Second International Conference on Probabilistic Numerics

The Second International Conference on Probabilistic Numerics (PN 2026) addresses a fundamental reconceptualization of numerical computation: rather than treating algorithms for integration, optimization, or differential equations as producing exact answers, probabilistic numerics frames computation as a statistical inference problem, returning calibrated uncertainty estimates alongside solutions. This perspective has important implications for scientific computing and machine learning.

The conference brings together numerical analysts, statisticians, machine learning researchers, and applied mathematicians who develop or use probabilistic numerical methods. Topics include Bayesian quadrature, probabilistic linear solvers, Gaussian process-based ODE solvers, and uncertainty quantification in large-scale simulations. PN 2026 is particularly relevant for researchers in scientific machine learning and safety-critical applications where knowing how much to trust a numerical answer is as important as the answer itself.