The Distribution Postulate in Algorithmic Bohmian Mechanics

This paper proposes formulating the distribution postulate in Bohmian mechanics as an objective constraining law using algorithmic randomness, thereby guaranteeing standard Born statistics for canonical quantum experiments and clarifying the nature of quantum probabilities within a deterministic framework.

The Gist

Imagine the universe as a giant, perfectly choreographed dance. In a theory called Bohmian Mechanics, every particle has a specific spot and a specific path it follows, just like a dancer in a play. There is no randomness in the dance itself; if you knew exactly where every dancer started and how the music (the "wave function") moves, you could predict every single step they will ever take.

But here's the problem: We don't know where the dancers started. To make the theory match what we see in the real world (like how often a coin lands on heads or tails), physicists have to assume a special rule at the very beginning of time. This rule is called the Distribution Postulate.

Traditionally, this rule is a bit vague. It's like saying, "God rolled a magical die to decide where the dancers started, and the die was weighted so that the results match the famous 'Born Rule' of quantum physics." But what does "God rolling a die" really mean? Is it a real physical law, or just a guess?

This paper proposes a new, sharper way to understand that starting rule using a branch of math called Algorithmic Randomness. Here is the breakdown of their idea:

1. The Problem with "Randomness"

In our daily lives, we think of a random sequence (like a string of coin flips) as one where you can't find a pattern. If you see Heads, Heads, Heads, Heads... a million times, that's not random. If you see a mix that looks messy and unpredictable, that's random.

But in math, "random" is tricky. A sequence can look random but still have a hidden pattern that a super-computer could eventually find. The authors want a definition of randomness that is objective and unbreakable. They use a concept called Martin-Löf randomness.

Think of Martin-Löf randomness as a "Gold Standard" for chaos. A sequence is Martin-Löf random if it passes every possible test for patterns that a computer could ever run. It's not just "looks messy"; it is mathematically impossible to compress or predict. It is the ultimate definition of "no pattern."

2. The New Rule: The Algorithmic Distribution Postulate

The authors suggest replacing the vague idea of "God rolling a die" with a strict mathematical law:

The Initial State of the Universe is Martin-Löf Random.

Instead of saying "the particles are placed with a probability of 50/50," they say: "The starting position of every particle is a point that looks completely random to any computer algorithm, relative to the quantum wave function."

The Analogy:
Imagine a giant, foggy map of a city (the "configuration space"). The fog is thickest in some areas and thinnest in others (this represents the quantum wave function, $|\psi|^2$).

• Old View: We say, "Pick a spot in the fog, but make sure you pick it according to the thickness of the fog."
• New View (aBM): We say, "Pick a spot that is algorithmically random relative to the fog." This means the spot you picked doesn't follow any hidden, computable pattern. It is a spot that a computer could never have guessed or described in advance, even though it fits the general shape of the fog.

3. Why This Matters: Certainty vs. Probability

In standard quantum mechanics, we usually say, "There is a 50% chance you will see a 'Heads' result." It's a guess.

In this new framework (Algorithmic Bohmian Mechanics or aBM), the result is much stronger. Because the starting point is guaranteed to be Martin-Löf random, the authors prove that:

• If you run a long series of experiments (like measuring the spin of electrons), the results will not just likely match the 50/50 rule.
• They will definitely match the 50/50 rule in the long run.

It's the difference between saying, "I bet you'll get heads half the time," and saying, "The math guarantees that if you flip the coin enough times, the pattern of heads and tails will be perfectly, objectively random."

4. The "Computable" Catch

The paper adds one important condition: This guarantee works for measurements that are computable.

• Analogy: Imagine a machine that measures the dancers. If the machine's instructions are something a computer could write down (a "computable" measurement), then the results will be perfectly random and follow the standard rules.
• The authors show that for any standard quantum experiment we can actually build (which are all computable), this new rule works perfectly.

5. What About "God" and "Chance"?

The paper argues that we don't need to imagine a deity rolling dice. Instead, the "chance" we see in the universe is actually a reflection of the complexity of the initial state.

• The universe started in a state so complex and patternless (Martin-Löf random) that it looks like it was determined by chance.
• This turns the "Distribution Postulate" from a vague suggestion into a hard, objective law of nature: "The universe began in a state that passes every possible test for randomness."

Summary

The authors have taken a fuzzy rule in quantum physics ("the particles start in a random spot") and sharpened it into a precise mathematical definition ("the particles start in a spot that is algorithmically random").

By doing this, they show that if the universe started this way, the results of our experiments are guaranteed to follow the standard rules of quantum mechanics. It replaces "probability" (a guess about what might happen) with "typicality" (a guarantee that what happens is the most mathematically normal outcome for a random start).

In short: The universe isn't rolling dice; it started with a starting line so perfectly chaotic that the results must look like a fair game.

Technical Summary

Technical Summary: The Distribution Postulate in Algorithmic Bohmian Mechanics

Problem Statement
Bohmian mechanics (BM) is a deterministic theory that reproduces the empirical predictions of standard quantum mechanics regarding particle positions. However, to generate the correct forward-looking probabilities (the Born rule), BM requires a special statistical boundary condition known as the distribution postulate: at an initial time $t_0$, the probability density for the particle configuration $Q$ must be $\rho(Q, t_0) = |\psi(Q, t_0)|^2$.

The central problem addressed by Barrett, Chen, and Lopez-Wild is the ontological and epistemological status of this postulate. In standard BM, the distribution is often treated as an epistemic probability reflecting ignorance, or as an objective physical law stipulating a "genuinely probabilistic procedure" (e.g., a divine randomizer) for setting initial conditions. The authors argue that the nature of this "objective chance" remains unclear. Specifically, it is ambiguous how to formulate the postulate as a precise, objective constraining law that guarantees the Born statistics without relying on vague notions of "typicality" or unexplained stochastic processes.

Methodology
The authors propose Algorithmic Bohmian Mechanics (aBM), a reformulation of BM that utilizes the theory of algorithmic randomness, specifically Martin-Löf (ML) randomness, to define the distribution postulate.

1. Algorithmic Randomness Framework: The paper adopts the definition of ML-randomness, where a sequence (or point) is considered random if it passes all effective statistical tests (Martin-Löf tests) relative to a given measure. A sequence is ML-random if it is not contained in any effectively describable null set (a set of measure zero defined by a computable sequence of open sets).
2. Computable Probability Spaces: The authors extend ML-randomness from binary sequences to the configuration space of quantum systems. They utilize the framework of computable Polish spaces and computable probability measures. The Born measure $\mu_\psi$ (induced by the wave function $|\psi|^2$) is treated as a computable measure on the configuration space $X$.
3. Layerwise Computability: To handle the dynamics of measurement, the authors introduce layerwise computable functions. A function is layerwise computable if it is computable on each "layer" of ML-random points (defined by their "randomness deficiency"). This allows the authors to represent sequences of measurements as computable maps that preserve randomness properties.
4. The Algorithmic Distribution Postulate: Instead of positing a stochastic process, the authors define the initial condition as an objective constraint: the actual initial configuration $x_0$ must be an element of the set of ML-random points relative to the Born measure $\mu_\psi$ (denoted $MLR_{\mu_\psi}$).

Key Contributions and Results

• Formulation of the Distribution Postulate: The paper provides a precise, objective definition of the distribution postulate: The actual initial configuration $x_0$ is Martin-Löf random with respect to the Born measure $\mu_\psi$. This replaces the vague notion of "typicality" with a sharp mathematical condition: $x_0$ lies outside every effectively describable $\mu_\psi$-null set.
• Preservation of Randomness: The authors prove that if the initial configuration is ML-random relative to $\mu_\psi$, then the sequence of measurement outcomes generated by a layerwise computable sequence of measurements is ML-random relative to the push-forward measure.
  • Theorem 2: If $x$ is $\mu_\psi$-ML-random and $\{f_n\}$ is a uniformly computable sequence of measurements, the outcome sequence $(f_1(x), f_2(x), \dots)$ is ML-random with respect to the induced joint measure $\mu_\omega$.
• Guarantee of Born Statistics: Unlike standard BM, which typically predicts Born statistics only with probability 1 (allowing for non-typical trajectories in the measure-zero set), aBM guarantees the standard Born statistics for all layerwise computable experiments.
  • Corollary 1 & 2: For identically prepared systems or repeated measurements on a single system (e.g., alternating x- and y-spin), the limiting relative frequencies of outcomes are exactly the Born probabilities (e.g., 1/2 for spin up/down) with certainty, not just with high probability. The outcome sequences are ML-random, satisfying the Law of Large Numbers by definition.
• Resolution of Single-System Issues: The authors address a shortcoming of earlier "toy" constructions where initial conditions determined by a single random sequence might fail to produce correct statistics for different observables on the same system. By restricting measurements to layerwise computable functions, aBM ensures that the randomness of the initial configuration is preserved across different types of measurements (e.g., alternating spin directions), yielding correct statistics for any computable subsequence.

Significance and Claims
The paper claims that aBM offers a clearer understanding of quantum probabilities in a deterministic theory by grounding them in objective algorithmic constraints rather than epistemic ignorance or unexplained stochastic laws.

• Objective Chance: The distribution postulate is reinterpreted as an objective law of nature that stipulates the initial configuration exhibits no effectively describable regularities relative to the quantum measure.
• Stronger Predictions: The authors note that aBM makes statistical predictions (guarantees of ML-randomness) rather than merely probabilistic predictions. While standard BM predicts Born statistics with probability 1, aBM guarantees them for the actual world (assuming the initial configuration is ML-random).
• Epistemic Connection: The paper argues that if an agent's credences are exchangeable (the order of measurements does not matter) and they accept aBM, de Finetti's representation theorem compels them to assign standard single-shot Born credences to measurement outcomes. Thus, aBM recovers the standard quantum probabilistic predictions for agents with exchangeable beliefs.

The authors conclude that this approach provides a concrete example of how algorithmic randomness can aid in specifying the content of physical laws, replacing informal appeals to typicality with mathematically precise objective constraints. They acknowledge that further work is needed to analyze the preservation of ML-randomness relative to conditional measures for subsystems, but assert that for the experiments considered, the framework successfully guarantees standard quantum statistical predictions.