⚛️ quantum physics

Generalised Probabilistic Theories

This paper introduces the Generalised Probabilistic Theories framework, which models physical states and transformations using real vectors and matrices within a convex structure, to explain classical and quantum theories while exploring phenomena like Bell non-locality, super-quantum "box world," and the role of phase space negativity in contextuality and tunnelling.

Original authors: Zujin Wen, Zihan Wang, Xiangjing Liu, Oscar Dahlsten

Published 2026-07-01

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Zujin Wen, Zihan Wang, Xiangjing Liu, Oscar Dahlsten

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

Imagine you are trying to understand the rules of a game. Usually, we learn the rules by playing two specific versions: Classical Mechanics (the game of everyday objects like coins and balls) and Quantum Mechanics (the game of tiny particles like electrons).

This paper introduces a massive, universal "Game Engine" called Generalised Probabilistic Theories (GPT). Think of GPT as a giant, empty box where you can build any possible game that follows the basic laws of probability. Inside this box, the "Classical Game" and the "Quantum Game" are just two specific rooms. But the box also contains rooms for games that don't exist in our universe yet, or perhaps never will.

Here is a breakdown of the paper's main ideas using simple analogies:

1. The Language of the Game: Vectors and Dots

In this framework, everything is translated into math that looks like lists of numbers (vectors).

States (The Setup): Imagine you are preparing a meal. In this theory, your "meal" (the state of the system) is a list of probabilities. For a coin, your list might be [Heads: 50%, Tails: 50%].
Measurements (The Tasting): When you check the result, you are essentially "dotting" your state list with a "measurement list." The result of this math operation gives you the probability of a specific outcome.
The Big Idea: The paper argues that both classical coins and quantum particles can be described this way. The difference isn't in the language (vectors), but in the shape of the allowed lists.

2. The Shape of Reality: The Coin vs. The Ball

The paper uses shapes to explain the difference between classical and quantum worlds.

The Classical Shape (The Square/Cube): Imagine a coin. It can be Heads, Tails, or a mix. If you plot all possible states of a classical system, they form a simple shape like a square or a cube. The corners are the pure states (Heads or Tails), and the middle is the mix.
- Key Feature: In this shape, every mixed state (like a 50/50 coin) has only one way to be built from pure states. It's like saying a smoothie is only made of apples and oranges. There is no other recipe.
The Quantum Shape (The Sphere/Ball): Now imagine a quantum particle (a qubit). Its states form a solid ball (the famous Bloch sphere).
- Key Feature: Inside this ball, a mixed state can be built from pure states in many different ways. It's like a smoothie that could be made of apples/oranges, or pears/bananas, or grapes/strawberries, all resulting in the exact same taste. This "non-unique decomposition" is a hallmark of quantum weirdness.

3. Reconstructing the Rules: Why is the World Quantum?

The authors ask: "Why is our universe a ball and not a cube? What rules force the shape to be a sphere?"
They use four simple "axioms" (basic rules) to rebuild quantum theory from scratch:

Simplicity: The world should be as simple as possible.
Subspaces: If you look at a small part of the system, it should behave like a smaller version of the whole.
Combinations: If you combine two systems, the complexity should multiply (like how two coins have 4 outcomes, not just 2).
Continuity: You should be able to smoothly turn one state into another without "jumping" or breaking the rules.

The Result: When you apply these rules, the math forces the shape to be a sphere (Quantum Mechanics). If you remove the "Continuity" rule, you get the cube (Classical Mechanics). This proves that the "weirdness" of quantum mechanics (like superposition) is a necessary consequence of wanting a smooth, continuous world.

4. The "Box World": What If We Break the Rules?

The paper explores a hypothetical universe called Box World.

The PR Box: Imagine a magic box with two people, Alice and Bob. They each press a button (0 or 1) and get a light (0 or 1).
- In our world (Quantum), their lights are correlated in a specific way, but there's a limit (Tsirelson's bound) to how strong that link can be.
- In Box World, the correlation is super-strong. They can coordinate their lights perfectly to break the limits of quantum mechanics, yet still obey the rule that they can't send secret messages faster than light (No-Signalling).
The Catch: While this "Box World" allows for super-powerful correlations, the paper finds a downside. In this world, you can't have smooth, continuous transformations. The "dynamics" (how things change over time) are very rigid and discrete. You can't smoothly rotate a state; you can only snap it to new positions. This suggests that to have a rich, continuous universe like ours, we might have to give up some of that super-powerful correlation.

5. The Ghost in the Machine: Negativity and Tunneling

Finally, the paper looks at Hamiltonian Mechanics (how energy drives motion) and Phase Space (a map of position and momentum).

The Wigner Function: In quantum mechanics, we use a special map called the Wigner function to visualize particles. Unlike a normal map where probabilities are always positive (you can't have -50% chance of rain), the Wigner map can have negative numbers.
The Metaphor: Think of these negative numbers as "anti-probability" or "ghosts." They cancel out real probabilities in some places and boost them in others.
Tunneling: This negativity explains Quantum Tunneling (how a particle can pass through a wall it shouldn't be able to cross). The paper shows that tunneling happens because the "ghosts" (negativity) allow the particle to "borrow" probability from forbidden regions.
Contextuality: The paper links this negativity to Contextuality. This is the idea that the answer to a question depends on how you ask it. The paper argues that if you try to explain quantum mechanics using a standard, positive-probability map (like a classical one), you fail. The "negativity" is the mathematical price we pay for the universe being contextual.

Summary

This paper is a tour of the "zoo" of possible physical theories.

It shows us that Classical and Quantum theories are just two specific shapes in a vast landscape.
It proves that Quantum Mechanics is the unique shape that allows for smooth, continuous changes while combining systems in a specific way.
It explores Box World, a place with stronger correlations than ours, but where time and motion are rigid and broken.
It identifies Negativity (negative probabilities) as the secret ingredient that allows quantum particles to tunnel through walls and behave in ways that classical maps cannot explain.

The paper doesn't promise new technology or medical cures; instead, it provides a deep, mathematical map of why our universe behaves the way it does, and what other universes could look like if the rules were slightly different.

Technical Summary: Generalised Probabilistic Theories

Problem and Scope
The paper addresses the foundational structure of physical theories by introducing the Generalised Probabilistic Theories (GPT) framework, also known as the convex framework. The central problem is to characterize probabilistic theories—of which classical and quantum mechanics are specific instances—using a unified operational language. The authors aim to move beyond the specific mathematical formalisms of Hilbert spaces or probability simplices to define states, transformations, and measurements in terms of real vectors, matrices, and functionals. This approach seeks to clarify the operational meaning of quantum theory, its relationship to classical probability, and the landscape of potential theories that lie beyond quantum mechanics (post-quantum theories).

Methodology
The paper employs a systematic operational approach where physical processes are modeled as data tables generated by system preparation, transformation, and measurement. The methodology relies on three core mathematical representations:

States: Represented as real vectors containing lists of probabilities (or expectation values) for measurement outcomes.
Transformations: Modeled as real matrices acting on state vectors, respecting probabilistic mixtures (convex combinations).
Measurements: Represented by effect vectors (functionals) where the probability of an outcome is the dot product of the state vector and the effect vector.

The authors utilize the geometry of cones (state cones and effect cones) to define the state space and measurement spaces. They apply this framework to:

Reconstruct standard quantum theory from a set of operational axioms (following Hardy's approach).
Analyze non-locality by comparing classical, quantum, and post-quantum correlations within the geometry of the non-signalling polytope.
Generalize Hamiltonian mechanics and phase-space distributions to non-Hilbert space theories, investigating the role of quasiprobability negativity.

Key Contributions and Results

Operational Framework and Definitions: The paper establishes that states are elements of a compact, convex set $\Omega_A$ within a real vector space $A$ , and measurements are collections of linear functionals (effects) summing to a unit functional. It demonstrates how classical probability theory corresponds to a simplex (where mixed states have unique decompositions) and how quantum theory corresponds to a specific convex geometry (the Bloch ball for qubits) where mixed states admit non-unique decompositions.
Axiomatic Reconstruction of Quantum Theory: Following Hardy's axioms (Simplicity, Subspaces, Composite Systems, and Continuity), the paper derives the functional relationship between the number of degrees of freedom $K$ and the dimension of distinguishable states $N$ .
- It proves that $K(N) = N^r$ for some integer $r$ .
- It rules out the classical case ( $r=1$ ) based on the continuity of reversible transformations.
- It shows that the minimal remaining case ( $r=2$ , i.e., $K=N^2$ ) for a two-level system ( $N=2$ ) uniquely recovers the geometry of the Bloch sphere and the group of reversible transformations as $SO(3)$ (the adjoint representation of $SU(2)$).
Non-Locality and "Box World": The paper analyzes the landscape of correlations using the non-signalling polytope.
- It distinguishes between the local hidden-variable polytope ( $X_{HDS}$ ), the quantum set ( $X_{QS}$ ), and the full non-signalling polytope ( $X_S$ ).
- It introduces the concept of the gbit (generalised bit), a system with a cubic state space allowing all deterministic assignments of fiducial measurements, contrasting it with the spherical Bloch ball of the qubit.
- It discusses the Popescu-Rohrlich (PR) box, a theoretical construct within "Box World" that achieves the algebraic maximum of the CHSH inequality ( $S=4$ ) while respecting the no-signalling principle.
- A key result regarding dynamics in Box World is that despite the vast state space, the allowed reversible transformations are highly restricted (essentially limited to local relabellings and permutations), preventing the generation of PR-box correlations from product states via reversible dynamics.
Generalised Hamiltonian Mechanics and Phase Space: The paper extends the concept of Hamiltonians to GPTs by defining them as observables that generate reversible dynamics.
- It unifies classical and quantum dynamics through generalised phase-space distributions, where the time evolution is governed by a kernel-dependent bracket (generalizing the Poisson and Moyal brackets).
- It highlights Wigner function negativity as a signature of non-classicality. The paper demonstrates that negativity is operationally linked to:
  1. Contextuality: Proving an equivalence between the existence of a non-negative quasiprobability representation and a non-contextual ontological model (Spekkens' framework).
  2. Tunneling: Defining tunneling operationally as a violation of classical probability bounds in phase space, showing that this phenomenon relies on the negativity of either the state or the measurement effect functions.

Significance and Claims
The paper claims to provide a "deeper understanding" of quantum theory by isolating its operational features from its specific Hilbert space formalism. By framing classical, quantum, and post-quantum theories within a single convex framework, the authors demonstrate that:

Quantum theory is not the only possible probabilistic theory but is distinguished by specific geometric constraints (e.g., the structure of the state space and the continuity of reversible dynamics).
The "super-quantum" correlations allowed in Box World (like the PR box) are mathematically consistent with the no-signalling principle but are constrained by the lack of rich reversible dynamics in that framework.
Features often viewed as uniquely quantum, such as contextuality and tunneling, can be understood as consequences of the inability to represent the theory with a strictly positive phase-space distribution (negativity).

The authors position this work as a systematic introduction to the field, clarifying how operational principles constrain the possible structures of physical laws. They note that while the reconstruction of quantum theory for $N=2$ is rigorous, extending these results to arbitrary finite dimensions requires advanced machinery (such as Jordan algebras) and is left to more specialized literature. The paper serves as a foundational overview rather than a proposal for new experimental tests, focusing instead on the theoretical classification of probabilistic theories.