Quandary is an open-source C++ software that simulates and optimizes the evolution of both open and closed quantum systems as governed by Lindblad’s equations, the open-system analog to the Schrödinger equation.
Quantum computers use pulses with a given shape and frequency to realize numerical operations. These pulses, therefore, serve as the interface between a quantum compiler and quantum hardware, and require adequate control to arrive in the desired state.
Quandary is designed to assist in this quantum state preparation. Quandary solves quantum control problems in large, open quantum systems by simulating a quantum device within a classical computing system to determine the effects of a specified pulse. It then iterates on and optimizes this pulse design to achieve the desired dynamics and drive the quantum system into the target state. These steps are necessary for a number of quantum computing applications, including quantum error corrections, logical gate operations, and quantum information processing.
Moreover, Quandary is massively parallelized, and one of the first quantum control codes to accomplish these steps in a high performance setting. These capabilities allow it to move beyond small computing systems and account for the interaction of the quantum system with its environment.
Quandary features a Julia and Python interface to simplify its use for physicists and other users unfamiliar with high performance computing. It’s scalable, flexible, and extendable to any optimization requirements beyond the Lab’s own quantum computer: the Quantum Development and Integration Testbed (QuDIT).
Quantum information science is critical for the Lab and future national security challenges. Quandary will help position the Lab as a leader in quantum sensing and control, and scientific quantum computing.
Quandary was made possible by financial support from DOE Advanced Scientific Computing Research (ASCR) and LLNL Laboratory Directed Research and Development, as well as from the Institutional Scientific Capability Portfolio (ISCP).