Quantum circuits for simulating linear interferometers
Hudson Leone, Peter S. Turner, Simon Devitt · 2025
Read Quantum circuits for simulating linear interferometers by Hudson Leone, Peter S. Turner, Simon Devitt (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-16880.txt.
Introduction
Hudson Leone, Peter S. Turner, and Simon Devitt write a 2025 study on how to mimic an optical device, called an interferometer, on a quantum computer. Their work was spurred by new quantum proof-of-work ideas. In their plan, the registers of the machine stand for the paths that light can take.
They break the math of the device into small two-by-two blocks, turn each block into circuit steps, and join the steps. They also built a software tool, named Aquinas, that does this work. As a test, it built a circuit for a five-path device with two photons, using ten qubits and a deep stack of steps.
Catalog record
| CREATOR | Hudson Leone, Peter S. Turner, Simon Devitt |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 28 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-16880
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Quantum circuits for simulating linear interferometers Hudson Leone, Peter S. Turner, Simon Devitt Abstract Motivated by recent proposals for quantum proof of work protocols, we investigate approaches for simulating linear optical interferometers using digital quantum circuits. We focus on a second quantisation approach, where the quantum computer’s registers represent optical modes. We can then use standard quantum optical techniques to decompose the unitary matrix describing an interferometer into an array of 2 × 2 2\times 2 unitaries, which are subsequently synthesised into quantum circuits and stitched together to complete the circuit. For an m m mode interferometer with n n identical photons, this method requires approximately 𝒪 ( m log ( n ) ) \mathcal{O}(m\log(n)) qubits and a circuit depth of 𝒪 ( m n 4 log 2 ( n ) polylog ( n 4 / ϵ ) ) \mathcal{O}(mn^{4}\log_{2}(n)\>\textrm{polylog}(n^{4}/\epsilon)) . We present a software package Aquinas ( a qu antum in terferometer as sembler) that uses this approach to generate such quantum circuits. For reference, an arbitrary five mode interferometer with two identical photons is compiled to a 10 qubit quantum circuit with a depth of 1972. I Introduction Linear optics has been at the forefront of quantum information research for many years. While interferometers natively execute certain types of high fidelity quantum operations, they cannot be practical universal quantum computers unless some nonlinearity is introduced Kok et al. 2007 . Nevertheless there remains significant interest in linear quantum photonics, particularly on account of the seminal work of Aaronson and Arkhipov which demonstrated that the interferometry problem of boson sampling is almost certainly classically intractable Aaronson and Arkhipov 2011 . This stems from the fact that even approximating the output occupation probabilities of a boson sampling experiment requires one to calculate the permanents of large matrices, which is a famously #P-complete problem. One implication of this is that linear optical interferometers, which are considerably easier to build than general purpose quantum computers, could in principle be used to demonstrate genuine (albeit impractical) quantum computational supremacy. Significant work has therefore been done to engineer interferometers that are both large enough and reliable enough to demonstrate an incontestable quantum advantage (see Brod et. al. for a review of progress Brod et al. 2019 ). Of note is a recent result from Madsen et. al. claiming supremacy using a variant of boson sampling that uses Gaussian states Madsen et al. 2022 . Although linear interferometry lacks the computational power of a general purpose quantum computer, the proliferation of these experimental results has motivated research around finding practical applications for boson sampling. Some computational examples include locating dense sub-graphs Arrazola and Bromley 2018 and the graph isomorphism problem Brádler et al. 2021 , though whether such algorithms can yield a tangible advantage in practice remains to be seen. An arguably more promising application for boson sampling is in distributed consensus between quantum computers. Recent work has indicated the suitability of coarse grained boson sampling as a proof of work algorithm, which may see use in future quantum infrastructure Singh et al. 2024 . Presently, to our knowledge the best reported algorithm for simulating linear interferometers is the classical method from Heurtel et. al. which, despite savings over conventional permanent based calculations, is still exponential in general (as expected) Heurtel et al. 2023 . It is worth mentioning that a major practical issue with developing boson samplers is that of photon loss and distinguishability. Though the hardness of boson sampling remains unchanged when a constant number of photons are lost Aaronson and Brod 2016 , losing a fraction of the total number of photons can render the sampling problem classically tractable Oszmaniec and Brod 2018 , Moylett et al. 2019 . One may then ask how to go about simulating boson sampling (and linear optics more generally) on a conventional quantum computer. Of course any quantum computer that can emulate boson sampling will generally be more powerful than an interferometer. Nevertheless, this could conceivably be useful for a quantum proof-of-work protocol, and simulating linear interferometers beyond what is experimentally feasible may be of fundamental interest. In this paper we report a new software package called Aquinas ( A Qu antum In terferometer As sembler) Leone which expresses a given linear interferometer as a quantum circuit that accurately simulates ideal linear interferometry for up to a pre-specified number of photons and modes. Broadly speaking, the algorithm works by first decomposing the given multimode interferometer into a network of interlaced two mode beamsplitters and phaseshifters. Following this, the Hamiltonian of each two mode piece is expressed in terms of ladder operators over the respective modes. By truncating the ladder operators at a fixed photon number (the total number of photons that enter the interferometer) and using some padding, we obtain a Hamiltonian that, when exponentiated, gives a unitary matrix that can be synthesised into a quantum sub-circuit. These sub-circuits are then knitted together in the configuration of the original beamsplitters to form the interferometer circuit. The main advantage of this ‘divide-and-conquer’ approach is that, by bringing compilation down to the level of individual beamsplitters, the problem of synthesising an interferometer circuit becomes computationally tractable with respect to both n , the number of photons, and m , the number of modes. As we will show, a clear drawback is that our circuits have an asymptotic depth that scales poorly with n . Our technique is therefore best suited for sparse interferometers with many modes and a limited number of photons. As a rule-of-thumb, a classically-hard instance of boson-sampling requires upwards of n\geq 30 photons over \mathcal{O}(n^{2}) modes Aaronson and Arkhipov 2011 . Our circuit depths become prohibitively large even for relatively easy experiments (as we will see in fig. 4 ). Nevertheless, we hope this work is a helpful first-step towards simulating quantum optics with digital quantum computers. In the following section, we describe how a linear interferometer evolves the creation operators of a photon number state and explain how these states can be described in either first or second quantization. In section III , we limit our attention to the second quantization and detail our algorithm. Finally, in section IV , we present numerical analysis that indicates our software is working correctly while allowing us to approximate circuit depths for moderate instances of boson sampling. II Linear optics and quantum computation background Figure 1 : A generic quantum circuit for simulating a linear interferometer in first quantization. The n registers are m -dimensional qudits that represent indistinguishable photons. Measurement in the computational basis indicates which of the m possible modes that photon is found in. By the symmetrization postulate for bosons, reordering the n registers can’t change the input state; this is implemented by the highly entangling Sym operation. Following this, the defining interferometer operation U is applied to each photon locally. Suppose we have a linear interferometer with m input and m output modes. Such a device is usually characterised in second quantisation by an m\times m unitary U that describes how creation operators are transformed for each mode. Specifically, if \hat{a}^{\dagger}_{i} is the creation operator for the i th mode, we have \hat{a}^{\dagger}_{i}\mapsto\sum_{i^{\prime}=1}^{m}U_{ii^{\prime}}\hat{a}^{\dagger}_{i^{\prime}} (1)