Reproduce “Evaluating Predictor-Corrector Dependence in Hard-Constrained Climate Emulators” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Tianyu Zhang, Xinyu Wang, Zhenghan Tai, Xiao-Wen Chang, Yoshua Bengio

The authors present an evaluation framework that separates corrected forecasts, raw outputs on corrected trajectories, and fully uncorrected rollouts. They cross correction during adaptation and forecasting to separate inference effects from predictor differences. In a three-seed ACE2 pilot, uncorrected precipitation error was worse at short horizons but better at 10 and 30 days; the pattern’s cause and generality remain unknown.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.