Heralded high-dimensional module-based quantum computation

Technical Summary: Heralded High-Dimensional Module-Based Quantum Computation

Problem Statement
While quantum computation offers advantages over classical counterparts, realizing universal quantum gates, particularly the controlled-NOT (CNOT) gate, remains a significant challenge. Existing architectures for high-dimensional quantum systems (qudits, where $d > 2$) often rely on post-selection, which limits efficiency, or require complex multi-qubit entangled states. Although high-dimensional systems offer benefits such as increased channel capacity and noise resilience, theoretical and experimental investigations into scalable, deterministic architectures for high-dimensional universal gates have lagged behind two-dimensional (qubit) systems. Specifically, there is a need for deterministic, heralded schemes that do not consume computational qudits and are insensitive to the dimensionality of the basis.

Methodology
The authors propose a scheme to construct deterministic, high-dimensional generalized CNOT gates using generalized parity modules as fundamental building blocks. The methodology proceeds in three main stages:

1. Development of Generalized Parity Modules: The authors define a generalized parity module that takes two input qudits $|i\rangle$ and $|j\rangle$ and an ancilla $|0\rangle$, outputting the state $|ij\rangle|(j \ominus i) \mod d\rangle$. They identify that out of $(d-1)!$ possible generalized parity "sorters" (permutations of the second index), only two specific forms are universal for quantum computing: the subtraction-based module $P = (j \ominus i) \mod d$ and the addition-based module $P = (j \oplus i) \mod d$.
2. Construction of the Generalized CNOT Gate: Utilizing these parity modules, the authors design a circuit (illustrated in Fig. 1) to implement a two-qudit generalized CNOT gate. The procedure involves:
   • Injecting control and target qudits into a first parity module ($P_1$) alongside an ancilla prepared in a balanced superposition state.
   • Applying single-qudit Fourier transformations ($F_d$) and a second parity module ($P_2$) to the ancilla and target.
   • Measuring the ancilla in the computational basis.
   • Performing classical feed-forward operations on the control and target qudits based on the measurement outcomes of $P_1$, $P_2$, and the ancilla. These operations correct the phase and permutation errors to yield the desired CNOT output: $|x\rangle_c |(x \oplus y) \mod d\rangle_t$.
3. Optical Implementation: The authors propose a specific optical architecture for implementing the generalized parity module using orbital angular momentum (OAM) degrees of freedom of single photons. This setup employs:
   • A six-photon ancillary state prepared via "Entanglement by Path Identity."
   • First-order ($PS_1$) and second-order ($PS_2$) parity sorters based on Mach-Zehnder interferometers containing Dove prisms to route photons based on OAM parity.
   • Photon-number-resolving detectors and single-photon detectors to herald the success of the operation.

Key Results

• Deterministic and Heralded Operation: The proposed scheme is deterministic in principle (the computational qudits are not consumed) and heralded. Success is signaled by specific detector clicks (absence of clicks in specific single-photon detectors $D_1$ and $D_2$), eliminating the need for post-selection of the computational state.
• Dimensional Independence: The protocol is insensitive to the dimensionality $d$ of the computing basis. The authors demonstrate the construction for qutrits ($d=3$) and provide the generalization for arbitrary $d \geq 4$.
• Identification of Universal Sorters: The paper establishes that among the $(d-1)!$ possible parity sorters, only $P = (j \ominus i) \mod d$ and $P = (j \oplus i) \mod d$ serve as universal building blocks for quantum computing. Other permutations may be useful for specific tasks like entangled state generation but are not sufficient for universal computation.
• Feasibility Analysis: The optical implementation relies on existing technologies, including programmable linear optical platforms, Dove prism-based parity sorters, and high-fidelity single-qudit gates. The authors acknowledge practical limitations such as spatial-temporal mismatch, technical imperfections in sorters, and detector dark counts, but maintain that the protocol is experimentally feasible.

Significance and Claims
The paper claims to provide a versatile framework for high-dimensional quantum computing that addresses the scalability and efficiency issues of previous approaches. By utilizing generalized parity modules, the scheme achieves:

• Resource Efficiency: It requires only one additional ancilla qudit and does not necessitate large-scale entangled states.
• Non-destructive Measurement: The parity module is implemented via quantum nondemolition measurements, preserving the input states for further processing.
• Modularity: The parity modules can serve as building blocks not only for quantum computation but also for quantum communication tasks.

The authors position their work as a step toward scalable high-dimensional quantum architectures, offering a deterministic alternative to probabilistic, post-selected methods. They emphasize that while the optical implementation is detailed for OAM-encoded photons, the underlying concepts are generalizable to other physical platforms.