Ioana Moflic and Alexandru Paler of Aalto University have developed Pandora, a new software tool capable of handling quantum circuits containing billions of gates. The tool addresses a critical gap between current software limitations and the demands of complex applications like quantum chemistry and Shor’s algorithm. Researchers demonstrated Pandora’s capabilities by fully compiling circuits for a Fermi-Hubbard 100×100 model and a 1024-bit Shor’s algorithm. Compared to TKET and Qiskit, Pandora exhibits a performance advantage when manipulating circuits exceeding 10,000 gates, opening “novel paths in quantum software.”
Pandora: Scaling Quantum Circuit Compilation and Manipulation
Pandora achieves circuit compilation for quantum systems containing billions of gates, a feat previously unattainable with existing software tools. The software’s architecture uses relational database management systems to overcome limitations in scaling quantum circuit manipulation, a critical hurdle for practical applications. This capability extends beyond mere compilation. Pandora also facilitates the analysis of these extraordinarily large circuits. “The performance and versatility of Pandora open novel paths in quantum software,” the authors state, highlighting the tool’s potential to drive innovation in the field.
Pandora uses a unique internal structure, representing quantum circuits as doubly linked lists, which allows for in-place processing and optimization through custom rewrite rules applied via pattern matching. Pandora can easily handle billions of gates and can stream circuit partitions in resource estimation pipelines at very high rates. This performance is achieved by storing detailed information about each gate, including its ID, previous and next gate connections for each qubit acted upon, and a unique label for circuit identification.
Each gate within the circuit is uniquely identified by its ID value, while the overall circuit is identified by its label, facilitating efficient tracking and manipulation. The structure of the doubly linked list allows for rapid traversal and modification of the circuit, as each gate points directly to its neighbors for each qubit it affects. For equivalence checking tasks, Pandora outperforms MQT.QCEC on specific circuits that have more than 32 qubits. “We return the generated partitions,” the authors state, highlighting the system’s ability to deliver processed circuit segments for further analysis.
Pandora’s Multi-threading and HPC-Ready Map-Reduce Model
Pandora achieves substantial performance gains through a map-reduce programming model, enabling the distribution of computational tasks across multiple nodes within a high-performance computing system. This architecture allows extremely large quantum circuits to be compiled and manipulated by dividing the workload, with each node running a Pandora container and using multi-threading for increased speed. The system’s design incorporates both rewrite threads, which execute circuit optimization rules, and insertion/extraction threads, responsible for importing and exporting circuit data and measuring associated costs. The efficiency of this approach stems from the use of a relational database management system as the core engine, reducing search times and minimizing the sequential portion of computation. This allows Pandora to achieve near-ideal speed-ups during parallel circuit transpilation, and maintain good performance even when applying templates to a substantial percentage of the total gate count.
The system’s architecture facilitates the caching of subcircuits and partitions, further enhancing performance and scalability. Containers are configured to support a distributed model, allowing for the parallel processing of complex quantum algorithms. This multi-threaded, HPC-ready design has demonstrably improved compilation times for demanding quantum simulations. For example, the team compiled utility-scale 100×100 Fermi-Hubbard instances, a computationally intensive task for quantum chemistry, in a matter of hours. The concurrent operation of decomposition methods alongside Pandora’s insertion processes further contributes to these performance improvements. “The decomposition method runs concurrently to the Pandora insertion, and this achieves performance speedups,” the authors state, highlighting the synergistic effect of these parallel processes. The ability to efficiently handle such large circuits is enabled by a batch processing system for gate insertion and extraction.
This system collects a fixed-size batch of gates, processes them as a single unit, and then frees the associated memory before preparing the next batch. This approach minimizes memory overhead and maximizes throughput. The system also provides a solid foundation for developing scalable and consistent circuit equivalence checkers.
Performance Gains Over TKET and Qiskit for Circuit Manipulation
Pandora achieves demonstrable speedups when optimizing quantum circuits, particularly when searching for infrequently occurring patterns, a common challenge in quantum algorithm design. This performance advantage extends to practical applications beyond randomized benchmarks. The system’s architecture is well-suited for optimizing adders by canceling gates at random positions within the circuit. Analysis of these compilations confirmed that the software’s high-performance computing capabilities allow for efficient processing of complex quantum simulations, a key step toward tackling real-world problems. Beyond compilation speed, Pandora also excels at equivalence checking, a vital process for verifying the correctness of quantum circuits. “Pandora becomes especially useful when the patterns we search for are rarely encountered in the circuit – and this is the case almost all the time for quantum circuit optimisation,” the authors state, highlighting the software’s unique strength in handling challenging optimization tasks. Performance analysis clearly demonstrates a benefit over contemporary frameworks for circuits exceeding 10,000 gates and for equivalence checking on circuits with more than 32 qubits, establishing Pandora as a significant advancement in quantum circuit manipulation.
The approach employed by the team focuses on determining equivalence without aiming for a universally complete method, instead demonstrating the potential of a multi-threaded, rewrite-based system. Pandora can easily handle billions of gates and can stream circuit partitions in resource estimation pipelines at very high rates.
Full Compilation of Fermi-Hubbard 100×100 and 1024-bit Shor’s Algorithm
Pandora, a new software tool, successfully compiled a 100×100 Fermi-Hubbard model circuit, a feat demonstrating its capacity to manage circuits important for simulating complex quantum chemistry systems, a domain where existing software often struggles. The ability to efficiently process these circuits stems from the software’s design, which focuses on circuit rewrites and a high-performance computing architecture. The software’s performance extends to algorithms with significant computational demands, including a full compilation of a 1024-bit Shor’s algorithm circuit.
Despite this, the system’s ability to handle circuits containing billions of gates remains a key advancement. “The hierarchic structure of the high-level circuits can be used for optimization purposes: a more efficient adder will automatically result in a more efficient implementation of Shor’s algorithm,” the researchers note, highlighting the software’s ability to exploit modularity within complex algorithms.
The software’s design also facilitates high-throughput circuit partitioning, a technique that uses the modular nature of algorithms like Shor’s to improve optimization.
TEXT TO CORRECT:
Multi-threading results closely mirrored ideal speed-up expectations, indicating a highly scalable architecture.
Output ONLY the corrected text:
Pandora’s Circuit Rewrite Engine Enables High-Speed Compilation & Optimization
The software’s architecture supports multi-threading for circuit operations, enabling parallel processing of rewrites and significantly reducing the time spent searching for specific gate patterns. Threads were assigned individual rewrites, mirroring a quantum circuit optimization approach, though the focus remained on pure rewrite speed rather than complex optimization heuristics. The results demonstrate that Pandora’s advantage arises from fast pattern search times, a critical factor for handling increasingly complex circuits.
The ability to quickly identify and manipulate gate patterns is a fundamental step towards handling the demands of future quantum algorithms and simulations. The researchers state that Pandora was made HPC-ready by supporting the map-reduce programming model, highlighting the system’s adaptability to large-scale computational environments.
See today’s quantum computing news on Quantum Zeitgeist for the latest breakthroughs in qubits, hardware, algorithms, and industry deals.
