Table of Contents
Fetching ...

A Complete Graphical Language for Linear Optical Circuits with Finite-Photon-Number Sources and Detectors

Nicolas Heurtel

TL;DR

This paper introduces the $\textbf{LO}_{fi}$-calculus, a graphical language to reason on the infinite-dimensional bosonic Fock space with circuits composed of four core elements of linear optics: the phase shifter, the beam splitter, and auxiliary sources and detectors with bounded photon number.

Abstract

Linear optical circuits can be used to manipulate the quantum states of photons as they pass through components including beam splitters and phase shifters. Those photonic states possess a particularly high level of expressiveness, as they reside within the bosonic Fock space, an infinite-dimensional Hilbert space. However, in the domain of linear optical quantum computation, these basic components may not be sufficient to efficiently perform all computations of interest, such as universal quantum computation. To address this limitation it is common to add auxiliary sources and detectors, which enable projections onto auxiliary photonic states and thus increase the versatility of the processes. In this paper, we introduce the $\textbf{LO}_{fi}$-calculus, a graphical language to reason on the infinite-dimensional bosonic Fock space with circuits composed of four core elements of linear optics: the phase shifter, the beam splitter, and auxiliary sources and detectors with bounded photon number. We present an equational theory that we prove to be complete: two $\textbf{LO}_{fi}$-circuits represent the same quantum process if and only if one can be transformed into the other with the rules of the $\textbf{LO}_{fi}$-calculus. We give a unique and compact universal form for such circuits.

A Complete Graphical Language for Linear Optical Circuits with Finite-Photon-Number Sources and Detectors

TL;DR

This paper introduces the -calculus, a graphical language to reason on the infinite-dimensional bosonic Fock space with circuits composed of four core elements of linear optics: the phase shifter, the beam splitter, and auxiliary sources and detectors with bounded photon number.

Abstract

Linear optical circuits can be used to manipulate the quantum states of photons as they pass through components including beam splitters and phase shifters. Those photonic states possess a particularly high level of expressiveness, as they reside within the bosonic Fock space, an infinite-dimensional Hilbert space. However, in the domain of linear optical quantum computation, these basic components may not be sufficient to efficiently perform all computations of interest, such as universal quantum computation. To address this limitation it is common to add auxiliary sources and detectors, which enable projections onto auxiliary photonic states and thus increase the versatility of the processes. In this paper, we introduce the -calculus, a graphical language to reason on the infinite-dimensional bosonic Fock space with circuits composed of four core elements of linear optics: the phase shifter, the beam splitter, and auxiliary sources and detectors with bounded photon number. We present an equational theory that we prove to be complete: two -circuits represent the same quantum process if and only if one can be transformed into the other with the rules of the -calculus. We give a unique and compact universal form for such circuits.
Paper Structure (11 sections, 7 theorems, 34 equations, 6 figures)

This paper contains 11 sections, 7 theorems, 34 equations, 6 figures.

Key Result

Proposition 6

For any $\textbf{LOPP}$-circuits $D$ and $D'$, if $\textbf{LOPP}\xspace\vdash D=D'$ then $\left\llbracket D \right\rrbracket_1=\left\llbracket D' \right\rrbracket_1$.

Figures (6)

  • Figure 1: \labelfig:ex-CNotOptical circuits implementing the CZ two-qubit logic gate with auxiliary sources and detectors. On the left is the original circuit\footnotemark of \citeknill2002note. There are two auxiliary photon generated on the bottom left: if exactly one photon is detected for each of the two last wires on the bottom right, then we know we have performed the operation $\left| 11 \right\rangle\mapsto -\left| 11 \right\rangle$ on the two first wires. This event has a probability $\frac227$ to occur. On the right is an equivalent representation in the \ensuremath\textbf\textupLO_\bmfi-calculus, where $\ensuremath\bmf$ and $\ensuremath\bmg$ are two-photon states and linear forms.
  • Figure 2: Linear optical circuit generating with a $\frac{1}{9}$ probability the Bell state $\left| \Phi^+ \right\rangle=\left| 1010 \right\rangle+\left| 0101 \right\rangle$, with the use of auxiliary sources and detectors. On the left is the original circuit of fldzhyan2021compact, on the right is an equivalent and modular description. Both are equivalent circuits in the ${\textbf{LO}}_{{\bm{fi}}}$-calculus.
  • Figure 3: New and minimal equational theory of the $\textbf{LOPP}$-calculus. For any angle of the LHS (resp. RHS) of the Equation (\ref{['nLOpp:E2']}) and (\ref{['nLOpp:E3']}), there exist angles for the RHS (resp. LHS) such that the equations are sound. The angles of are unique if we restrict $\alpha_0,\alpha_2,\beta_0,\beta_1,\beta_3\in[0,2\pi)$, $\alpha_1 \in [0,\frac{\pi}{2}),\alpha_3\in[0,\pi)$, $\beta_2\in[0,\frac{\pi}{2}]$, and by taking $\alpha_1=0$ if $\alpha_0-\alpha_2=0~\text{mod}~\pi$ and $\beta_1=0$ if $\beta_2\in\left\{0,\frac{\pi}{2}\right\}$. The rotations associated with Equations (\ref{['nLOpp:E2']}) and (\ref{['nLOpp:E2']}) are detailed in the proof of prop:soundLOpp:[] defDefinition \ref{['prop:soundLOpp']} figFigure \ref{['prop:soundLOpp']} lemLemma \ref{['prop:soundLOpp']} axiomAxiom (\ref{['prop:soundLOpp']}) secSection \ref{['prop:soundLOpp']} subsecSection \ref{['prop:soundLOpp']} thmTheorem \ref{['prop:soundLOpp']} propProposition \ref{['prop:soundLOpp']} tabTable \ref{['prop:soundLOpp']} oLOppEquation (\ref{['prop:soundLOpp']}) nLOppEquation (\ref{['prop:soundLOpp']}) appAppendix \ref{['prop:soundLOpp']} remRemark \ref{['prop:soundLOpp']} exExample \ref{['prop:soundLOpp']} eqEquation (\ref{['prop:soundLOpp']}) [] , and the explicit values of the angles are detailed in subsec:E2angles:[] defDefinition \ref{['subsec:E2angles']} figFigure \ref{['subsec:E2angles']} lemLemma \ref{['subsec:E2angles']} axiomAxiom (\ref{['subsec:E2angles']}) secSection \ref{['subsec:E2angles']} subsecSection \ref{['subsec:E2angles']} thmTheorem \ref{['subsec:E2angles']} propProposition \ref{['subsec:E2angles']} tabTable \ref{['subsec:E2angles']} oLOppEquation (\ref{['subsec:E2angles']}) nLOppEquation (\ref{['subsec:E2angles']}) appAppendix \ref{['subsec:E2angles']} remRemark \ref{['subsec:E2angles']} exExample \ref{['subsec:E2angles']} eqEquation (\ref{['subsec:E2angles']}) [] and \ref{['subsec:E3angles']}.
  • Figure 4: Old axioms of the $\textbf{LOPP}$-calculus that are not in fig:nLOpp:[] defDefinition \ref{['fig:nLOpp']} figFigure \ref{['fig:nLOpp']} lemLemma \ref{['fig:nLOpp']} axiomAxiom (\ref{['fig:nLOpp']}) secSection \ref{['fig:nLOpp']} subsecSection \ref{['fig:nLOpp']} thmTheorem \ref{['fig:nLOpp']} propProposition \ref{['fig:nLOpp']} tabTable \ref{['fig:nLOpp']} oLOppEquation (\ref{['fig:nLOpp']}) nLOppEquation (\ref{['fig:nLOpp']}) appAppendix \ref{['fig:nLOpp']} remRemark \ref{['fig:nLOpp']} exExample \ref{['fig:nLOpp']} eqEquation (\ref{['fig:nLOpp']}) [] . They are derived in app:proofcompLOpp:[] defDefinition \ref{['app:proofcompLOpp']} figFigure \ref{['app:proofcompLOpp']} lemLemma \ref{['app:proofcompLOpp']} axiomAxiom (\ref{['app:proofcompLOpp']}) secSection \ref{['app:proofcompLOpp']} subsecSection \ref{['app:proofcompLOpp']} thmTheorem \ref{['app:proofcompLOpp']} propProposition \ref{['app:proofcompLOpp']} tabTable \ref{['app:proofcompLOpp']} oLOppEquation (\ref{['app:proofcompLOpp']}) nLOppEquation (\ref{['app:proofcompLOpp']}) appAppendix \ref{['app:proofcompLOpp']} remRemark \ref{['app:proofcompLOpp']} exExample \ref{['app:proofcompLOpp']} eqEquation (\ref{['app:proofcompLOpp']}) [] . In Equations (\ref{['oLOpp:E2']}) and (\ref{['oLOpp:E3']}), the LHS circuit has arbitrary parameters which uniquely determine the parameters of the RHS circuit. For any $\alpha_i\in\mathbb R$, there exist $\beta_j\in[0,2\pi)$ such that oLOpp:E2:[] defDefinition \ref{['oLOpp:E2']} figFigure \ref{['oLOpp:E2']} lemLemma \ref{['oLOpp:E2']} axiomAxiom (\ref{['oLOpp:E2']}) secSection \ref{['oLOpp:E2']} subsecSection \ref{['oLOpp:E2']} thmTheorem \ref{['oLOpp:E2']} propProposition \ref{['oLOpp:E2']} tabTable \ref{['oLOpp:E2']} oLOppEquation (\ref{['oLOpp:E2']}) nLOppEquation (\ref{['oLOpp:E2']}) appAppendix \ref{['oLOpp:E2']} remRemark \ref{['oLOpp:E2']} exExample \ref{['oLOpp:E2']} eqEquation (\ref{['oLOpp:E2']}) [] is sound, and for any $\gamma_i\in\mathbb R$, there exist $\delta_j\in[0,2\pi)$ such that oLOpp:E3:[] defDefinition \ref{['oLOpp:E3']} figFigure \ref{['oLOpp:E3']} lemLemma \ref{['oLOpp:E3']} axiomAxiom (\ref{['oLOpp:E3']}) secSection \ref{['oLOpp:E3']} subsecSection \ref{['oLOpp:E3']} thmTheorem \ref{['oLOpp:E3']} propProposition \ref{['oLOpp:E3']} tabTable \ref{['oLOpp:E3']} oLOppEquation (\ref{['oLOpp:E3']}) nLOppEquation (\ref{['oLOpp:E3']}) appAppendix \ref{['oLOpp:E3']} remRemark \ref{['oLOpp:E3']} exExample \ref{['oLOpp:E3']} eqEquation (\ref{['oLOpp:E3']}) [] is sound. We can ensure that the angles $\beta_j$ are unique by assuming that $\beta_1,\beta_2\in [0,\pi)$ and if $\beta_2\in\{0,\frac{\pi}{2}\}$ then $\beta_1=0$, and we can ensure that the angles $\delta_j$ are unique by assuming that $\delta_1,\delta_2,\delta_3,\delta_4,\delta_5,\delta_6\in[0,\pi)$. If $\delta_3\in\{0,\frac{\pi}{2}\}$ then $\delta_1=0$, if $\delta_4\in\{0,\frac{\pi}{2}\}$ then $\delta_2=0$, if $\delta_4=0$ then $\delta_3=0$, and if $\delta_6\in\{0,\frac{\pi}{2}\}$ then $\delta_5=0$. The existence and uniqueness of such $\beta_j$ and $\delta_j$ are given by Lemmas 10 and 11 of clement2022lov.
  • Figure 5: Axioms of the ${\textbf{LO}}_{{\bm{fi}}}$-calculus. The angles of (\ref{['axiom:E2']}) and (\ref{['axiom:E3']}) are the same as in the axioms of the $\textbf{LOPP}$-calculus ( fig:nLOpp:[] defDefinition \ref{['fig:nLOpp']} figFigure \ref{['fig:nLOpp']} lemLemma \ref{['fig:nLOpp']} axiomAxiom (\ref{['fig:nLOpp']}) secSection \ref{['fig:nLOpp']} subsecSection \ref{['fig:nLOpp']} thmTheorem \ref{['fig:nLOpp']} propProposition \ref{['fig:nLOpp']} tabTable \ref{['fig:nLOpp']} oLOppEquation (\ref{['fig:nLOpp']}) nLOppEquation (\ref{['fig:nLOpp']}) appAppendix \ref{['fig:nLOpp']} remRemark \ref{['fig:nLOpp']} exExample \ref{['fig:nLOpp']} eqEquation (\ref{['fig:nLOpp']}) [] ). $h$ is any linear function $\mathcal{B}^{\text{pre}}_{2}\rightarrow\mathcal{B}^{\text{pre}}_{2}$. The conventions for $\left\{\emptyset,\left| . \right\rangle,\left| ... \right\rangle,\left\langle . \right|,\left\langle ... \right|\right\}$, and the omitted sums are detailed in def:conventions:[] defDefinition \ref{['def:conventions']} figFigure \ref{['def:conventions']} lemLemma \ref{['def:conventions']} axiomAxiom (\ref{['def:conventions']}) secSection \ref{['def:conventions']} subsecSection \ref{['def:conventions']} thmTheorem \ref{['def:conventions']} propProposition \ref{['def:conventions']} tabTable \ref{['def:conventions']} oLOppEquation (\ref{['def:conventions']}) nLOppEquation (\ref{['def:conventions']}) appAppendix \ref{['def:conventions']} remRemark \ref{['def:conventions']} exExample \ref{['def:conventions']} eqEquation (\ref{['def:conventions']}) [] . The interpretations of the axioms are given in prop:soundness:[] defDefinition \ref{['prop:soundness']} figFigure \ref{['prop:soundness']} lemLemma \ref{['prop:soundness']} axiomAxiom (\ref{['prop:soundness']}) secSection \ref{['prop:soundness']} subsecSection \ref{['prop:soundness']} thmTheorem \ref{['prop:soundness']} propProposition \ref{['prop:soundness']} tabTable \ref{['prop:soundness']} oLOppEquation (\ref{['prop:soundness']}) nLOppEquation (\ref{['prop:soundness']}) appAppendix \ref{['prop:soundness']} remRemark \ref{['prop:soundness']} exExample \ref{['prop:soundness']} eqEquation (\ref{['prop:soundness']}) [] .
  • ...and 1 more figures

Theorems & Definitions (32)

  • Definition 1
  • Example 2
  • Definition 3: Semantics of $\textbf{LOPP}$
  • Remark 4
  • Definition 5: $\textbf{LOPP}$-calculus
  • Proposition 6: Soundness of $\textbf{LOPP}$
  • Theorem 7: Completeness of $\textbf{LOPP}$
  • Theorem 8: Minimality
  • Definition 9: $\triangle$-circuits
  • Remark 10: Coefficients of $\left\llbracket \triangle \right\rrbracket_1$
  • ...and 22 more