences of measurements. This concept, developed by Hoyrup and Rojas, serves as an effective analogue of measurability, ensuring that the mapping from initial configurations to outcomes preserves Martin-Löf randomness.

Key Contributions

Algorithmic Bohmian Mechanics (aBM): The paper formulates a new version of Bohmian mechanics where the distribution postulate is replaced by an Algorithmic Distribution Postulate. This postulate is a constraining law stating that the actual initial configuration $Q_0$ must be Martin-Löf random with respect to the Born measure $\mu_\Psi$ ( $Q_0 \in \text{MLR}_{\mu_\Psi}$ ).
Precise Law Formulation: Unlike standard presentations where the distribution postulate is a vague statistical stipulation, the aBM postulate is a sharp, objective constraint that excludes a precisely defined set of physically impossible worlds (the effectively null sets).
Preservation of Randomness: The authors establish that if the initial configuration is Martin-Löf random, then the sequence of outcomes generated by any uniformly layerwise computable sequence of measurements is also Martin-Löf random with respect to the induced push-forward measure.

Results
The central technical result is Theorem 1:

Let $(X, \mu_\Psi)$ be the configuration space with the Born measure.
Let $\{f_n\}$ be a uniformly layerwise computable sequence of functions representing measurements.
Let $\mu_\omega$ be the induced joint measure on outcomes (which yields Born statistics).
Result: If the initial configuration $Q_0$ is $\mu_\Psi$ -Martin-Löf random, then the outcome sequence $(f_1(Q_0), f_2(Q_0), \dots)$ is $\mu_\omega$ -Martin-Löf random.

Corollary 1 applies this to independent and identically distributed (iid) binary measurements, guaranteeing that the limiting relative frequency of outcomes converges to the Born expectation with certainty (not merely with probability 1). This holds for every infinite computable subsequence of measurements.

Significance and Claims
The paper claims that this approach provides a precise content for statistical laws in physics:

Empirical Guarantee: By ruling out effectively null sets rather than just measure-zero sets, aBM provides a stronger empirical guarantee than standard typicality arguments. The Born statistics hold with certainty for all computable experimental protocols.
Objective Law vs. Epistemic Probability: The framework separates the objective content of the law (the algorithmic constraint on initial conditions plus deterministic dynamics) from the epistemic role of probability. Probabilities arise as rational credences for agents with exchangeable beliefs, derived via de Finetti's representation theorem from the guaranteed limiting frequencies.
Scope of Experimentation: The restriction to computable measurements is not merely technical but reflects a substantive hypothesis: only experiments that can be finitely specified, implemented, and repeated are physically realizable. "Sneaky" non-computable experiments, which could theoretically defeat the distribution postulate, are excluded from the scope of physical possibility.
Generalizability: While demonstrated via Bohmian mechanics, the authors suggest this approach can be applied to any physical theory requiring a statistical boundary condition, offering a way to formulate such conditions as algorithmic constraints.

The paper concludes that algorithmic randomness transforms the Bohmian distribution postulate from a vague statistical posit into a precise, constraining law of nature that specifies the initial condition of the universe.

Technical Summary: Algorithmic Randomness and Physical Typicality

More like this