XaosQ

XaosQ is designed as a modular platform with three solver families: XaosQ-or, XaosQ-bifurcate, and XaosQ-quantum. This allows the system to select or combine different computational approaches depending on problem type, scale, and runtime constraints.

XaosQ is not only a rescheduling engine but a broader robustness framework built around a stochastic perturbation model, instability-cost metrics, and stability-aware scheduling. This means the product is designed to optimize both solution cost and schedule resilience under disruptions.

A core feature of XaosQ is dynamic resource allocation with explicit cost functions and constraint-handling logic. The design also includes configurable convergence criteria and mechanisms for escaping local minima, making it suitable for complex non-convex optimization workflows.

The XaosQ-bifurcate module is based on physics-inspired optimization using chaos/simulated-bifurcation-style methods. It is intended to provide high performance on large optimization instances using classical hardware, especially where standard heuristics scale poorly.

The XaosQ-quantum module is designed as a hybrid solver that integrates classical optimization with quantum/annealing backends. It is positioned as a “quantum-ready” component that can exploit available NISQ annealers while still operating within a broader classical platform.

XaosQ is designed for execution and validation on HPC infrastructure and cloud CPU/GPU environments. This supports large-scale testing, performance tuning, and practical deployment of computationally intensive optimization pipelines.

We use ML solver tuning to get optimization quality improvements. In practice, this adds an adaptive layer for hyperparameter selection and potentially better solver behavior across changing problem distributions.

XaosQ uses a Time-Space Network (TSN) model combined with mixed-integer linear programming (MILP) to optimize aircraft and crew recovery jointly. This allows the solver to handle interdependencies that are usually missed when aircraft and crew are solved in separate sequential steps.

XaosQ uses disruption-aware search-space construction to avoid enumerating an excessively large set of recovery options. It focuses solver effort on relevant alternatives around disrupted parts of the schedule, which helps maintain tractable runtimes.

XaosQ-OR is designed for airline operations control centers and targets real decision windows during day-of-operations disruptions. Its architecture prioritizes fast generation of feasible, cost-efficient recovery plans under time pressure.