Alexander Tyner

Accelerated characterization of two-level systems in superconducting qubits via machine learning

  1. Avinash Pathapati,
  2. Olli Mansikkamäki,
  3. Alexander Tyner,
  4. and Alexander V. Balatsky
We introduce a data-driven approach for extracting two-level system (TLS) parameters-frequency ωTLS, coupling strength g, dissipation time TTLS,1, and the pure dephasing time TϕTLS,2,
labelled as a 4-component vector q⃗ , directly from simulated spectroscopy data generated for a single TLS by a form of two-tone spectroscopy. Specifically, we demonstrate that a custom convolutional neural network model(CNN) can simultaneously predict ωTLS, g, TTLS,1 and TϕTLS,2 from the spectroscopy data presented in the form of images. Our results show that the model achieves superior performance to perturbation theory methods in successfully extracting the TLS parameters. Although the model, initially trained on noise-free data, exhibits a decline in accuracy when evaluated on noisy images, retraining it on a noisy dataset leads to a substantial performance improvement, achieving results comparable to those obtained under noise-free conditions. Furthermore, the model exhibits higher predictive accuracy for parameters ωTLS and g in comparison to TTLS,1 and TϕTLS,2.
arxiv pdf

Two-tone spectroscopy for the detection of two-level systems in superconducting qubits

  1. Olli Mansikkamäki,
  2. Alexander Tyner,
  3. Alexander Bilmes,
  4. Ilya Drozdov,
  5. and Alexander Balatsky
Two-level systems (TLS) of unclear physical origin are a major contributor to decoherence in superconducting qubits. The interactions of individual TLS with a qubit can be detected
via various spectroscopic methods, most of which have relied on the tunability of the qubit frequency. We propose a novel method that requires only a microwave drive and dispersive readout, and thus also works fixed-frequency qubits. The proposed two-tone spectroscopy involves a microwave pulse of varying frequency and length to excite TLSs of unknown frequencies, followed by a second pulse at the qubit frequency. TLS parameters can be estimated from the qubit population as a function of the first pulse frequency and length.
arxiv pdf