Quantum control is concerned with the realisation of desired dynamics in quantum systems, serving as a linchpin for advancing quantum technologies and fundamental research. Analytic approaches and standard optimisation algorithms do not yield satisfactory solutions for more complex quantum systems, and especially not for real world quantum systems which are open and noisy. We devise a physics-constrained Reinforcement Learning (RL) algorithm that restricts the space of possible solutions. We incorporate priors about the desired time scales of the quantum state dynamics - as well as realistic control signal limitations - as constraints to the RL algorithm. These constraints improve solution quality and enhance computational scaleability. We evaluate our method on three broadly relevant quantum systems and incorporate real-world complications, arising from dissipation and control signal perturbations. We achieve both higher fidelities - which exceed 0.999 across all systems - and better robustness to time-dependent perturbations and experimental imperfections than previous methods. Lastly, we demonstrate that incorporating multi-step feedback can yield solutions robust even to strong perturbations. Our implementation can be found at: https://github.com/jan-o-e/RL4qcWpc.

Network design is a storied discipline with a decades-long history of impact and relevance, ranging from telephony networks in the 1950s to modern networks optimized for distributed ML training in the 2020s. Despite this rich history, there are surprising shortcomings in the available analysis and design tools. To name a few: (1) designers often need to reimplement known techniques to evaluate new networks (2) even basic network performance metrics (e.g., sparsest cut or bisection bandwidth) are not easy to visually analyze, resulting in inadvertent errors in published literature, (3) modern techniques for network topology synthesis under various constraints are not easily accessible, and (4) comprehensive comparisons across all known previous network designs is cumbersome. In this work, we introduce an online tool - Library of Networks- that addresses these shortcomings by combining two key functionalities in a unified tool for network design and analysis. First, Library of Networks implements a database of known network topologies that serves as a design pool from which network designers can browse/select networks that satisfy their design requirements. In addition, it enables researchers to comprehensively compare with prior designs without the implementation burden. Second, Library of Networks implements a broad (and expandable) collection of analysis and synthesis routines/kernels that can be used to analyze/generate networks. The analysis kernels in the initial release of the tool include basic performance metrics, deadlock-free routing, highthroughput routing, VC assignment, cycle-level simulation, collective communication scheduling, and area/power estimation. Each of these analysis implementations are easily reusable for new network designs because of the integrated tool design. Library of Networks also supports network synthesis via optimization for objective-seeking design-space exploration. We believe that the tool will be of wide interest to academic researchers, students, and designers in industry.