A transformation that replaces a polynomial map or multilinear-unaware function by an explicitly multilinear map that encodes its homogeneous multilinear components (the polarized or polarized-like form). It is used to isolate the k-linear parts of degree-k homogeneous pieces and to represent nonlinear operations as multilinear tensors or symmetric forms.

Principle

Separate homogeneous degrees and express each homogeneous polynomial of degree k as the diagonal restriction of a unique (when denominators permit) symmetric k-multilinear map; equivalently, obtain multilinear structure by polarization and symmetrization so multilinear algebra techniques apply.

Demonstration

Given a homogeneous cubic polynomial f on a vector space V over a field of characteristic zero, produce the trilinear symmetric form F(u,v,w) defined by polarizing f, so that f(v)=F(v,v,v). Concretely, for f(x)=a x^3 on one variable, the associated trilinear map is F(x,y,z)=a (x y z) up to the combinatorial normalization factor.

Misapplication

Applying multilinearization blindly over rings with torsion and expecting the same formulas: dividing by factorials may be invalid, yielding non-unique or nonexistent multilinear companions. Also treating an analytic but nonpolynomial map as if it had a finite multilinearization without specifying truncation is misleading.

Consequence

Multilinearization produces multilinear invariants and tensor representatives of nonlinear algebraic data, enabling the use of representation theory, tensor contractions, and universal multilinear constructions; it clarifies symmetries and reduces some nonlinear problems to linear algebra over tensor powers.

Reversal

The inverse perspective is diagonalization or symmetrization collapse: recovering the original homogeneous polynomial by evaluating the multilinear form on repeated arguments (the diagonal), which loses multilinear flexibility and conflates tensor positions.

Boundary

Applies primarily to homogeneous polynomial maps or maps admitting a finite-degree polynomial expansion; over rings where division by integers is not allowed the canonical normalization may fail. It does not generally apply to arbitrary nonpolynomial functions, nonhomogeneous maps without decomposition, or to structures inherently nonmultilinear (e.g., certain nonassociative bilinear forms may not polarize nicely).

Semantic Tension

Competing meanings arise between polarization (constructing a symmetric multilinear form from a polynomial) and naive linearization (first-order tangent linear approximation). Multilinearization is algebraic and degree-sensitive, whereas linearization usually refers to first-order approximations that drop higher-degree structure.

Synthesis

Multilinearization is the algebraic process of extracting degree-k multilinear tensors from homogeneous polynomial parts by polarization and symmetrization, subject to base ring constraints; it converts nonlinear homogeneous data into multilinear objects suitable for tensor techniques while cautioning about denominators, uniqueness, and domain of validity.