An operation that aggregates or compresses fine-scale information into coarse variables by averaging, moment matching, projection, or spectral truncation, producing a reduced representation suitable for coarse models.

Principle

Condense detailed state information into coarse observables while preserving chosen invariants (mass, momentum, energy or moments) and controlling aliasing and sampling error.

Demonstration

Computing cell-average densities and moment closures for a finite-volume coarse model by integrating the fine-grid field over cell volumes and retaining low-order moments as coarse variables.

Misapplication

Using a non-conservative or inconsistent averaging kernel that breaks global conservation, introduces bias, or permits aliasing of high-frequency energy into coarse modes.

Consequence

Produces compact representations that enable cheaper simulation, parameter estimation, and model coupling while making explicit which fine-scale information is discarded or retained.

Reversal

Opposite is Coarse-to-Fine Reconstruction, which attempts to reintroduce plausible subgrid structure; aggregation is generally many-to-one and thus irreversible without additional priors.

Boundary

Requires a well-defined mapping (averaging operator or projection) and a notion of scale separation; not applicable when fine-scale variability cannot be meaningfully summarized by the chosen moments or projections.

Semantic Tension

Tension between lossless linear projections (orthogonal projection in a Hilbert space) and lossy nonlinear compressions that retain task-relevant features but violate classical linear invariants.

Synthesis

A controlled compression operator that maps high-resolution fields to coarse observables by integrating or projecting while prioritizing conservation and clarity about discarded information.