Dynamic Systems Forecasting
Quantum Reservoir Computing (QRC) is a near-term quantum machine learning paradigm for temporal data processing that requires no gradient-based optimization of the quantum system and is suited to noisy intermediate-scale quantum hardware. Participants will design and benchmark simulation-friendly QRC systems and apply them to real-world industry problems where chaotic dynamics and nonlinear temporal dependencies are central, including financial volatility prediction and climate and weather time-series forecasting. Teams will demonstrate performance across different qubit counts and realistic noise models, benchmark against classical baselines, and implement a common benchmark to validate that the quantum reservoir exhibits sufficient expressivity for forecasting complex, regime-shifting systems.
Phase 1 focuses on three things: building your team, selecting your Challenge track, and developing your initial concept.
Team Composition
The Global Industry Challenge is a collaborative competition; teams can be formed locally or internationally with between 1 and 5 teammates.
Find teammates here.
Join a team here.
Teams may have between 1 and 5 participants. We recommend 3 to 5 members to balance the workload effectively. Strong teams are diverse in skill. We encourage you to fill the following roles:
• Coder
• Data / Technical Lead
• Content / Domain Expert
• Business / Commercial Lead
• Project Manager / Team Coordinator
What do I submit for Phase 1?
Create a 1‑page (plus cover page) maximum PDF, using 11‑point Times New Roman font and single spacing, and submit it through the relevant Aqora Competition.
File Format Requirement: TeamName__Phase1_VersionX.pdf
Include the following details:
- Cover Page (not included in page count) GIC 2026 Cover Page.docx
- Team Qualifications for Tackling the Challenge (Education, Job Title and Organization)
- Brief Description of the Steps You Will Use to Solve this Challenge
- High‑level overview of your proposed solution
- Technical approach, including how you will utilize quantum computing feasibility and advantage over classical methods
- Projected industry impact
First Phase Submission Deadline: Sunday, April 5, 11:59 PM (EST).
Judging Criteria
Teams will be evaluated on two dimensions. First, technical proficiency: judges will assess the depth and quality of your team's quantum computing knowledge and execution capability. Second, a higher weight is placed on conceptual feasibility: your Phase 1 submission must define a credible, technically grounded approach that benchmarks quantum methods against classical baselines and articulates measurable industry impact.
Quantum Reservoir Computing (QRC) is one of the most promising near-term quantum machine learning paradigms for temporal data processing. Unlike variational quantum algorithms, QRC requires no gradient-based optimization of the quantum system. Only a single linear readout layer is trained, making it naturally suited to today’s noisy intermediate-scale quantum hardware. Recent results have demonstrated QRC scaling to 108 qubits on neutral atom hardware with promising results against the classical baselines on chaotic time-series forecasting.
This challenge asks participants to design, build, and benchmark simulation-friendly QRC systems (5–20 qubits) and apply them to real-world industry problems where chaotic dynamics, nonlinear temporal dependencies, and regime shifts make QRC’s unique strengths directly relevant. Participants choose one or both of the following application tracks:
- Financial Volatility Prediction: Markets swing between calm and chaotic periods. Predicting when these shifts happen is a core challenge in quantitative finance, with direct impact on how portfolios are managed and how risk is priced. Build a QRC that detects volatility regime changes in publicly available equity data and forecasts transitions before they happen.
- Weather Time-Series Forecasting: Atmospheric systems are inherently chaotic, making accurate forecasting one of the classic hard problems in science. Using real-world weather station data from sources like NOAA, build a QRC that predicts temperature, pressure, or humidity sequences. Short-term weather prediction directly impacts agriculture, energy, logistics, and disaster response, making it an ideal testbed for QRC on a problem that matters.
In addition to the chosen industry track, all participants implement a common MNIST digit classification benchmark using their QRC architecture, providing a standardized comparison across teams and validating that the quantum reservoir exhibits sufficient expressivity. Participants are expected to demonstrate how their QRC performs across different qubit counts (e.g., 5, 10, 15 qubits) and under realistic noise models, including depolarizing channels and amplitude damping.
Platform & Compute Access
Participants will receive access to the qBraid platform, which provides a unified environment with quantum simulators, GPU-accelerated training, and quantum processing units (QPUs) from multiple hardware providers. Simulator-based development is the primary mode of work for this challenge; GPU and QPU resources are available to support training and hardware validation runs as participants progress through the challenge.
Alongside their GitHub repository, teams are asked to submit a qBraid Skill: a structured, agent-executable package that allows an AI coding agent to navigate the codebase, configure the reservoir, run training, and reproduce results end-to-end. This is not just documentation; it is a functional interface for reproducibility. Teams are encouraged to leverage qBraid’s agentic coding capabilities throughout development, including agent-driven hyperparameter sweeps, automated benchmarking across qubit counts, and AI-assisted noise analysis.
Design Space
QRC offers a rich architectural design space. Participants are free to explore any approach, including but not limited to:
-Hamiltonian design: Ising, Heisenberg, random graphs, Bose-Hubbard, or custom topologies
-Input encoding schemes: amplitude encoding, angle encoding, temporal multiplexing
-Measurement strategies: projective, weak, or continuous measurement; spatial multiplexing across parallel reservoirs
-Feedback mechanisms: re-injecting measurement outcomes to restore fading memory and enhance nonlinearity
-Hybrid architectures: combining classical memory-augmentation layers with quantum feature extraction
-Readout strategies: polynomial regression, ridge regression, or other post-processing to expand feature space
Key References
- Kornjaca et al., “Large-scale quantum reservoir learning with an analog quantum computer,” arXiv:2407.02553 (2024).
- Zhu et al., “Practical few-atom quantum reservoir computing,” Phys. Rev. Research 7, 023290 (2025).
- Tandon et al., "Quantum Reservoir Computing for Corrosion Prediction in Aerospace: A Hybrid Approach for Enhanced Material Degradation Forecasting," arXiv:2505.22837 (2025)
- Li et al., “Quantum Reservoir Computing for Realized Volatility Forecasting,” arXiv:2505.13933 (2025).
- Ahmed et al., “Robust quantum reservoir computers for forecasting chaotic dynamics: generalized synchronization and stability,” Proc. R. Soc. A 481 (2025).
