ATLAS-Q Documentation#
ATLAS-Q (Adaptive Tensor Learning And Simulation – Quantum) is a GPU-accelerated quantum tensor network simulator implementing Matrix Product States (MPS), Matrix Product Operators (MPO), Projected Entangled Pair States (PEPS), and variational quantum algorithms. The framework provides memory-efficient quantum state representation with adaptive bond dimensions, custom GPU kernels, and specialized backends for Clifford circuits and period-finding.
Version 0.7.0 (December 2025) - IR v1.1.0
NEW: IR Integration - Pre-Computation Diagnosis
ATLAS-Q v0.7.0 introduces complete Informational Relativity (IR) integration with pre-computation regime diagnosis. This breakthrough enables you to know whether quantum algorithms will succeed before running them.
IR v1.1.0 Features:
Regime Analyzer: Diagnose problems as IR (observable), Transition, or AIR (hidden) regimes
Pre-Computation GO/NO-GO: Physics-derived e^-2 threshold (0.135) for trustworthiness
4× VQE Circuit Reduction: Coherence-based measurement grouping
42% Period Finding Shot Reduction: IR preprocessing for QPE
100 Qubit MPS Simulation: 1.56 seconds with 10^25× memory compression
Spectral Lifting: Full M_ij relational matrix analysis for structure detection
Key capabilities:
Coherence-Aware VQE/QAOA: Real-time coherence tracking with GO/NO-GO classification
IR Integration: Circular statistics and RMT-based quality metrics (R̄, V_φ)
Hardware-Validated: Tested on IBM Brisbane with near-ideal coherence (R̄=0.988 for H2O)
Adaptive MPS with per-bond dimension control and global memory budgets
Variational algorithms: VQE, QAOA with hardware-efficient and UCCSD ansätze
Grover’s quantum search algorithm with MPO-based oracles (94-100% accuracy)
Time evolution via TDVP (1-site and 2-site)
PEPS for 2D tensor networks
Stabilizer backend for Clifford circuits (20× speedup)
Circuit cutting and entanglement forging
Distributed MPS for multi-GPU simulation
Molecular Hamiltonians via PySCF integration
Custom Triton kernels for gate operations (1.5-3× speedup)
cuQuantum backend integration
Performance: 100 qubits in 1.56 seconds, 10^25× memory compression at 100 qubits, 4× circuit reduction with IR grouping.
Getting Started
User Guide
API Reference
- API Reference
- IR Module (Informational Relativity)
- atlas_q.adaptive_mps
- MPS PyTorch Backend
- atlas_q.mpo_ops
- atlas_q.tdvp
- atlas_q.vqe_qaoa
- atlas_q.grover
- atlas_q.stabilizer_backend
- Noise Models
- PEPS (2D Tensor Networks)
- Circuit Cutting
- 2D/Planar Circuits
- Distributed MPS
- atlas_q.cuquantum_backend
- atlas_q.quantum_hybrid_system
- Triton GPU Kernels
- atlas_q.diagnostics
- atlas_q.linalg_robust
- atlas_q.truncation
- Module Overview
- Quick Access
- Module Import Patterns
- Type Annotations
- Index
Developer Documentation
Additional Resources
Quick Links#
Citation#
@software{atlasq2025,
title={ATLAS-Q: Adaptive Tensor Learning And Simulation – Quantum},
author={ATLAS-Q Development Team},
year={2025},
url={https://github.com/followthesapper/ATLAS-Q},
version={0.7.0}
}