High-performance Python simulations for studying symmetry-constrained quantum many-body dynamics, subsystem equilibration, and projected ensemble universality.
Based on:
R.-A. Chang et al.,
Deep Thermalization under Charge-Conserving Quantum Dynamics,
PRX Quantum 6, 020343 (2025)
In modern quantum simulators, we can measure the environment of a subsystem rather than tracing it out.
Instead of only studying the reduced density matrix ρ_A,
we condition on measurement outcomes of the environment B.
This generates a projected ensemble —
an ensemble of pure states on subsystem A, weighted by measurement probabilities.
Subsystem A interacts with environment B.
After measuring B, we obtain conditional pure states on A.
This framework goes beyond standard thermalization by analyzing the full distribution of subsystem states rather than only expectation values.
Deep thermalization refers to the emergence of universal projected ensembles in chaotic quantum dynamics.
In symmetry-free systems:
- The projected ensemble approaches the Haar distribution.
With conservation laws (e.g., U(1) charge conservation):
- The limiting ensemble acquires structured universal forms.
- Only coarse-grained conserved quantities determine equilibrium statistics.
- Microscopic circuit details become irrelevant.
This work studies how U(1) symmetry modifies deep thermalization.
We simulate charge-conserving random quantum circuits:
Key features:
- Two-site unitary gates
- Exact U(1) charge conservation
- Random circuit architecture
- Large bath limit
This architecture models:
- Symmetry-preserving quantum hardware
- Analog quantum simulators
- Charge-conserving superconducting qubit circuits
Trace-norm distance to the universal projected ensemble as a function of circuit depth for different system sizes.
Key observations:
- The distance decays rapidly in time under U(1)-symmetric chaotic dynamics.
- A finite late-time plateau appears due to finite-size effects.
- The late-time average scales as a power law in system size (inset).
This demonstrates quantitatively that:
- The projected ensemble approaches its universal limit.
- Finite-size corrections shrink systematically with increasing N.
- Symmetry-constrained dynamics still exhibit universal equilibration.
Such finite-size scaling analysis is directly relevant for benchmarking symmetry-preserving quantum processors, where conservation laws constrain accessible Hilbert space growth.
We show that projected ensembles converge to a generalized universal form determined only by:
- Initial charge distribution
- Measurement basis statistics
This demonstrates a symmetry-aware extension of standard thermalization.
We perform large-scale simulations (up to 24 qubits) and analyze:
- Trace-norm distance to limiting ensemble
- Finite-size scaling of convergence
- Hilbert-space growth under symmetry constraints
This provides quantitative insight into:
- When classical simulation remains efficient
- How conservation laws alter scrambling dynamics
- Scaling toward potential quantum advantage
Implemented in Python with:
- Charge-sector-resolved Hilbert space decomposition
- Efficient tensor contractions
- Monte Carlo sampling
- Parallelized HPC workflows (Slurm-based)
Designed for scalability and reproducibility.
Conservation laws are intrinsic to physical quantum systems.
Understanding symmetry-constrained dynamics is relevant for:
- Hardware benchmarking under physical constraints
- Diagnosing scrambling and equilibration
- Designing symmetry-aware algorithms
- Identifying classical-to-quantum crossover regimes
This repository provides a classical simulation platform for studying these effects quantitatively.
src/ # Core simulation code
figures/ # Diagrams and result plots
notebooks/ # Analysis scripts (optional)
README.md
Rui-An (Ryan) Chang
PhD Candidate in Quantum Information Science
University of Texas at Austin
Publication: PRX Quantum (2025)
Future extensions include:
- Noise-aware symmetry diagnostics
- Operator growth and magic under conservation laws
- Benchmarking protocols for hardware-constrained architectures
- Large-scale simulation with GPU acceleration


