QuEra Computing

‼️ Determining and addressing performance drops will be critical for the applicability of the approach at scale, a problem that may be faced via improvements in the classical reduction and utilization of techniques like qubit encoding compression or problem decomposition.

When looking at the results, qReduMIS produces a 100% success rate for "easy" problems in terms of a hardness measurement H. However, when looking at "harder" problems, including larger and more complex graphs, the success rate drops to 89% - 95%.

The efficiency is derived from classical kernelization used in tandem with high-accuracy QPU measurements. By identifying and removing vertices that have a high or low likelihood of being part of the solution, the QPU allows for further accuracy in kernelization.

💡 In this work, the team proposes and implements a new quantum-informed reduction algorithm for the MIS problem: qReduMIS. This approach integrates classical techniques with QuEra's Aquila QPU as a co-processor to utilize strengths from each paradigm.

The maximum independent set (MIS) problem, a well-known graph optimization problem in classical computing, has no proven algorithms to solve the problem in polynomial time for arbitrary graph structures.

qReduMIS: A Quantum-Informed Reduction Algorithm for the Maximum... We propose and implement a quantum-informed reduction algorithm for the maximum independent set problem that integrates classical kernelization techniques with information extracted from quantum...

Researchers from our partners Amazon Web Services (AWS) and JPMorganChase present

"qReduMIS: A Quantum-Informed Reduction Algorithm for the Maximum Independent Set Problem"

🔗 arXiv: arxiv.org/abs/2503.1...

‼️ Determining and addressing performance drops will be critical for the applicability of the approach at scale, a problem that may be faced via improvements in the classical reduction and utilization of techniques like qubit encoding compression or problem decomposition.

When looking at the results, qReduMIS produces a 100% success rate for "easy" problems in terms of a hardness measurement H. However, when looking at "harder" problems, including larger and more complex graphs, the success rate drops to 89% - 95%.

The efficiency is derived from classical kernelization used in tandem with high-accuracy QPU measurements. By identifying and removing vertices that have a high or low likelihood of being part of the solution, the QPU allows for further accuracy in kernelization.

💡 In this work, the team proposes and implements a new quantum-informed reduction algorithm for the MIS problem: qReduMIS. This approach integrates classical techniques with QuEra's Aquila QPU as a co-processor to utilize strengths from each paradigm.

The maximum independent set (MIS) problem, a well-known graph optimization problem in classical computing, has no proven algorithms to solve the problem in polynomial time for arbitrary graph structures.

Mathematically, they demonstrate that this method can reach the minimum of loss with a constant probability.

❗However, one possible drawback is the need to encode and store the parameters needed during inference implicitly in a quantum state.

💡 The authors propose encoding inputs (e.g., images) unit-by-unit and then taking mid-circuit measurements and using post-selection procedures (instead of feedback control) to train their models.

🤔 A great challenge in the field of quantum machine learning is to devise protocols that bypass the need to estimate gradients of parametrized quantum circuits to train a learning model.

Quantum automated learning with provable and explainable trainability Machine learning is widely believed to be one of the most promising practical applications of quantum computing. Existing quantum machine learning schemes typically employ a quantum-classical...

"Quantum automated learning with provable and explainable trainability"

arxiv.org/abs/2502.0...

This paper proposes a new gradient-free method to train quantum neural networks without separately encoding inputs in independent circuits, potentially leading to a massive saving of shots in practice.

Simulating Neutron Scattering on an Analog Quantum Processor Neutron scattering characterization of materials allows for the study of entanglement and microscopic structure, but is inefficient to simulate classically for comparison to theoretical models and...

The results achieved with Aquila successfully resolve the key peak features expected in neutron scattering, as benchmarked by a study involving a chain of up to 25 qubits. Read the complete paper on arXiv! arxiv.org/abs/2410.0...
3/3

These measurements are directly relevant for comparing quantum simulation analyses with neutron-scattering experiments, a common technique in the study of quantum magnetism.
2/3

QuEra’s Aquila, currently the only neutral-atom quantum processor available on a public cloud, continues to demonstrate its value across various applications. In this work, a team led by George Siopsis showcases how to measure the dynamic structure factor of a magnetic system using Aquila.
1/3

The advantages of zone-based architectures over monolithic processing systems are further validated, with the compiler predicting fidelity improvements exceeding a factor of 20 in benchmarking tests.
3/3

In this work, a UCLA-based team led by Jason Cong continues to enhance their compilation stack, introducing processes to minimize data movement between zones (boosting speed and fidelity), increase the flexibility of zone architecture, and improve scheduling strategies.
2/3

Reuse-Aware Compilation for Zoned Quantum Architectures Based on... Quantum computing architectures based on neutral atoms offer large scales and high-fidelity operations. They can be heterogeneous, with different zones for storage, entangling operations, and...

A paper that caught our attention: hubs.ly/Q02_Hm8D0

Thanks to coherent atomic shuttling, the exploration space for optimal algorithm compilation in neutral-atom quantum computers remains a promising area of innovation.
1/3

given that several neutral-atom species are fermionic.

This work addresses the uncertainty by demonstrating how to perform error correction with fermionic qubits, establishing clear performance improvement bounds for natively fermionic algorithms, and highlighting the potential of neutral-atom fermionic computers,

The mapping between conventional qubits and fermions is inefficient, posing a significant challenge. However, this challenge can be mitigated if qubits themselves exhibit fermionic behavior. Until now, it was uncertain whether scalable quantum computing was feasible in such scenarios.

Fermion-qubit fault-tolerant quantum computing Simulating the dynamics of electrons and other fermionic particles in quantum chemistry, material science, and high-energy physics is one of the most promising applications of fault-tolerant...

A paper that caught our attention arXiv: hubs.ly/Q02_Hkj00

Most of the anticipated high-impact applications of quantum computing in the field of quantum simulation involve degrees of freedom known as fermions, such as electrons in materials and quarks in high-energy systems.

Careers Join our world-class team of scientists, engineers, and business people and help us make a dent in the world. Join us.

If you’d like to work on similar frontier questions, check out our openings: quera.com/careers.

This work reflects the efforts of the entire QuEra team and collaborators at Harvard and MIT. Special thanks to the theory team: Sunny, Casey, Chen, Harry, and the Gemini experimental team: Pedro, John, Niki, and Sergi

There is a long road ahead, but this experiment underscores the potential of neutral atoms for advancing universal quantum computation. The evolving landscape of magic state preparation methods also provides ample opportunities for exploration.

Our results highlight a key building block for fault-tolerant quantum computing. While progress is exciting, further fidelity improvements are needed to enable multiple distillation rounds and reduce overheads.

To the best of our knowledge, this is also the largest color code that has been demonstrated to date.