TFQ

TensorFlow Quantum (TFQ) is an open-source library developed to enable researchers to build and train quantum machine learning models and hybrid quantum-classical pipelines within the TensorFlow ecosystem. It provides high-level abstractions for creating parameterized quantum circuits using Cirq and for integrating them into TensorFlow models.

TFQ converts Cirq circuits into TensorFlow tensors, allowing end-to-end differentiable workflows. This enables gradient-based optimization of quantum circuit parameters with standard TensorFlow optimizers. The library supports parameterized quantum circuits, expectation-value measurements, and sampling, with components that can be inserted into Keras models to form hybrid architectures.

A key aim of TFQ is to facilitate hybrid quantum-classical learning, where data is encoded into quantum states and processed by quantum circuits, while classical components provide optimization and post-processing. Common use cases include quantum-enhanced feature maps, variational quantum circuits for classification or regression, and exploratory studies of quantum data encoding techniques. TFQ emphasizes interoperability with existing quantum simulation tools and backends provided by Cirq, and can target both simulated environments and, when available, real quantum hardware through Cirq-compatible interfaces.

History and status: TFQ was released to accelerate research in quantum machine learning by Google and the broader community, integrating tightly with Cirq for circuit construction and TensorFlow for training and deployment. The project is open-source and maintained by Google's quantum AI team along with contributors from the research community.

See also: Cirq, TensorFlow, quantum machine learning, variational quantum algorithms.