Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Graph-Based Pulse Representation for Diverse Quantum Control Hardware

Aniket S. Dalvi, Leon Riesebos, Jacob Whitlow, Kenneth R. Brown

TL;DR

The paper addresses the challenge of representing and transforming pulse-level quantum controls across diverse hardware. It introduces pulselib, a graph-based, parametric pulse representation that preserves high-level pulse information through Scalars, Waveforms, Clocks, and Schedules, and provides compiler-like transformations and pipelines to lower descriptions to hardware. Key contributions include an AST-like DAG architecture, explicit phase synchronization via clock waveforms and clock sequences, multi-channel scheduling, and a Python implementation with visitors and maximal-munch transformations demonstrated on trapped-ion pulse schemes such as AC-Stark compensation and shelving. This approach enables flexible, target- and application-agnostic pulse descriptions that can evolve with hardware needs, improving expressiveness and transformability of complex pulse sequences.

Abstract

Pulse-level control of quantum systems is critical for enabling gate implementations, calibration procedures, and Hamiltonian evolution which fundamentally are not supported by the traditional circuit model. This level of control necessitates both efficient generation and representation. In this work, we propose pulselib - a graph-based pulse-level representation. A graph structure, with nodes consisting of parametrized fundamental waveforms, stores all the high-level pulse information while staying flexible for translation into hardware-specific inputs. We motivate pulselib by comparing its feature set and information flow through the pulse layer of the software stack with currently available pulse representations. We describe the architecture of this proposed representation that mimics the abstract syntax tree (AST) model from classical compilation pipelines. Finally, we outline applications like trapped-ion-specific gate and shelving pulse schemes whose constraints and implementation can be written and represented due to pulselib's graph-based architecture.

Graph-Based Pulse Representation for Diverse Quantum Control Hardware

TL;DR

The paper addresses the challenge of representing and transforming pulse-level quantum controls across diverse hardware. It introduces pulselib, a graph-based, parametric pulse representation that preserves high-level pulse information through Scalars, Waveforms, Clocks, and Schedules, and provides compiler-like transformations and pipelines to lower descriptions to hardware. Key contributions include an AST-like DAG architecture, explicit phase synchronization via clock waveforms and clock sequences, multi-channel scheduling, and a Python implementation with visitors and maximal-munch transformations demonstrated on trapped-ion pulse schemes such as AC-Stark compensation and shelving. This approach enables flexible, target- and application-agnostic pulse descriptions that can evolve with hardware needs, improving expressiveness and transformability of complex pulse sequences.

Abstract

Pulse-level control of quantum systems is critical for enabling gate implementations, calibration procedures, and Hamiltonian evolution which fundamentally are not supported by the traditional circuit model. This level of control necessitates both efficient generation and representation. In this work, we propose pulselib - a graph-based pulse-level representation. A graph structure, with nodes consisting of parametrized fundamental waveforms, stores all the high-level pulse information while staying flexible for translation into hardware-specific inputs. We motivate pulselib by comparing its feature set and information flow through the pulse layer of the software stack with currently available pulse representations. We describe the architecture of this proposed representation that mimics the abstract syntax tree (AST) model from classical compilation pipelines. Finally, we outline applications like trapped-ion-specific gate and shelving pulse schemes whose constraints and implementation can be written and represented due to pulselib's graph-based architecture.
Paper Structure (17 sections, 9 figures, 1 table)

This paper contains 17 sections, 9 figures, 1 table.

Table of Contents

  1. Introduction
  2. Motivation
  3. Architecture
  4. Scalars
  5. Waveforms
  6. Schedule
  7. Transformations
  8. Visitors and Transformers
  9. Pipeline
  10. Implementation
  11. Scalars and Waveforms
  12. Schedule
  13. Visitors and Transformers
  14. Applications
  15. Pulse-Dependent Clock
  16. ...and 2 more sections

Figures (9)

  • Figure 1: The envisioned workflow from graph-based user pulse descriptions to a target application.
  • Figure 2: (a) The graph for the sum of a number and a variable with key "foo." Numbers at the edges represent the operand order. (b) The graph of a sine waveform with a static frequency and phase.
  • Figure 3: Traingle frequency-modulated waveform with added Gaussian amplitude modulation sampled at 1 GHz.
  • Figure 4: The graphs and renders of two sequential waveforms with absolute and continuous phase mode.
  • Figure 5: The graph of a frequency-modulated sine waveform based on a 10 MHz carrier frequency.
  • ...and 4 more figures