Abstract: How do we build a practical quantum computer out of faulty components available in the lab? An answer that is being pursued across essentially all quantum computing platforms is to use quantum error correction (QEC). In QEC, a single “logical” qubit is encoded into a state of a large number of imperfect “physical” qubits. The expectation is that, as the number of physical qubits involved increases, the probability with which the logical qubit fails should decrease exponentially. Superconducting quantum processors that we develop at Google follow this expectation well – as long as small sizes of the QEC code are concerned. At the same time, our team has recently discovered that for large code sizes, the exponential decay stops; the failure probability ceases to reduce further with the increase of the number of physical qubits. The failure probability “floor” reflects the existence of rare catastrophic errors that affect the quantum processor as a whole. We know the origin of these catastrophic events: they are caused by impacts of high-energy radioactive particles with the device (e.g., coming from space). But how can we circumvent them?

In this talk, I will tell you about our work at Google aimed at understanding the physics of the impact events, and explain the hardware mitigation strategies we are developing to suppress them.