method-qcs

name	method-qcs
description	Use when a quantum circuit simulation, VQE, TensorCircuit-NG, JAX backend, contraction path, statevector, MPS circuit, gradient, or circuit-performance reproduction needs method-level route and tool selection.

Method QCS

Quantum circuit simulation is the differentiable-circuit and performance track. Use it to choose representation and software stack before tuning backend settings.

Sources

• Methodology reference (reproduction-grade algorithm, parameters, validation, gap analysis): references/qcs-methodology.md
• Track README: tracks/qcs/README.md
• Interview notes: docs/qcs/interview.html
• Review notes: docs/qcs/review.html
• Challenge brief: docs/qcs/backup.md
• Tool skills: /using-tensorcircuit-ng, /using-jax

Route

1. Choose representation first: tensor-network contraction, full statevector, or MPS circuit. This determines memory and whether the result is exact or approximate.
2. Recommend /using-tensorcircuit-ng for large/deep differentiable VQE, energy+gradient profiling, contraction tuning, batching, scan/checkpointing, and the QCS track benchmark.
3. Invoke /using-jax for backend, CPU/GPU, precision, device, and compile/warm-runtime setup.
4. If the paper requires Qiskit, PennyLane, Yao, cuQuantum, or quimb specifically, offer official code / web search or a new tool skill instead of silently forcing TensorCircuit-NG.
5. Keep compile time, path-search time, warm runtime, and optimizer-loop time separate in any proposal or report.

Tool Handoff

Invoke /using-tensorcircuit-ng after the route is chosen. /using-tensorcircuit-ng owns representation-specific parameters, contraction/path settings, observable format, gradients, validation, and time estimate. Invoke /using-jax for backend setup.

Details

Classical simulation of quantum circuits, including variational circuit optimization (VQE) and general circuit contraction. In this harness, the default route uses TensorCircuit-NG on a JAX backend with automatic differentiation, JIT compilation, and tensor-network contraction controls. It is not a hardware-execution workflow.

This card is generic methodology. Paper-specific benchmarks, hardware layouts, and figure protocols belong in /reproduce-paper protocols or run specs, not here.

Scope

Use this card for:

• Parameterized circuit ansatz construction.
• Hamiltonian expectation values from dense, sparse, Pauli-sum, or MPO-like representations.
• Reverse-mode gradients through circuit contractions.
• JIT-compiled value-and-gradient kernels.
• Runtime, compile-time, and peak-memory profiling of differentiable VQE steps.
• Contraction-path, slicing, batching, scan, checkpointing, and precision studies.

Do not use this card for:

• Hardware execution, cloud submission, readout mitigation, or QPU calibration.
• QAOA or generic quantum algorithms unless they are treated as VQE-like variational energy minimization.
• TensorCircuit-NG paper-specific benchmark reproduction.
• General quantum-circuit simulation without a variational energy / differentiable objective.

Notation

• theta: variational parameters.
• Ansatz: parameterized circuit preparing |psi(theta)>.
• H: Hamiltonian represented as Pauli terms, sparse matrix, dense matrix, or tensor-network operator.
• Forward pass: compute E(theta).
• Backward pass: compute grad_theta E(theta) by reverse-mode AD.
• JIT compile time: first traced/compiled execution cost.
• Steady runtime: execution time after compilation/cache warmup.
• Peak memory: maximum device or host memory during forward/backward execution.
• Contractor: tensor-network contraction path and execution strategy.

Pitfalls

• Dense Hamiltonian blowup: prefer sparse, Pauli-sum, or MPO-like representations when a dense matrix would dominate memory. The representation choice is part of the method, not only a runtime tweak.

Verification

• Small-system ED check: compare VQE energy to ED for the same Hamiltonian and boundary condition.
• Gradient check: finite-difference a small parameter subset and compare with AD gradients.
• Energy bound: VQE energy should not fall below the exact ground-state energy for a Hermitian Hamiltonian.
• Seed/depth stability: run multiple seeds and a depth sweep before claiming ansatz convergence.

Citations

• .knowledge/literature/quantum-circuit-simulation/tensorcircuit-tensorcircuit-ng.md - official TensorCircuit-NG repository and documentation entry.
• .knowledge/literature/quantum-circuit-simulation/2602.14167_tensorcircuit-ng-a-universal-composable-and-scalable-platfor.md - TensorCircuit-NG software reference.
• .knowledge/literature/quantum-circuit-simulation/2205.10091_tensorcircuit-a-quantum-software-framework-for-the-nisq-era.md - TensorCircuit software and differentiable-circuit reference.
• .knowledge/literature/quantum-circuit-simulation/2002.01935_hyper-optimized-tensor-network-contraction.md - cotengra contraction-path reference.