How-To Guides#
Problem-oriented guides for specific tasks. Each guide addresses a particular challenge and provides practical solutions.
- How to Optimize Performance
- Problem
- Prerequisites
- Enable GPU Acceleration
- Optimize Bond Dimensions
- Use Mixed Precision
- Batch and Parallelize Operations
- Profile and Identify Bottlenecks
- Optimize Memory Usage
- Algorithm-Specific Optimizations
- Use Stabilizer Backend for Clifford
- Hardware Considerations
- Verification
- Summary
- See Also
- How to Handle Large Quantum Systems
- Configure Precision
- Integrate cuQuantum
- Debug Simulations
- Save and Load State
- How to Build Custom Hamiltonians
- Parallel Computation
- Benchmark Comparison
Guide Overview#
- How to Optimize Performance
Maximize simulation speed using custom Triton kernels, GPU optimization, and efficient tensor operations.
- How to Handle Large Quantum Systems
Simulate systems beyond single-GPU memory limits using adaptive truncation, distributed MPS, and memory budgets.
- Configure Precision
Choose appropriate numerical precision (complex32/64/128) and configure mixed-precision policies.
- Integrate cuQuantum
Enable NVIDIA cuQuantum acceleration for 2-10× speedup on supported operations.
- Debug Simulations
Diagnose numerical issues, track error propagation, and validate simulation correctness.
- Save and Load State
Checkpoint MPS states, save optimization progress, and resume long-running simulations.
- How to Build Custom Hamiltonians
Build custom Hamiltonians using MPO operations, including non-local interactions and time-dependent terms.
- Parallel Computation
Leverage multi-GPU parallelism with distributed MPS and data-parallel measurement sampling.
- Benchmark Comparison
Compare ATLAS-Q performance against Qiskit, Cirq, and ITensor for specific use cases.
Prerequisites#
Guides assume familiarity with basic ATLAS-Q usage. If you are new to ATLAS-Q, complete the Tutorials first.