⚛️ quantum physics

Squeezing-enhanced Pairwise Fusion of Photonic Qudits

This paper demonstrates that applying identical single-mode squeezers to the outputs of linear-optical pairwise fusion gates for photonic qudits enables the recovery of structured measurement failures via photon-number parity analysis, thereby increasing the ideal success probability of Bell projections without requiring ancillary input photons.

Original authors: Pradip Laha, Peter van Loock

Published 2026-06-30

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: Pradip Laha, Peter van Loock

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 send a secret message using tiny packets of light called photons. In the world of quantum computing, these photons can carry more than just a simple "0" or "1"; they can carry complex information by existing in one of many different "tracks" or "rails" at once. This is called a qudit (a quantum digit with many sides).

The goal of this research is to perform a specific magic trick called Pairwise Fusion. Think of this as trying to take two separate quantum messages and "glue" them together into a single, perfectly entangled pair.

The Problem: The "Glue" Sometimes Fails

In the standard way of doing this (using only mirrors and beam splitters), the process works great for about 75% to 83% of the time, depending on how complex the message is.

However, when it fails, it doesn't just fail randomly. The paper describes a very specific pattern:

The Successes: When the two photons are on different tracks, the "glue" works perfectly.
The Failures: When the two photons happen to be on the same track, the process fails. But here's the catch: the failure isn't a total mess. It's a structured failure. The photons are still there, but they are stuck in a specific "diagonal" state that the standard machine can't read.

Usually, when a quantum experiment fails, you just throw that attempt away and try again. This paper asks: Can we fix these specific "stuck" failures without throwing them away?

The Solution: The "Squeezing" Trampoline

The authors propose adding a new tool to the machine: Squeezers.

Imagine the photons are like balls bouncing on a trampoline.

The Passive Machine: Without the trampoline, if two balls land on the same spot (the "diagonal failure"), they just sit there and the machine gives up.
The Squeezing Layer: The authors add a special "squeezing" layer (a type of optical lens that stretches and compresses the light) right before the detectors.

Think of the squeezer as a trampoline that reshapes the landing.

It leaves the winners alone: If the photons were already on different tracks (the successful cases), the squeezer doesn't mess them up. They still land in the "success" zone.
It rescues the losers: If the photons were stuck on the same track (the failure), the squeezer stretches their energy in a very specific way. It redistributes the light so that some of these "stuck" photons now land in a new, readable pattern.

The "Imbalance" Rule

How does the machine know which of these reshuffled failures are actually good? The paper discovers a simple rule, like a secret code:

The machine looks at the number of photons landing in each track. It calculates the imbalance (the difference) between the tracks.

If the imbalance vector has exactly two tracks with a difference, and those differences are equal in size, the machine says, "Aha! This is a valid success!"
If the pattern is messy or has more than two tracks involved, it says, "Nope, that's still a failure," and discards it.

The Results: A Small but Steady Win

By using this "squeezing trampoline," the researchers managed to recover some of the lost attempts.

For a 4-track system, the success rate went from 75% up to 79.6%.
For a 6-track system, it went from 83.3% up to 87.1%.

It's not a massive jump, but in the world of quantum computing, where every photon counts, squeezing out an extra few percent of success is a big deal. It means you need fewer photons and less time to build a quantum network.

The Catch (The Cost)

To get this boost, you have to pay a price:

Complexity: You need to add 2 "squeezing" machines for every track (so, a lot of extra hardware).
Precision: The detectors need to be very good at counting high numbers of photons, because the squeezing can sometimes create bursts of many photons at once. If the detector gets "saturated" (too full to count), it throws the result away.

Summary

The paper demonstrates a clever trick: instead of trying to fix the entire quantum measurement process, they identified the specific "shape" of the failures, added a tool (squeezing) that only affects those failures, and turned a portion of the "trash" into "treasure." It's like finding a way to salvage the broken pieces of a puzzle and fit them back in, rather than just throwing the broken pieces in the bin.

Technical Summary: Squeezing-Enhanced Pairwise Fusion of Photonic Qudits

Problem Statement
Photonic quantum information processing relies heavily on measurement-driven operations, particularly Bell-state measurements (BM) and fusion gates, to generate entanglement and perform computation. In linear-optical architectures, deterministic photon-photon nonlinearities are unavailable, making these operations probabilistic. While high-dimensional encoding (d-rail qudits) offers a larger logical space without increasing photon counts, it does not inherently remove the constraints of passive linear optics.

Specifically, the ancilla-free pairwise fusion gate (PFG) introduced by Yamazaki and Azuma projects two d-rail qudits onto Bell states supported on two-rail subspaces. Without active processing, this passive measurement succeeds with probability $1 - d^{-1}$ . Crucially, the failure sector is not generic; it is entirely confined to the "diagonal" logical subspace (where both photons occupy the same rail), while the "off-diagonal" sector (different rails) is successfully projected. The central challenge addressed by this work is whether an active optical layer can selectively recover a portion of this structured diagonal failure sector without introducing ancillary input photons, thereby increasing the overall success probability of the fusion gate.

Methodology
The authors propose a scheme where identical single-mode squeezers are applied to all $2d$ output modes of the ancilla-free PFG interferometer, immediately preceding photon-number-resolving (PNR) detection.

Logical Structure: The input consists of two single-photon qudits. The passive interferometer (a product of balanced beam splitters) maps off-diagonal inputs to odd-parity output patterns and diagonal inputs to even-parity, bunched patterns.
Squeezing Layer: A local Gaussian layer of $2d$ single-mode squeezers is appended. This layer preserves photon-number parity locally. Consequently, the odd-parity signatures identifying successful off-diagonal fusion remain disjoint from the all-even diagonal sector.
Failure Engineering: The squeezing redistributes photon-number amplitudes within the all-even diagonal sector. The authors analyze the resulting Kraus operators to determine which detection patterns correspond to valid pairwise Bell projections (specifically the diagonal Bell states $|\Phi^{\pm}_{ij}\rangle$ ).
Acceptance Criterion: A detection pattern is accepted as a successful fusion outcome if:
- It is an off-diagonal odd-parity pattern (passive success).
- It is an all-even pattern where the "photon-number-imbalance vector" (the difference in photon counts between the two output ports of each rail) has exactly two nonzero components of equal magnitude. This condition ensures the restricted Kraus bra is proportional to a definite pairwise diagonal Bell state.
Finite Resolution: The authors extend the theory to finite detector resolution, where a saturation threshold $n_{sat}$ defines a trusted window. Events with saturated detectors are declared erasures to ensure a conservative, certified success probability.

Key Contributions

Exact POVM Derivation: The paper derives the exact logical-space Positive Operator-Valued Measure (POVM) for the squeezing-enhanced PFG. It proves that the diagonal recovery is governed by a global imbalance criterion, independent of the squeezing strength $r$ , though the probability weight depends on $r$ .
Closed-Form Success Probability: The authors derive an exact closed-form expression for the ideal-PNR success probability involving complete elliptic integrals of the first and second kind. This expression quantifies the recovery of the diagonal failure sector.
Optimization Analysis: The study identifies the optimal squeezing parameter $r^*$ that maximizes the success probability for a given dimension $d$ . It demonstrates that the optimal squeezing strength scales as $r^* \approx 1/\sqrt{d}$ for large $d$ .
Finite-Resolution Theory: A rigorous framework is developed for finite-resolution detectors, showing that while saturation suppresses the advantage, it does not eliminate it.

Results
The application of local squeezing yields a systematic, moderate increase in the ideal success probability:

Ideal-PNR Performance:
- For $d=4$ , the success probability increases from the passive benchmark of 75% to 79.62%.
- For $d=6$ , it increases from 83.33% to 87.15%.
- Asymptotically, for large $d$ , the optimized measurement recovers a fraction $e^{-1}$ of the diagonal failure sector, resulting in an absolute gain of approximately $1/(ed)$.
Finite-Resolution Performance:
- With a representative saturation threshold of $n_{sat} = 7$ (resolving photon numbers 0–6), the certified success probabilities remain high: 78.84% for $d=4$ and 86.71% for $d=6$ .
- The advantage persists even with limited dynamic range, provided the detector can resolve the relevant low-photon-number patterns.

Significance and Claims
The paper establishes active Gaussian processing (specifically local squeezing) as a viable method for "recycling" structured measurement failures in photonic quantum computing. The significance lies in the following points:

Ancilla-Free Enhancement: Unlike other high-efficiency schemes that require populated ancillary states or enlarged interferometers, this method enhances the PFG without adding ancillary input photons.
Failure-Space Engineering: The work illustrates a strategy of preserving the already successful sector of a measurement while selectively reshaping the support of its failure POVM. The structured nature of the passive failure (confined to the diagonal sector) makes it amenable to this targeted recovery.
Experimental Feasibility: The required resources are $2d$ squeezing operations and a larger photon-number dynamic range at detection. The authors note that while the gains are moderate, they are systematic and do not rely on idealized assumptions beyond lossless detection and perfect squeezing.
Scalability: The method is applicable to arbitrary dimensions $d$ , with the optimal squeezing strength decreasing as dimension increases, suggesting compatibility with scaling to larger logical spaces.

The authors conclude that while realistic imperfections (loss, detector inefficiency, phase drift) must be quantified for practical implementation, the theoretical framework provides a clear path for improving fusion-based quantum computation and entanglement distribution rates in high-dimensional photonic systems.