QuantumCircuitsMPS.jl

Documentation for QuantumCircuitsMPS.jl, a quantum circuit simulation package for MIPT/CIPT research.

"PyTorch for Quantum Circuits" — a pure Julia library for simulating one-dimensional (1D) quantum circuits with four interchangeable backends: Matrix Product States (MPS, via ITensors.jl, L=100+), an exact dense state vector (L≲25, for cross-validation), a stabilizer tableau (Clifford-only gates, L=100-1000+), and a fermionic Gaussian (free-fermion Majorana-covariance-matrix) backend for Gaussian-preserving circuits, exact and polynomial-time. It's purpose-built for researchers studying Measurement-Induced (MIPT) and Control-Induced (CIPT) Phase Transitions in monitored quantum circuits, where feedback, measurements, and unitary dynamics compete to create distinct entanglement phases.

Physicists write apply!(state, HaarRandom(), Bricklayer(:odd)) and never see ITensor index objects, SVD calls, or tensor contractions — one shared simulation engine (the "One Model, Four Backends" design described in Design Philosophy) manages the gap between physics intent (Gates + Geometry) and the numerical representation actually doing the work underneath.

Installation

Not yet registered in the Julia General registry:

using Pkg
Pkg.add(url="https://github.com/hainingpan/QuantumCircuitsMPS.jl")

Local development: cd /path/to/QuantumCircuitsMPS.jl && julia --project=., then:

using Pkg
Pkg.instantiate()
using QuantumCircuitsMPS

Dependencies: ITensors.jl, ITensorMPS.jl, QuantumClifford.jl (required); Luxor.jl (optional, circuit visualization). Julia 1.12+. For interactive development, Pkg.add("Revise") globally and add using Revise before using QuantumCircuitsMPS.

Quick Example

using QuantumCircuitsMPS

L = 12                   # 12 qubits
p = 0.15                 # Measurement probability (near critical point p_c ≈ 0.16)
n_steps = 50              # Time evolution steps

circuit = Circuit(L=L, bc=:periodic) do c
    apply!(c, HaarRandom(), Bricklayer(:even))
    apply_with_prob!(c; outcomes=[(probability=p, gate=Measure(:Z), geometry=AllSites())])
    record!(c, :entropy)
    apply!(c, HaarRandom(), Bricklayer(:odd))
    apply_with_prob!(c; outcomes=[(probability=p, gate=Measure(:Z), geometry=AllSites())])
    record!(c, :entropy)
end

state = SimulationState(L=L, bc=:periodic, maxdim=64,
    rng=RNGRegistry(gates_spacetime=42, born_measurement=1, gates_realization=2))
initialize!(state, ProductState(binary_int=0))
track!(state, :entropy => EntanglementEntropy(; cut=L÷2))

simulate!(circuit, state; n_steps=n_steps, record_when=:marks)

entropies = state.observables[:entropy]
println("Final entropy: $(entropies[end])")

Haar gates entangle, measurements disentangle: below p_c the system is volume-law, above it, area-law. Full walkthroughs (CIPT, AKLT, feedback, all four backends) live in the Tutorials page and the notebooks in examples/.

Documentation

• Design Philosophy — the layered-abstraction diagram, the unified stochastic rule, and broadcast-vs-set geometry semantics.
• Backends: MPS Backend, State Vector Backend, Clifford Backend, Gaussian Backend.
• Tutorials — full walkthroughs (CIPT, AKLT, feedback, all four backends).
• API Reference — the full list of exported types and functions.

This documentation is published in two versions: stable (the latest tagged release) and dev (the dev branch, may include unreleased changes).

Citation

@software{quantumcircuitsmps,
  author = {Pan, Haining and Pixley, Jedediah H},
  title = {QuantumCircuitsMPS.jl: MPS-based Quantum Circuit Simulation},
  url = {https://github.com/hainingpan/QuantumCircuitsMPS.jl},
  year = {2026}
}

Known Limitations

• 1D circuits only: the entire geometry vocabulary (Bricklayer, AllSites, SingleSite, ...), boundary conditions (:open/:periodic), and all four backends assume a one-dimensional chain of sites. Higher-dimensional (2D+) circuit geometries are a planned future direction — see ROADMAP.md.
• RNG stream name hardcoded: the stochastic engine always draws from :gates_spacetime. Independently-named streams per probabilistic operation are deferred until a concrete research use case requires it.

See ROADMAP.md for planned features and the Changelog for the full release history.