Probabilistic Numerics
A methodological framework that treats numerical computation as a statistical inference problem, in which algorithms return posterior distributions rather than single numerical values, thereby quantifying uncertainty arising from discretization, finite observations, or model error.
Timeline (last 90 days)
Gaussian process solvers belong to the Probabilistic Numerics framework, which treats numerical computation as a statistical inference problem and returns a posterior distribution.
This method transforms the acceptability of solutions from a traditional post-processing step into conditioning information for a probabilistic PDE solver, allowing the posterior distribution itself to respect the constraints.
All Facts (2)
Gaussian process solvers belong to the Probabilistic Numerics framework, which treats numerical computation as a statistical inference problem and returns a posterior distribution.
50%UnverifiedThis method transforms the acceptability of solutions from a traditional post-processing step into conditioning information for a probabilistic PDE solver, allowing the posterior distribution itself to respect the constraints.
50%