Simulating imperfect quantum optical circuits using unsymmetrized bases

John Steinmetz, Maike Ostmann, Alex Neville, Brendan Pankovich, Adel Sohbi · 2024

Read Simulating imperfect quantum optical circuits using unsymmetrized bases by John Steinmetz, Maike Ostmann, Alex Neville, Brendan Pankovich, Adel Sohbi (2024) 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-2412-13330.txt.

Introduction

John Steinmetz, Maike Ostmann, Alex Neville, Brendan Pankovich, and Adel Sohbi shrink the simulation space for imperfect photonic quantum circuits by refusing to symmetrize what need not be. Fault-tolerant photonic computing needs huge entangled states, and errors like loss and partial distinguishability must be simulated on them. The standard photon basis explodes combinatorially with photon number and mode count. Their unsymmetrized basis is exponentially smaller and loses no information.

Joint simulation of loss and distinguishability then reaches states far beyond the old frontier. The team delivers the first simulations of imperfect qubits encoded in quantum parity codes. In one example the working space is over sixty orders of magnitude smaller than the full photon space. A derivation of the loss mechanism for partially distinguishable photons accompanies the method. Bigger resource states can finally be error-checked in silico.

Catalog record

CREATORJohn Steinmetz, Maike Ostmann, Alex Neville, Brendan Pankovich, Adel Sohbi
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE94 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-13330

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Simulating imperfect quantum optical circuits using unsymmetrized bases John Steinmetz, Maike Ostmann, Alex Neville, Brendan Pankovich, Adel Sohbi Abstract Fault-tolerant photonic quantum computing requires the generation of large entangled resource states. The required size of these states makes it challenging to simulate the effects of errors such as loss and partial distinguishability. For an interferometer with N N partially distinguishable input photons and M M spatial modes, the Fock basis can have up to ( N + N ​ M − 1 N ) {N+NM-1\choose N} elements. We show that it is possible to use a much smaller unsymmetrized basis with size M N M^{N} without discarding any information. This enables simulations of the joint effect of loss and partial distinguishability on larger states than is otherwise possible. We demonstrate the technique by providing the first-ever simulations of the generation of imperfect qubits encoded using quantum parity codes, including an example where the Hilbert space is over 60 60 orders of magnitude smaller than the N N -photon Fock space. As part of the analysis, we derive the loss mechanism for partially distinguishable photons. Preprint: APS/123-QED I Introduction A linear optical quantum computer is made up of circuits that generate and manipulate multiphoton entangled resource states [ 1 , 2 ] . These circuits are composed of linear optical components such as beam splitters and phase shifters, which in practice can each act imperfectly on the quantum state. It is important to understand how different imperfections in these circuit components affect the quantum state in order to predict the output of the circuit and determine the best way to correct any errors that occur. Quantum circuit simulators can be used to classically model these imperfect circuits, so they are a valuable tool to guide the design of a physical device. Several simulators intended specifically for photonic circuits have been released in recent years, coinciding with the rapid growth of photonic quantum computing as a field [ 3 , 4 , 5 , 6 , 7 , 8 ] . Partial distinguishability is an important source of error to accurately model in optical circuits due to the practical difficulty of producing identical photons. If the photons in a circuit have different properties, then they do not perfectly interfere. The output of a boson sampling experiment with partial distinguishability can be modelled using matrix permanents [ 9 , 10 , 11 ] . There have also been several publications in recent years that incorporate some version of partial distinguishability in a linear optical circuit simulation [ 7 , 12 , 13 ] . When these simulations use density matrices (or state vectors), the partial distinguishability is modelled by adding extra modes to the system corresponding to the additional properties of the photons. This increases the space size, so there is a significant restriction on the size of circuits that can be simulated. The larger space size also makes it more difficult to model other imperfections such as loss, so the problem of combining loss with partial distinguishability is largely unexplored. An alternative to density matrix simulations is to use matrix permanents, which are frequently used in the context of boson sampling [ 14 , 15 ] . For a system of indistinguishable photons, permanents can be used to reconstruct the full density matrix at the output of a circuit. However, when the photons are partially distinguishable, this method only yields output probabilities, not amplitudes; reconstructing the full state can be done with tomographic reconstruction [ 16 ] , which has a higher computational cost. It is also more difficult to deal with processes like loss that yield mixed states. These can be modelled in a way that keeps the state pure, like adding environment modes and expanding the working space. In certain special cases, loss can be commuted to the beginning of the circuit, which allows each component of the mixed state to be treated separately. Despite these limitations, the process of computing matrix permanents is highly optimized, so this is often the method of choice for simulating circuits. In this paper, we propose a new set of density matrix simulation techniques that can be used to model systems that include both partially distinguishable photons and non-uniform loss in a way that scales more favourably with the circuit size than previously published methods. This is done by working in an unsymmetrized space instead of the usual Fock space, delaying the photons from interfering until the end of the circuit. We propose a specific unsymmetrized basis for a system of partially distinguishable photons such that the dimension of the Hilbert space scales better than Fock space. The new framework naturally handles mixed states since states are still represented as density matrices, so we can implement non-uniform loss without introducing any extra modes. In order to model loss and distinguishability together, we derive generalized Kraus operators and show how to use them in the unsymmetrized space. We demonstrate the new method by obtaining complete density matrices for some example circuits in the presence of arbitrary loss and partial distinguishability, focusing on generators of entangled resource states that are included in existing proposals for fault-tolerant devices. In Section II , we review the theory of partially distinguishable photons in first and second quantization. In Section III , we propose an unsymmetrized basis and show how it can result in a smaller space size. In Section IV , we demonstrate how to obtain physically meaningful results from the unsymmetrized density matrix. In Section V , we show how to combine loss with partial distinguishability and integrate it into the proposed framework. In Section VI , we demonstrate our new simulation techniques on some example circuits, including generators of encoded qubits. In Section VII , we conclude and discuss the merits of our proposal relative to other simulation techniques. II Partially distinguishable photons II.1 Internal states Each particle in a circuit, in addition to its spatial location, has other properties that could potentially distinguish it from other particles. For example, each particle might have a different spectral profile, time delay or polarization. We can express each particle’s state as the tensor product of an external state |m\rangle , which refers to its location in the circuit, and an internal state |\phi\rangle , which contains all its other properties [ 17 ] . Note that at this stage we are using the word “particle” instead of the word “photon”, because until the state is symmetrized, the analysis applies to both bosons and fermions. The single-photon Hilbert space is \mathcal{H}=\mathcal{H}_{\text{ext}}\otimes\mathcal{H}_{\text{int}}, (1) where the external and internal spaces are spanned by the bases \mathcal{B}_{\text{ext}}=\{|m\rangle\},\quad\mathcal{B}_{\text{int}}=\{|\phi\rangle\}, (2) where m\in\{1,\dots,M\} . While the external modes are all mutually orthogonal, this is not necessarily the case for the internal modes. Each pair of internal modes has an overlap \langle\phi_{i}|\phi_{j}\rangle\equiv S_{ij} , which can be considered elements of a distinguishability matrix S [ 18 ] . The square of these overlaps |S_{ij}|^{2} is the pairwise visibility, which in a photonic system quantifies the size of the fringes that appear when two photons are interfered at a beam splitter, as in a Hong-Ou-Mandel experiment [ 19 , 20 ] . A visibility of |S_{ij}|^{2}=0 between two

Read the complete text