Quantum Information Science · PHY 568
Quantum Gradients in Variational Quantum Algorithms
From the two-point parameter-shift rule to Fourier-based gradient reconstruction, with applications to QAOA and quantum natural gradients.
Overview
Coursework on computing and using gradients in variational quantum algorithms. It connects the frequency structure of parameterized quantum expectation values to generalized parameter-shift rules.
Topics & methods
- Simple and generalized parameter-shift rules, including equidistant and arbitrary spectra.
- Fourier reconstruction of expectation values and their derivatives.
- Applications to QAOA and MaxCut, followed by the geometry of quantum natural gradients.
Discussion
Generalized parameter-shift rules use the frequency structure of quantum expectation values to compute gradients for variational algorithms.