skills.rest
unitarylab
unitarylabOFFICIAL
Explore all skills by this author →
GitHub

reverse

Computes expectation gradients and QGT entries via reverse-mode statevector differentiation in Qiskit.

18|3|Updated Aug 14, 2026
Search Tags:#gradient#quantum computing#qiskit#variational algorithms#reverse-mode#statevector#qgt
One-click install
npx skills add https://github.com/unitarylab/quantum-practices --skill reverse-unitarylab
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: reverse
Source: https://github.com/unitarylab/quantum-practices/tree/main/algorithms/gradients/reverse
Command: npx skills add https://github.com/unitarylab/quantum-practices --skill reverse-unitarylab

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires qiskit, qiskit_algorithms, numpy, unitarylab, and includes scripts (resource) components.

What problem does it solve? Computing exact gradients and Quantum Geometric Tensor entries for parameterized quantum circuits typically requires many extra circuit evaluations with parameter-shift or finite-difference methods. This Skill provides reverse-mode statevector differentiation using Qiskit's ReverseEstimatorGradient and ReverseQGT, delivering exact analytic gradients in a single reverse sweep for small circuits. ## Core Features & Use Cases - Exact Gradient Computation: Uses ReverseEstimatorGradient to compute expectation value gradients with O(P) scaling in the number of parameterized gates, with no truncation error. - QGT Computation: Uses ReverseQGT with configurable phase-fix and derivative type (REAL, IMAG, COMPLEX) for natural gradient optimization workflows. - Reference Implementation: Includes a manual NumPy-based reverse-sweep implementation with known test cases validated against finite differences and analytic solutions. - Use Case: A researcher training a variational quantum eigensolver on a 2-qubit RY+CX ansatz needs exact gradients of the ZZ expectation value; this Skill generates runnable Qiskit code and validates results against finite-difference baselines. ## Quick Start Ask the assistant to compute the gradient of the ZZ observable for a two-qubit RY parameterized circuit using reverse-mode differentiation in Qiskit.

Frequently Asked Questions about reverse

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I compute gradients of a parameterized quantum circuit in Qiskit?▼

Use ReverseEstimatorGradient from qiskit_algorithms.gradients by calling run() with your circuit, observable, and parameter values. It performs a reverse sweep over statevectors to produce exact analytic gradients without extra circuit evaluations.

What is the difference between reverse-mode gradient and parameter-shift?▼

Reverse-mode gradient differentiates directly on statevectors in a single backward sweep with O(P) scaling, while parameter-shift requires additional circuit evaluations per parameter. Reverse-mode is exact but limited to statevector simulation of small circuits.

How do I compute the Quantum Geometric Tensor in Qiskit?▼

Use ReverseQGT from qiskit_algorithms.gradients and call run() with your circuit and parameter values. Set phase_fix=True for the phase-fixed metric used by natural gradient optimizers, and choose the derivative_type for real, imaginary, or complex output.

Why does ReverseEstimatorGradient raise KeyError with parameterized two-qubit gates?▼

In qiskit_algorithms 0.4.0, circuits with parameterized cp, crx, cry, or crz gates can raise KeyError during gradient-parameter preprocessing. Use only parameterized rx, ry, rz gates and keep entanglers like cx non-parameterized.

What are the limitations of reverse-mode statevector gradients?▼

Runtime and memory scale exponentially as O(2^n) with qubit count, making circuits beyond roughly 10 qubits impractical. It also supports only the declared gate set and cannot use external estimator or sampler primitives.

Related Skills

Explore More Education & Research Skills→
Estrella-231Estrella-231

qiskit

Execute quantum circuits and extract measurement or expectation results using Qiskit.

Community
Intermediate
K-Dense-AIK-Dense-AI
33.0k

pennylane

Optimize quantum circuits with automatic differentiation across simulators and hardware.

Official
Advanced
lanceguilancegui
2

structural-estimation

Estimates structural economic model primitives and simulates counterfactual equilibria in R, Julia, or Python.

Community
Advanced
Estrella-231Estrella-231

pennylane

Construct and train differentiable quantum circuits with automatic optimization.

Community
Advanced