Scalable Design-Program-Compilation Optimizations for Quantum Algorithms: Using Quantum Neural Network as a Case Study
TimeMonday, July 11th1:30pm - 5pm PDT
Location3004, Level 3
DescriptionAlong with the rapid development of quantum computers, using only 5 years to scale up the number of qubits from 5 to 127 in IBM Quantum, it provides opportunities for more applications to take full use of powerful quantum computers. Along with the increasing numbers of qubits, saying IBM plans to debut quantum computers with more than 1,000 qubits in 3 years, new challenges and questions are posed in designing, programming, synthesizing, and mapping applications to quantum computers at scale:
(1) how to synthesize and map (i.e., compile) the logical circuit to physical qubits; (2) how to program applications to adapt to quantum computing; and (3) how to design a quantum circuit with potential quantum advantage? This tutorial is composed of three talks to address all the above issues. We will start from a specific task, i.e., compilation. Based on the team’s previous work, OLSQ , we will show how to automatically optimize and transform quantum programs to meet hardware limitations to respond
to issue (1). Then, we will introduce how to systematically derive new large-scale quantum program optimizations, including the projection-based assertion and Paulihedral compiler to address the issue (2). Finally, toward the quantum advantage in issue (3), we will narrow our focus down to a case-study application, i.e., quantum neural network. We will demonstrate how to design, program, and compile the quantum neural network, which is based on the team’s recent work QuantumFlow , published at
Nature Communications. In the above three talks, the hands-on experience could be gained in optimizing quantum circuits to physic qubits and implementing the neural network on the quantum circuit through the on-site coding demonstration. All attendees will leave with code examples that they can use as the backbone implementation to their own projects.