A family of transformations on a space of physical or statistical models that successively coarse-grain short-scale degrees of freedom and rescale variables, producing a trajectory of effective parameters as the observation scale changes.

Principle

Integrate out or average fine-scale fluctuations and perform rescaling (change of units and field normalization) so that the effective description at a larger length or lower energy scale is expressed by new parameter values; repeating yields a flow in theory-parameter space.

Demonstration

Block-spin transformation in the two-dimensional Ising lattice: group nearby spins into blocks, define block variables, sum over internal configurations and rescale to obtain new effective coupling constants; iterating produces trajectories toward fixed effective descriptions.

Misapplication

Applying coarse-graining without the compensating rescaling or using a single finite coarse operation and treating the result as universal flow information; ignoring relevant operator directions that grow under scale change.

Consequence

Proper RG flow classifies large-scale behavior by attracting and repelling regions (fixed descriptions) and explains universality: disparate microscopic models can flow to the same large-scale effective description and scaling laws.

Reversal

A microscopic description that keeps all degrees of freedom at the original scale rather than eliminating short-scale structure; such a description preserves microscopic details rather than exhibiting emergent universal behavior.

Boundary

Applies when a notion of scale and scale-separable fluctuations exists; not directly applicable to finite small systems without scale separation, to strictly nonlocal interactions lacking a scale hierarchy, or when coarse-graining changes the model class incompatibly.

Semantic Tension

Contrasted with mean-field or perturbative approximations: RG flow stresses scale-dependent changes of effective parameters and operator relevance, whereas mean-field ignores fluctuation-driven renormalization and may miss universality classes.

Synthesis

Renormalization group flow is the iterative process of eliminating short-scale fluctuations and rescaling that maps a model to scale-dependent effective parameters, producing trajectories that organize universal large-scale behavior.