QCrank Protocol for DPQAs: Decoding Quantum Algorithms
DPQA QCrank Protocol
Optimisation of Neutral Atom Qubit Array Algorithm Deployment using QCrank Encoding Protocol
Creating quantum hardware instructions from quantum algorithms is still tricky. Anupam Mitra, Wan-Hsuan Lin, and Jan Balewski from NERSC, Lawrence Berkeley National Laboratory, and UC Los Angeles, along with QuEra and Harvard colleagues, created an innovative solution.
A compilation method for QCrank, an encoding protocol for dynamically programmable qubit array (DPQA) based on neutral atoms, is examined. It stores classical data in a quantum state. This work reveals how to better deploy algorithms and achieve encouraging accuracy when writing and retrieving classical data using these arrays' unique architecture, which includes high qubit counts, parallel operation, and changeable connection.
By comparing performance to current quantum processors, the team shows how these dynamically programmable arrays could be scalable and effective for near-term quantum computing applications.
Parallel Control and Neutral Atom Qubit Systems
Neutral-atom technologies are promising platforms for quantum computers as experimental systems approach hundreds of qubits and attain competitive two-qubit gate fidelities of 99.5%. Dynamically programmable qubit arrays (DPQA) have discrete functional zones, reconfigurable qubit connections, and global laser beams to operate on many qubits simultaneously.
These powers derive from manipulating atoms without losing quantum information. DPQAs allow real-time qubit connection and operation changes during computation, unlike fixed-configuration arrays.
A single optical pulse physically moves atoms while keeping their quantum state and applies two-qubit gates between close-together atoms to selectively operate on qubit subsets. The quantum processor can address individual qubits, giving it more control. To maximise quantum circuit compilation and execution for DPQA, researchers are finding the optimal number of functional zones, designing the best qubit arrangement for these zones, and enhancing gate application while restricting atom movement.
Evaluating the accuracy of QCrank implementation for sequenced data encoding on a simulated DPQA, IBM Heron superconducting devices, and Quantinuum's H1-1E trapped-ion device emulation, with realistic hardware constraints.
QuEra's Gemini-class quantum computers and Harvard's Rb atom-based devices model neutral-atom DPQA. Modern neutral-atom quantum computers use spatial light modulators (SLMs) to load atoms into laser traps, creating static trapping configurations that can be separated into zones. One zone may have tightly connected trapping sites for entangling processes and another a dense, regular square lattice for qubits. Single-qubit gates can be implemented locally or globally using laser beams. Acoustic-optical deflectors (AODs) can move several atoms at once, rearranging site-selected atoms between zones. Studies study atom-selective measurements in the midst of the circuit, while this model predicts global destructive measurement of all qubits.
Photonic Quantum Noise and Atom Movement Errors
Study focusses on modelling noise channels for digital photonic quantum computing architecture. Selected noise values represent current state-of-the-art, with analyses undertaken with ±30% noise levels to explore potential enhancements. A Pauli channel error depending on the amount of atom motions caused by atom shuttling increases circuit execution time. Due to long relaxation and decoherence times, movement-related delays affect result fidelity less than channel faults. A actual quantum computing job was benchmarked using the QCrank encoding approach.
QCrank encodes real-valued data sequences onto data qubits using uniformly controlled rotation gates and entangling gates and Hilbert space's exponential capacity. A QCrank circuit with na address and nd data qubits requires L two-qubit entangling gates and L single-qubit rotations to store L = nd × 2^na real-valued values. The neutral-atom native gate-set with CZ entangling gates and variable angle rotations suits this design. Entangling gates in parallel layers or identical single-qubit gates on all qubits create a quantum circuit with great execution parallelism.
Architectural reconfigurability meets address-data qubit bipartiteness. To quantify accuracy, the root mean square error (RMSE) is utilised, which accounts for QCrank circuit noise. A compiler optimisation strategy was tested using a QCrank setup with na = 4 address qubits and nd = 8 data qubits compressing 128 real values onto 12 qubits. Data qubits are divided into two na-sized portions. Application of CZ gates between address qubits and a selection of data qubits requires dynamic movement.
The approach only moves address qubits, horizontally moving cyclic permutations and vertically addressing data qubit sets. QCrank circuit performance at different input sizes was examined using Qiskit simulator with density matrix backend. The results show that longer input sequences require more entangling gates, which causes inaccuracy. To compare hardware platforms, IBM Fez with Pauli Twirling and Quantinuum noisy simulator simulations were done. The ratio of address to data qubits influences performance, with digital neutral atom and trapped-ion QPUs having identical QCrank accuracy.
Qubit Arrays Change Show Accuracy Scaling Promise
Dynamic Qubit Arrays Encourage Accuracy Scaling Quantinuum's H1-1E and IBM Fez simulations show impressive dynamically programmable qubit array accuracy scaling. These findings show that dynamically configurable qubit architectures may increase quantum computation scalability and performance.
Scaling Materials Discovery Simulations
Future research will apply these methods to larger systems and more complex geometries to replicate realistic materials with unprecedented accuracy. To increase computation efficiency and scalability, alternative numerical systems and parallelisation methods will be examined. The team wants to apply these methods to high-temperature superconductivity and energy material development.
