Assistant Professor Mohammad Mirhosseini (Caltech EE/APh) explains how his group built a mechanical quantum memory that stores microwave-photon quantum states far longer than typical superconducting qubits, and why that matters for hybrid quantum architectures. The discussion covers microwave photons, phonons, optomechanics, coherence versus lifetime (T2 vs. T1), current speed bottlenecks, and implications for quantum transduction and error mechanisms. The discussion centers on a paper from Mirhosseini's paper from December of 2024 titled, “A mechanical quantum memory for microwave photons,” detailing strong coupling between a transmon and a long‑lived nanomechanical oscillator for storage and retrieval of nonclassical states.

Guest

Mohammad Mirhosseini is an Assistant Professor of Electrical Engineering and Applied Physics at Caltech, where his group engineers hybrid superconducting–phononic–photonic systems at millikelvin temperatures for computing, communication, and sensing. He completed his PhD at the University of Rochester’s Institute of Optics and was a postdoc in Oscar Painter’s group at Caltech before starting his lab. His recent team effort demonstrates mechanical oscillators as compact, long‑lived quantum memories integrated with superconducting circuits.

Key topics

• What “microwave photons” are and how qubits emit/absorb single microwave photons in circuit QED analogously to atoms and optical photons.
• Why “memory” is missing in today’s quantum processors and how a dedicated long‑lived storage element can complement fast but dissipative superconducting qubits.
• Optomechanics 101: mapping quantum states between electrical and mechanical degrees of freedom, with phonons as the quantized vibrational excitations.
• T1 vs. T2: demonstrated order‑of‑magnitude gains in lifetime (T1) and more modest current gains in coherence (T2), plus paths to mitigate dephasing.
• Present bottleneck: state conversion between qubit and oscillator is about 100× slower than native superconducting operations, with clear engineering avenues to speed up.
• Quantum transduction: leveraging the same mechanical intermediary to bridge microwave and optical domains for interconnects and networking.
• Two‑level system (TLS) defects: shared decoherence mechanisms across mechanical oscillators and superconducting circuits and why comparing both can illuminate materials limits.

Why it matters

Hybrid architectures that pair fast processors with long‑lived memories are a natural route to scaling, and mechanical oscillators offer lifetimes far exceeding conventional superconducting storage elements while remaining chip‑integrable.. Demonstrating nonclassical state storage and retrieval with strong qubit–mechanics coupling validates mechanical oscillators as practical quantum memories and sets the stage for on‑chip transduction. Overcoming current speed limits and dephasing would lower the overhead for synchronization, buffering, and possibly future fault‑tolerant protocols in superconducting platforms.

Episode highlights

• A clear explanation of microwave photons and how circuit QED lets qubits create and absorb them one by one.
• Mechanical memory concept: store quantum states as phonons in a gigahertz‑frequency nanomechanical oscillator and read them back later.
• Performance today: roughly 10–30× longer T1 than typical superconducting qubits with current T2 gains of a few×, alongside concrete strategies to extend T2.
• Speed trade‑off: present qubit–mechanics state transfer is ~100× slower than native superconducting gates, but device design and coupling improvements are underway.
• Roadmap: tighter coupling for in‑oscillator gates, microwave‑to‑optical conversion via the same mechanics, and probing TLS defects to inform both mechanical and superconducting coherence.

In this episode, host Sebastian Hassinger sits down with Xiaodi Wu, Associate Professor at the University of Maryland, to discuss Wu’s journey through quantum information science, his drive for bridging computer science and physics, and the creation of the quantum programming language SimuQ.

Guest Introduction

• Xiaodi Wu shares his academic path from Tsinghua University (where he studied mathematics and physics) to a PhD at the University of Michigan, followed by postdoctoral work at MIT and a position at the University of Oregon, before joining the University of Maryland.
• The conversation highlights Wu’s formative experiences, early fascination with quantum complexity, and the impact of mentors like Andy Yao.

Quantum Computing: Theory Meets Practice

• Wu discusses his desire to blend theoretical computer science with physics, leading to pioneering work in quantum complexity theory and device-independent quantum cryptography.
• He reflects on the challenges and benefits of interdisciplinary research, and the importance of historical context in guiding modern quantum technology development.

Programming Languages and Human Factors

• The episode delves into Wu’s transition from theory to practical tools, emphasizing the major role of human factors and software correctness in building reliable quantum software.
• Wu identifies the value of drawing inspiration from classical programming languages like FORTRAN and SIMULA—and points out that quantum software must prioritize usability and debugging, not just elegant algorithms.

SimiQ: Hamiltonian-Based Quantum Abstraction

• Wu introduces SimuQ, a new quantum programming language designed to treat Hamiltonian evolution as a first-class abstraction, akin to how floating-point arithmetic is fundamental in classical computing.
• SimiQ enables users to specify Hamiltonian models directly and compiles them to both gate-based and analog/pulse-level quantum devices (including IBM, AWS Braket, and D-Wave backends).
• The language aims to make quantum simulation and continuous-variable problems more accessible, and serves as a test bed for new quantum software abstractions.

Analog vs. Digital in Quantum Computing

• Wu and Hassinger explore the analog/digital divide in quantum hardware, examining how SimuQ leverages the strengths of both by focusing on higher-level abstractions (Hamiltonians) that fit natural use cases like quantum simulation and dynamic systems.

Practical Applications and Vision

• The conversation highlights targeted domains for SimuQ, such as quantum chemistry, physics simulation, and machine learning algorithms that benefit from continuous-variable modeling.
• Wu discusses his vision for developer-friendly quantum tools, drawing parallels to the evolution of classical programming and the value of reusable abstractions for future advancements.

Listen to The New Quantum Era podcast for more interviews with leaders in quantum computing, software development, and scientific research.