Quantum Error Correction in Neutral Atom Quantum Computer

该项目旨在开发一个双物种平台，用于中性原子的量子计算和模拟，为实现主动量子纠错提供一条途径，这对未来超过100量子位的扩展至关重要。由于能够独立控制物种间和物种内的相互作用，这种硬件将同时为模拟计算和仿真提供一个通用平台，为复杂的多体物理研究提供了一条途径，并增加了现实世界优化问题的多样性，这些问题可以使用中性原子硬件来解决。量子计算（QC）为信息处理提供了一种革命性的方法，为有效解决各种领域遇到的经典难题提供了一条途径，包括金融服务（例如投资组合优化）、能源（例如网络设计和能源分配）、物流（例如路线规划）和IT（例如搜索、机器学习）。使用在量子力学定律下运行的硬件也为研究物理系统创造了一个平台，比如复杂的材料和分子，早期的演示显示了材料科学和量子化学的应用，最终可以扩大规模，加速药物设计或优化航空航天和制造业的材料。虽然大规模应用将需要数千个量子比特，但在短期内，小型（100个量子比特）量子处理器将达到量子硬件能够解决即使在最大的传统超级计算机上也无法解决的问题的水平。然而，量子系统的缩放仍然是一个主要的实验挑战，在小型、最先进的离子和超导体系统中已经证明了高保真性能，但在将这种性能扩展到大约10-20个量子比特时存在重大的技术障碍。在过去的十年中，中性原子已经成为量子信息处理最有前途的平台之一，与竞争技术相比，中性原子的主要优势在于能够按实际量子计算的需要扩展到大量相同的量子位。到目前为止，几个实验已经证明了捕获> 256量子位的量子位阵列。为了耦合中性原子量子位，使用了具有极大电偶极矩的高激发里德伯态，从而产生强而可控的相互作用。这些可以用于执行高保真多量子位门，其中F>0.95用于两个量子位，F>0.995用于多量子位门的固有保真度，或者用于执行研究材料或解决优化问题所需的可控自旋模型的量子模拟。虽然已经取得了重大的实验进展，但目前仍有许多挑战限制了基于单一原子种类的硬件扩展到更大的阵列尺寸。第一个原因是由于与背景原子的碰撞使原子从阱中弹出而导致真空寿命有限。对于室温操作，对于1个原子，这通常是10s，但是对于1000个原子阵列，这意味着只有10ms。这可以通过在低至4k的低温下运行来解决，在低温下，冷表面会导致寿命显著增加，高达bbb6000秒，这意味着即使对于1000个原子，也可以恢复>6s倍。下一个问题是中性原子量子位的读出时间长，通常需要10-50毫秒来读出量子位状态。对于单一种类，串扰和散射光平均读出对整个阵列具有破坏性，没有明确的途径来执行纠错以达到容错操作所需的局部测量。

This project seeks to develop a dual-species platform for quantum computing and simulation with neutral atoms, providing a route to implementing active quantum error correction essential for future scaling beyond 100 qubits. This hardware will simultaneously provide a versatile platform for analogue computing and simulation due to the ability to independently control inter- and intra-species interactions, providing a route to performing studies of complex many-body physics as well as increasing the diversity of real-world optimisation problems that can be tackled using neutral atom hardware.Quantum computation (QC) offers a revolutionary approach to information processing, providing a route to efficiently solve classically hard problems encountered across a diverse range of sectors, including financial services (e.g. portfolio optimisation), energy (e.g. network design and energy distribution), logistics (e.g. route planning), and IT (e.g. search, machine learning). Using hardware operating under the laws of quantum mechanics also creates a platform for studying physical systems, such as complex materials and molecules, with early demonstrations showing applications in materials science and quantum chemistry that could eventually be scaled up to accelerate drug design or optimised materials for aerospace and manufacturing.Whilst large-scale applications will require thousands of qubits, in the near-term small (100 qubit) quantum processors will reach a regime in which the quantum hardware is able to solve problems not accessible even on the largest available conventional supercomputers. However, scaling of quantum systems remains a major experimental challenge, with high-fidelity performance demonstrated for small, state-of-the-art ion and superconductor systems but with significant technical barriers to extending this performance beyond around 10-20 qubits. Over the last decade, neutral atoms have emerged as one of the most promising platforms for quantum information processing, with a major advantage over competing technologies arising from the ability to scale to large numbers of identical qubits as required for performing practical quantum computing. To date, several experiments have demonstrated trapping of qubit arrays with > 256 qubits. To couple neutral atom qubits, highly excited Rydberg states are used which have extremely large electric dipole moments giving rise to strong and controllable interactions. These can be exploited to perform high fidelity multi-qubit gates, with F>0.95 demonstrated for two qubits and intrinsic fidelities of F>0.995 for multi-qubit gates, or for performing quantum simulation of controllable spin models as required for studying materials or solving optimisation problems.Whilst there has been significant experimental progress, a number of challenges currently limit scaling to larger array sizes for hardware based on a single atomic species. The first arises from finite vacuum lifetime due to collisions with background atoms ejecting atoms from the trap. For room temperature operation, this is typically 10s for 1 atom but means only 10ms for a 1000 atom array. This can be solved by moving to operation at cryogenic temperatures down to 4 K where the cold surfaces cause significant increase in lifetime upwards of > 6000 seconds meaning recovery of times > 6s even for 1000 atoms. The next issue lies in the long readout time for neutral atom qubits, typically requiring 10-50 ms to readout qubit states. With a single species, the cross-talk and scattered light mean readout is destructive across the whole array, with no clear pathway to performing local measurements required for error correction to reach fault tolerant operation.