Diagrammatic Differentiation for Quantum Machine Learning
This work addresses the challenge of automatic differentiation for hybrid classical-quantum circuits in quantum machine learning, offering a novel diagrammatic approach that could enhance computational efficiency and practical implementations.
The paper tackles the problem of differentiating parametrized quantum circuits by introducing diagrammatic differentiation, a method that generalizes dual numbers to monoidal categories and applies it to ZX diagrams, enabling gradient calculations for quantum machine learning. The result includes an open-source implementation in DisCoPy, allowing simplified gradients via PyZX and execution on quantum hardware through tket, with applications in variational quantum algorithms.
We introduce diagrammatic differentiation for tensor calculus by generalising the dual number construction from rigs to monoidal categories. Applying this to ZX diagrams, we show how to calculate diagrammatically the gradient of a linear map with respect to a phase parameter. For diagrams of parametrised quantum circuits, we get the well-known parameter-shift rule at the basis of many variational quantum algorithms. We then extend our method to the automatic differentation of hybrid classical-quantum circuits, using diagrams with bubbles to encode arbitrary non-linear operators. Moreover, diagrammatic differentiation comes with an open-source implementation in DisCoPy, the Python library for monoidal categories. Diagrammatic gradients of classical-quantum circuits can then be simplified using the PyZX library and executed on quantum hardware via the tket compiler. This opens the door to many practical applications harnessing both the structure of string diagrams and the computational power of quantum machine learning.