TensorCircuit-NG Unites Quantum Computing and AI in One Powerful Framework
A new software platform called TensorCircuit-NG has launched, combining quantum computing, simulation, artificial intelligence, and high-performance computing. The system merges circuits, tensor networks, and neural networks into a single framework designed for advanced optimization and large-scale simulations.
TensorCircuit-NG treats quantum computations as differentiable graphs. This approach allows gradient-based tuning of quantum circuits and parameters, improving efficiency in design and training. The platform's dual-layer architecture unifies infrastructure and representation under a tensor-native philosophy.
The software tackles key limitations of existing simulators by introducing scalable noise modelling and distributed computing. It delivers near-linear speed improvements in variational quantum algorithms, enabling complex tasks like end-to-end quantum machine learning for applications such as image recognition. Additionally, it supports differentiable optimization of tensor network states, which is vital for solving problems in many-body physics.
TensorCircuit-NG also broadens compatibility with various quantum systems, including qudit setups and fermion Gaussian states. Its advanced simulation engines and noise-handling strategies aim to push the boundaries of quantum system modelling.
The platform promises to accelerate research in quantum simulation and optimization by offering greater flexibility and scalability. Its ability to integrate multiple computational approaches could open new possibilities for both theoretical and applied quantum science. Major institutions worldwide continue to explore similar frameworks, though specific adoption figures remain unverified as of early 2026.
