AWS re:Invent 2025 -JPMorganChase & Amazon Advanced Solutions Lab drive quantum R&D on Braket-CMP327

Summary of AWS re:Invent 2025 - JPMorganChase & Amazon Advanced Solutions Lab drive quantum R&D on Braket-CMP327

Overview of Quantum Computing

• Quantum computers leverage the laws of quantum mechanics to perform computations in novel and improved ways, going beyond the limitations of classical computers
• Classical computers store information in binary bits (0 or 1), while quantum bits (qubits) can exist in superposition of 0 and 1 simultaneously
• Simulating even small quantum systems on classical computers becomes exponentially difficult as the number of qubits increases
• Quantum computers hold the promise of exponential speedups for certain problems in areas like physics, chemistry, cryptography, and optimization

Quantum R&D Approach

• The goal of quantum R&D is to lay the groundwork for future production workloads through hybrid quantum-classical workflows
• Key steps include:
  1. Identifying target use cases that are computationally hard and business-relevant
  2. Researching new quantum algorithms and benchmarking them against classical algorithms
  3. Bringing quantum technology from a niche to broad adoption
• AWS offers tools and services like Amazon Braket to provide access to various quantum hardware and programming models
• The Amazon Advanced Solutions Lab engages in custom R&D to develop novel quantum algorithms and deliver business value with quantum-inspired classical algorithms

Collaboration between JPMorgan Chase and AWS

• Finance is rich in computationally hard optimization problems, such as portfolio selection, pricing and hedging of options, risk management, and asset-liability management
• The portfolio selection problem can be framed as a maximum independent set (MIS) problem, which is known to be computationally hard
• Prior research demonstrated quantum speedups for solving the MIS problem with up to 289 qubits
• Through their collaboration, JPMorgan Chase and AWS have published four scholarly papers and developed a suite of tools for solving large, hard problems on current quantum hardware

The Q-REDOMAS Algorithm

• Q-REDOMAS is a hybrid quantum-classical algorithm for solving the MIS problem
• The algorithm has two main steps:
  1. Classical reduction: Use a "Pac-Man" algorithm to reduce the problem size by identifying and removing exposed corner nodes
  2. Quantum sampling: Use a quantum computer as a probabilistic sampling machine to identify high-likelihood nodes for the final solution
• The classical reduction step significantly reduces the problem size, allowing the quantum computer to focus on the "hard kernel" of the problem
• Experiments on the Aquila quantum device from Qerashow that Q-REDOMAS outperforms standard quantum annealing algorithms, maintaining high success probability even for large and hard problem instances

Results and Impact

• Q-REDOMAS was tested on real quantum hardware accessed through Amazon Braket, with experiments involving over 200 qubits
• The algorithm demonstrated a significant performance boost compared to quantum annealing, maintaining a success probability above 89% even for the hardest problem instances
• The hybrid quantum-classical approach allows Q-REDOMAS to solve large-scale portfolio selection problems that would be intractable for classical computers alone
• This work showcases how quantum R&D can lead to practical solutions for industry-relevant optimization problems in finance and beyond

Future Outlook

• The Q-REDOMAS algorithm is hardware-agnostic, allowing it to be tested on various quantum platforms beyond the Aquila device
• The team plans to explore other applications of the hybrid quantum-classical framework beyond portfolio optimization
• Continued collaboration between industry leaders like JPMorgan Chase and quantum research experts at AWS aims to drive the advancement and real-world adoption of quantum computing