⚛️ quantum physics

Bell Inequalities for Smells

This paper introduces and analyzes a new class of Bell inequalities based on direct equality comparisons of measurement outcomes, such as smells, which yields thousands of new tight inequalities and demonstrates their utility as powerful tools for detecting nonlocality, serving as dimension and outcome witnesses, and characterizing genuine multipartite nonlocality.

Original authors: Ricardo Faleiro, Flavien Hirsch, Emmanuel Zambrini Cruzeiro, Nicolas Gisin

Published 2026-03-19

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Ricardo Faleiro, Flavien Hirsch, Emmanuel Zambrini Cruzeiro, Nicolas Gisin

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 at a party where everyone is playing a game. Usually, in these quantum physics games, players have to report back exact numbers or colors (like "I got a 7" or "I got Red"). But what if the players' answers were smells?

If Alice smells "Vanilla" and Bob smells "Vanilla," they can agree they are the same. But if Alice smells "Vanilla" and Bob smells "Lavender," they know they are different. The problem is, trying to describe the exact chemical composition of every smell is impossible and messy. All you can really do is ask: "Did you smell the same thing as me?"

This paper, titled "Bell Inequalities for Smells," is about a new way to test the spooky, weird rules of the quantum world using only these simple "Same or Different?" comparisons.

Here is the breakdown in simple terms:

1. The Problem: Too Many Variables

In standard quantum tests (like the famous CHSH experiment), scientists look at specific numbers. But in the real world, outcomes can be complex (like smells, sounds, or images). It's hard to write a math rule for "Is this smell the same as that one?" if you try to analyze every detail of the smell.

The authors realized: We don't need to know what the smell is, only if two people agree on it.

2. The Solution: The "Smell" Test

They created a new type of math rule (a Bell Inequality) that only cares about equality.

The Setup: Alice and Bob (and maybe Charlie and Dave) get secret instructions (inputs). They sniff a mystery scent (output).
The Rule: They can only report back: "We smelled the same thing" or "We smelled different things."
The Goal: They want to see if their "Same/Different" answers can be explained by a secret plan they made beforehand (Classical/Local), or if they are doing something weirdly connected that defies normal logic (Quantum/Non-local).

3. The Big Discovery: The "Saturation" Point

One of the coolest findings is a concept called Saturation.
Imagine you are trying to guess a password.

If the password is 1 digit long, you have 10 guesses (0-9).
If it's 2 digits, you have 100 guesses.
But the authors found that for these "Smell" games, once you have a certain number of possible smells (outcomes), adding more smells doesn't make the game any harder or more complex.

It's like having a bucket of water. Once the bucket is full, adding more water just makes it spill; it doesn't change the fact that the bucket is full. They proved that for any number of players and inputs, there is a "magic number" of smells after which the rules of the game stop changing. This makes the math much easier to solve!

4. The "Unanimous" Game

They also defined a special, stricter version called Unanimous Bell Inequalities.

The Rule: This game only counts the moments when everyone smells the exact same thing.
The Analogy: Imagine a choir. In a normal game, you check if any two singers match. In the "Unanimous" game, you only score points if the entire choir sings the exact same note at the exact same time.
Why it matters: They proved that these "Unanimous" games are actually just a fancy way of playing a specific type of logic puzzle called a "Deterministic Nonlocal Game." This connects their smell theory to existing math in a very elegant way.

5. Why Should We Care? (The "Dimension Witness")

The paper isn't just about smells; it's about detecting the size of the universe (or rather, the size of the quantum system).

Think of a quantum system like a library. A small library (low dimension) has few books. A huge library (high dimension) has millions.
The authors found that some of their "Smell" inequalities act like a Library Size Detector.
If the players win the game with a score that is too high for a small library, you know for a fact they are using a huge library (a high-dimensional quantum system), even if you can't see the books inside.

6. The "S33" Inequality: A Super-Sensitive Detector

They highlighted one specific rule (called S33) that is a superstar.

It can detect "spooky connections" (non-locality) in systems that are too weak to break the standard rules (like the famous CHSH inequality).
It's like having a smoke detector that is so sensitive it can smell a single match being struck, whereas the old detectors only go off when the house is on fire.

Summary

This paper is a masterclass in simplification.

Old way: "Tell me the exact value of your measurement." (Hard, messy, often impossible).
New way: "Just tell me if you agree with your partner." (Simple, elegant, powerful).

By stripping away the complex details and focusing only on agreement, the authors unlocked thousands of new mathematical rules. These rules help us:

Prove that quantum mechanics is weird (non-local).
Measure how "big" a quantum system is.
Find new ways to build secure quantum computers and communication networks.

In short: They took the complex world of quantum physics and turned it into a simple game of "Same or Different," proving that sometimes, the simplest questions yield the deepest answers.

1. Problem Statement

Standard Bell inequalities typically rely on discrete numerical outcomes or specific labels (e.g., $a, b \in \{0, 1, \dots, d-1\}$ ) to calculate correlations. However, in many physical or abstract scenarios, measurement outcomes may be difficult to characterize individually (e.g., "smells," colors, or complex states) but easy to compare for equality.

The authors address the gap in Bell nonlocality theory where the only accessible information is whether outcomes are identical or different, rather than their specific values. They define a new class of Bell inequalities based solely on equality-comparisons of outcomes. The core challenge is to characterize the "local polytope" (the set of classical behaviors) under this restricted observation model and to identify tight inequalities (facets) that can detect quantum nonlocality.

2. Methodology

The authors employ a combination of combinatorial geometry, graph theory, and computational polytope enumeration:

Mathematical Formulation:
- They define a scenario $(n, m, k)$ with $n$ parties, $m$ inputs, and $k$ outcomes.
- Instead of full probability distributions $p(a|x)$ , they focus on the probability of observing specific equality patterns (partitions) $\sigma$ . For $n$ parties, the number of possible equality patterns is given by the $n$ -th Bell number ( $B_n$ ).
- The behavior space is reduced to a sub-polytope $L_\Sigma$ (the "smells" local polytope), defined by the convex hull of classical behaviors respecting these equality constraints.
Combinatorial Analysis (Saturating Outcomes):
- Using graph theory, they represent deterministic local strategies as multipartite graphs where connected components represent groups of parties producing equal outputs.
- They prove that there exists a saturated number of outcomes ( $k^*$ ) beyond which increasing $k$ does not generate new vertices in the local polytope. This drastically reduces the computational complexity of finding inequalities.
Computational Enumeration:
- They use the PANDA software (Parallel AdjaceNcy Decomposition Algorithm) to enumerate facets of the reduced polytopes for various scenarios (bipartite and multipartite).
- They classify these facets by symmetry (party permutation, input permutation) and check if they are also facets of the standard full Bell local polytope.
Unanimous Inequalities:
- They introduce a stricter subclass called Unanimous Bell Inequalities, where only the event "all parties output the same value" is considered. They prove these can always be mapped to deterministic nonlocal games.

3. Key Contributions

A. Theoretical Framework: Bell Inequalities for Smells

Definition: Formalized Bell inequalities where probabilities are of the form $p(a=b|x,y)$ rather than specific outcome values.
Saturation Theorem (Theorem 1): Proved that for an $(n, m, k)$ $(n, m, k)$ scenario, the local polytope $L_\Sigma$ $L_{Σ}$ saturates at a finite number of outcomes $k^* = \lfloor \frac{n(m+1)}{2} \rfloor$ $k^{*} = ⌊ \frac{n ( m + 1 )}{2} ⌋$ . Increasing $k$ $k$ beyond this yields no new classical vertices.
- Corollary: For bipartite scenarios, $k^* = m+1$ .
Vertex Counting: Provided analytical expressions for the number of vertices in bipartite scenarios (Theorem 2), linking the problem to Stirling numbers of the second kind.

B. Unanimous Bell Inequalities

Defined inequalities considering only the partition where all parties agree ( $a_1 = a_2 = \dots = a_n$ ).
Theorem 3 & 4: Proved that unanimous inequalities always admit a saturating number of outcomes $k^* = m_{min} + 1$ and can be rewritten as deterministic nonlocal games. This connects the geometric approach (polytopes) with the game-theoretic approach to nonlocality.

C. Discovery of New Inequalities

Bipartite Results: Solved scenarios up to $(2, 5, 2)$ and conjectured completeness for $(2, 4, k)$ . Found thousands of new tight inequalities. Notably, many are Party-Permutation Invariant (PPI) and are also facets of the standard local polytope.
Multipartite Results: Solved scenarios up to 4 parties. Found $\sim 6.5$ million facet classes in $(4, 2, 2)$ , with $\sim 5$ million being facets of the standard local polytope.

4. Key Results and Examples

Dimension Witnesses:
- The inequality $S_{33}$ (in scenario $2,3,3$) acts as a dimension witness. It has a local bound of 3, but quantum strategies with dimension $d=2$ achieve $\ge 3.5$ , and higher dimensions yield higher violations.
- Crucially, $S_{33}$ detects nonlocality in states that do not violate CHSH, demonstrating that "smell" inequalities can reveal nonlocality missed by standard binary inequalities.
Genuine Multipartite Nonlocality (GMNL):
- Identified inequalities (e.g., $S_{222}$ in $3,2,3$ and $U_4$ in $4,2,2$) that act as witnesses for GMNL.
- $U_4$ is a unanimous inequality where the quantum value ( $5/4$ ) exceeds the local bound ($1$) and the bilocal-no-signaling bound, proving genuine 4-partite nonlocality.
Outcome Witnesses:
- The inequality $S_{222}$ (lifted to $3,2,3$) demonstrates that certain quantum violations are impossible with 2 outcomes but possible with 3, acting as a witness for the cardinality of the outcome space.
Structural Properties:
- For bipartite "smell" inequalities, the No-Signaling (NS) and Signaling (S) bounds coincide ($NS = S$).
- For all bipartite facets (except positivity), the hierarchy is strict: $L < Q < NS = S$. This implies no "trivial" bipartite facets exist where quantum mechanics cannot violate the local bound.

5. Significance

New Tool for Nonlocality: The paper provides a powerful, computationally tractable method to discover new Bell inequalities. By reducing the dimensionality of the behavior space (focusing on equality rather than specific values), they can solve scenarios that are intractable for full polytope enumeration.
Physical Relevance: The framework applies to experiments where outcomes are hard to label but easy to compare (e.g., quantum sensing, biological systems, or high-dimensional quantum states).
Device-Independent Protocols: The discovered inequalities serve as robust dimension witnesses, outcome witnesses, and GMNL witnesses. They can certify the dimensionality of quantum systems or the nature of multipartite entanglement without assuming the internal workings of the devices.
Connection to Mathematics: The work bridges Bell nonlocality with combinatorial mathematics, specifically utilizing Bell numbers and Stirling numbers to characterize the structure of local strategies.

In summary, "Bell Inequalities for Smells" establishes a rigorous theoretical framework for analyzing nonlocality based on outcome equality, leading to the discovery of thousands of new inequalities with practical applications in device-independent quantum information processing.