log in  |  register  |  feedback?  |  help  |  web accessibility
Logo
Dissertation Defense: Quantum routing for architecture-respecting circuit transformations
Eddie Schoute - University of Maryland
ATL 3100A (faculty only) Virtual Via Zoom: https://umd.zoom.us/j/98433503998?pwd=eUQzKy9lS3BtSHJhR0JuLzlPZE0zUT09
Tuesday, October 12, 2021, 1:30-3:30 pm Calendar
  • You are subscribed to this talk through .
  • You are watching this talk through .
  • You are subscribed to this talk. (unsubscribe, watch)
  • You are watching this talk. (unwatch, subscribe)
  • You are not subscribed to this talk. (watch, subscribe)
Abstract

The connectivity between qubits is one of the many design aspects that go into building a quantum computer. Better connectivity makes it easier to perform arbitrary interacting operations in quantum algorithms, but it may also come with additional noise and may be costly to manufacture. Therefore, many proposals for scalable quantum computer architectures sacrifice connectivity to obtain better modularity and suppress noise. This poses a challenge to running quantum algorithms because simulating missing connectivity can come with significant overhead.

A natural stepping stone is permuting qubits on the architecture, a task we call quantum routing. We first give a rigorous analysis for the special case of classical routing using SWAP gates. Then we present a time-independent Hamiltonian protocol that reverses a chain of qubits asymptotically 3 times faster than classical routing. Using this protocol, we exhibit the first separation between classical and quantum routing time. This leads us to lower bound unitary quantum routing to be inversely proportional to the vertex expansion of the architecture graph in a gate model and inversely proportional to the edge expansion in a Hamiltonian evolution model. We rule out a superpolynomial separation between classical and quantum routing for architectures with poor expansion properties such as grid graphs.

We then show how to use routing to transform quantum circuits such that their interactions respect the architecture constraints while attempting to minimize the depth overhead. We benchmark the performance of our circuit transformations on grid and modular architectures.

Finally, we give a circuit transformation for fault-tolerant quantum computation in the surface code. We use a construction for parallel long-range operations in constant logical time that allows us to avoid the need for routing altogether. Our benchmarks show improved performance over our previous circuit transformations using classical routing.

Bio

Eddie Schoute is a PhD candidate at the Joint Center for Quantum Information and Computer Science (QuICS) supervised by prof. Andrew Childs. Before coming to Maryland, he received his B.S. Computer Science, B.S. Electrical Engineering, M.S. Computer Science, and M.S. Embedded Systems from Delft University of Technology in the Netherlands and performed research on quantum networks with Stephanie Wehner at QuTech. Eddie Schoute held a QuICS Lanczos Graduate Fellowship between 2016–2018 and was awarded the IBM PhD Fellowship for 2020–2022. His research interests are bridging the gap between quantum algorithms and quantum hardware, compilers, circuit optimization, and quantum algorithms.

This talk is organized by Andrea F. Svejda