Cross-posted at https://pennylane.ai/blog/2022/07/lightning-fast-simulations-with-pennylane-and-the-nvidia-cuquantum-sdk/.
Cross-posted at https://aws.amazon.com/blogs/quantum-computing/accelerate-your-simulations-of-hybrid-quantum-algorithms-on-amazon-braket-with-nvidia-cuquantum-and-pennylane/.
Cross referenced from http://gpue-group.github.io/.
One of the major components of GPUE is the ability to track and manipulate vortices in Bose-Einstein condensates. While we can work with creating vortices in 3D, the majority of the vortex creation and manipulation framework exists solely in 2D. This is due to the 2D vortex code being developed for the project on vortex dynamics, published in PRA here (arXiV version here).
To allow us to track and manipulate vortices in (2D) condensates, we require some numerical techniques and tools:
Cross-posted from https://gpue-group.github.io/development/.
CuFFT usage with angular momentum/gauge fields To implement angular momentum operators in split-operator based pseudo-spectral methods, we must take special care of these evolution operators. As outlined in O’Riordan, 2017, the non-commutative nature of the position and momentum space operators required for angular momentum present a challenge — we cannot implement a numerical model without trade-offs. This is well documented in the above literature, though what is not discussed is an implementation of this method using CUDA and the CuFFT library.
Investigating the use of different FFT implementations for quantum dynamics across different hardware architectures
GPU enabled Gross-Pitaevskii equation solver