Introduction to Geometric Quantum Machine Learning

Introduction to Geometric Quantum Machine Learning

This PennyLane tutorial introduces Geometric Quantum Machine Learning (GQML), an approach that builds the known symmetries of a problem directly into a quantum model's architecture to make it learn more efficiently and generalize better. The central idea, borrowed from geometric deep learning, is that if a task is invariant or equivariant under some group of transformations (for example, the labels of a graph do not change when its nodes are relabeled), then the model should be too; baking this symmetry into the circuit reduces the space of functions it must search and prevents it from wasting capacity learning the symmetry from data. The tutorial explains the relevant group-theory concepts (symmetry groups, invariance, and equivariance) at a practical level, and shows how to construct symmetry-respecting (equivariant) quantum circuits in PennyLane by choosing gates and encodings that commute appropriately with the symmetry. It works through a concrete example demonstrating the advantage of an equivariant model over a generic one. By connecting symmetry principles to circuit design, the tutorial gives a principled, modern foundation for building better-structured quantum machine-learning models in PennyLane.

Run it

pip install -r requirements.txt
python demo.py

Source and license

Imported from demonstrations_v2/tutorial_geometric_qml/demo.py in PennyLaneAI/demos at c52c0abeb5122218aa96b38eea848864cce7323f, under the Apache License 2.0. Original authors: Xanadu and the PennyLane community. The upstream LICENSE is included alongside this example.