Research

Quantum error suppression

Full fault-tolerant quantum error correction architectures like the surface code are extremely resource-intensive: at current error rates, they require thousands of physical qubits and operations in order to encode a single fault-tolerant logical qubit. However, there are many techniques that can suppress errors—reduce error rates, or prevent certain types of errors from occurring—using very few resources. We are working on optimizing and combining many of these techniques, including dynamical decoupling, decoherence-free subspace encodings, bath engineering, and novel error-protected qubit designs. One of these designs is the trimon, a 3-qubit circuit with strong all-to-all dispersive couplings. Much of this work is done in collaboration with Daniel Lidar’s group at USC and R. Vijayaraghavan’s group at the Tata Institute.

We are also working with the Lidar group and Aashish Clerk’s group at the University of Chicago to better characterize quantum errors. This includes work on better understanding protocols like randomized benchmarking, developing our own deterministic benchmarking procedure, and studying how maximal amounts of information can be gained about a quantum operation.

Engineering dissipation

Dissipation can cause decoherence and errors in qubits, but can also be a resource for state preparation, coherence preservation, and even bath-engineered gates. Dissipation is also at the root of the quantum-classical transition and is intricately linked to the quantum measurement problem. In order to make full use of dissipation for quantum computing and for studying the foundations of quantum mechanics, we have developed an on-demand “dissipator” device. This device consists of a parametric coupler (essentially a tunable transmon qubit) with a fast flux line for modulating the coupler frequency. A strong coupling to a 50-ohm-terminated feedline provides the dissipative element. By coupling the dissipator to another off-resonant target mode and parametrically modulating the dissipator’s resonant frequency at the detuning frequency between the two, we introduce dissipation into that mode. This allows us to:

• Tune the amount of dissipation fast in-situ, taking the target mode from high-coherence to very lossy in a few ns
• Quickly switch between different target modes simply by changing the parametric drive frequency
• Use a high-frequency dissipator as a quantum refrigerator, cooling the target mode below the dissipator temperature
• Add extra drives to pump energy into the system, engineer higher-order loss processes, or “cool” to non-thermal states.

We have already demonstrated using this device to reset a measurement cavity back to 0 photons and cool it continuously. We are now exploring how to suppress decoherence and even stabilize non-trivial states with the dissipator.

Quasiparticles in qubit circuits

Quasiparticles–single-particle excitations of the superfluid condensate–are a major source of decoherence in superconducting qubits. A quasiparticle tunneling across a Josephson junction can absorb energy from the qubit, causing relaxation; it can donate energy to the qubit, causing spurious excitation; and it can modify the qubit energy spectrum, causing dephasing. Quasiparticles can also cause loss during bulk transport, and can become trapped in sensitive circuit elements, causing further decoherence. While superconducting qubits are typically operated at temperatures low enough to completely eliminate all quasiparticles, experimental research has consistently shown significant non-thermal populations of quasiparticles.

These non-equilibrium quasiparticles may have many sources. Blackbody radiation from warmer stages of a cryostat can impact the circuit and excite quasiparticles. Likewise, cosmic radiation and nearby radioactive materials may generate quasiparticle populations that are significant. “Hot spots” in the circuit may be insufficiently thermalized and therefore generate thermal quasiparticles at a higher temperature. All of these mechanisms are likely significant and must be combatted in order to optimize qubit coherence.

In order to tackle the problem of quasiparticles, their behavior must be carefully studied. In LFL, we study quasiparticles through a few techniques. In one set of experiments, we trap quasiparticles in the internal Andreev states of weak-link Josephson junctions. These Andreev traps, when integrated into a resonant circuit, affect the resonant frequency of the circuit, enabling fast, single-shot detection of quasiparticle trapping. We use these traps to study the correlations between quasiparticles, their energy distribution, and their relaxation and recombination mechanisms. We also design qubits which are especially sensitive or insensitive to quasiparticle behavior, and measure qubit spectrum and lifetime in a variety of quasiparticle generation environments. These experiments allow us to better understand the mechanisms by which quasiparticles are created and destroyed, and to test strategies for mitigating quasiparticle-related decoherence.