A mesh-scale oscillatory pattern (often alternating high/low values on adjacent nodes or elements) in computed fields—commonly pressure or stress—caused by incompatible discretizations, lack of stabilization, or poor choice of interpolation spaces.

Principle

When the discrete approximation spaces for coupled variables (e.g., velocity and pressure) do not satisfy stability conditions (such as the inf-sup/Babuška–Brezzi condition) or lack filtering/stabilization, spurious high-frequency modes at mesh scale can appear as checkerboard patterns.

Demonstration

In incompressible flow, using equal-order interpolation for velocity and pressure without stabilization can produce a pressure field that alternates sign node-to-node across the mesh, a classic checkerboarding symptom.

Misapplication

Treating checkerboard patterns as meaningful local oscillations and refining the mesh without addressing the stability mismatch entrenches the artefact and increases computational cost without improving physical accuracy.

Consequence

Unchecked checkerboarding pollutes field estimates, corrupts derived quantities (forces, fluxes), and undermines solver convergence or error estimation, motivating stabilization, mixed formulations, or appropriate interpolation pairings.

Reversal

Adopting stable mixed elements, applying pressure/gradient stabilization, or using appropriate filtering removes checkerboarding and yields physically consistent, mesh-insensitive fields.

Boundary

Refers to spurious mesh-scale oscillations in discrete fields due to discretization instability; distinct from legitimate physical high-frequency content, round-off noise, or oscillations from poorly converged iterative solvers.

Semantic Tension

Tension between simple interpolation and stability: equal-order or cheap interpolations are attractive for implementation, but they risk checkerboarding versus mixed or stabilized schemes that impose more constraints and complexity.

Synthesis

Checkerboarding is the emergence of mesh-scale spurious oscillations from incompatible or unstabilized discretizations; resolving it requires ensuring stability of approximation spaces or adding targeted stabilization so computed fields reflect physical structure rather than numerical artefacts.