Techniques for obtaining approximate solutions to mathematical problems that cannot be solved exactly, using finite sequences of arithmetic operations; encompassing root-finding, interpolation, quadrature, linear-system solvers, ODE/PDE integrators, and optimisation routines with explicit control over accuracy, stability, and computational cost.
Semantic Classification
Content
- Numerical methods provide algorithms for tasks such as solving linear systems, finding roots, integrating functions, and approximating solutions to differential equations. Key concerns include accuracy, stability, and computational cost.
- These methods underpin scientific computing and simulation, and many optimisation procedures used in machine learning are numerical methods at their core.