A coupling paradigm that connects models operating at different spatial and/or temporal scales, enabling information transfer so that phenomena whose effects span scales are represented in the combined prediction.

Principle

Identify scale separation or overlap, define mappings (upscaling, downscaling, homogenization, embedding) between scales that preserve relevant conserved quantities and statistical properties, and choose sequential or concurrent coupling strategies according to the physics and computational constraints.

Demonstration

Fracture modeling that couples molecular dynamics near a crack tip to a continuum finite‑element model away from the tip: atomistic simulations provide stress/constitutive information to the continuum (upscaling) while boundary displacements from the continuum constrain the atomistic region (downscaling) to capture crack initiation and propagation.

Misapplication

Applying homogenized effective properties from fine scale to coarse models when important nonlocal or transient fine‑scale dynamics exist will miss critical behaviors; likewise, naive embedding without proper scale interface conditions can introduce artifacts.

Consequence

Allows capturing emergent macroscale effects driven by microscale physics and reduces the need for prohibitively fine global resolution, at the cost of added complexity in scale mappings, validation and possible numerical stiffness at scale interfaces.

Reversal

Single‑scale modeling that ignores influence across scales, which can be valid when clear scale decoupling exists but will fail when microscale phenomena significantly affect macroscale response.

Boundary

Applies to coupling across spatial and/or temporal scales; excludes purely multiphysics couplings that occur without scale separation and excludes trivial parameter passing lacking principled mapping or conservation.

Semantic Tension

Tension exists between multiscale coupling and scale bridging: multiscale coupling emphasizes simultaneous or nested solution across scales, while nearby meanings emphasize one‑way parameter transfer or heuristic upscaling; both overlap in practice.

Synthesis

Multiscale coupling links models at different scales through principled upscaling/downscaling or concurrent embedding so that scale‑spanning phenomena are represented without resolving all scales everywhere.