1 Definition and basic properties
A bilinear form is a map that combines two vectors to produce a scalar in a way that is linear in each input separately. It is one of the most basic constructions in linear algebra and serves as a standard language for expressing interactions between vectors. Bilinear forms appear naturally in geometry, algebra, and many applied settings.
1.1 Formal definition
Let \(V\) and \(W\) be vector spaces over the same field \(F\). A map \(B: V \times W \to F\) is called bilinear if, for each fixed vector in one argument, the resulting map in the other argument is linear. When \(V = W\), the form is often simply called a bilinear form on \(V\).
1.2 Linearity in each argument
Linearity in the first variable means that for vectors \(u,v \in V\) and scalars \(a,b \in F\), one has \(B(au+bv, w) = aB(u,w) + bB(v,w)\). A similar identity holds in the second variable. These two conditions together imply that the form respects addition and scalar multiplication independently in each slot.
1.3 Matrix representation
In finite-dimensional spaces, choosing bases for the two vector spaces turns a bilinear form into a matrix. If \(x\) and \(y\) are coordinate column vectors, then the form can be written as \(B(x,y) = x^{T}Ay\), where \(A\) is the matrix of the form in the chosen bases. This representation makes computation straightforward and connects bilinear forms to matrix algebra.
1.4 Examples
Bilinear forms arise in familiar settings, from ordinary dot products to algebraic pairings defined by determinants. Many standard constructions in mathematics can be understood as special bilinear forms with additional properties.
1.4.1 Dot product
The standard dot product on \(F^{n}\) is a bilinear form given by \(x \cdot y = \sum_{i=1}^{n} x_i y_i\), when the field and context support this expression. It measures alignment between vectors and is central to Euclidean geometry and many computational methods.
1.4.2 Determinant-like forms
On two-dimensional spaces, the determinant of two vectors can define an alternating bilinear form. For vectors \(u = (u_1,u_2)\) and \(v = (v_1,v_2)\), the quantity \(u_1v_2 - u_2v_1\) is bilinear and changes sign when the two vectors are swapped. Such forms are closely tied to oriented area.
1.5 Basic identities
Bilinear forms satisfy several useful identities obtained by expanding linear combinations in either argument. For example, \(B(u+v,w+z)\) expands into four terms. These identities are used repeatedly in proofs, computations, and the derivation of structural properties.
2 Types of bilinear forms
Bilinear forms are often classified by how they behave under interchange of their inputs and by whether they vanish on nonzero vectors in an essential way. These distinctions play an important role in geometry and algebra.
2.1 Symmetric bilinear forms
A bilinear form \(B\) is symmetric if \(B(u,v) = B(v,u)\) for all vectors \(u,v\). Symmetric forms are the natural bilinear objects behind many notions of length, angle, and quadratic measurement. Their matrix representations are symmetric matrices in suitable bases.
2.2 Alternating bilinear forms
An alternating bilinear form satisfies \(B(v,v)=0\) for every vector \(v\). This condition forces the form to vanish on repeated arguments and makes such forms especially important in geometry and symplectic theory.
2.2.1 Skew-symmetry
For bilinear forms over fields of characteristic not equal to 2, alternating forms are skew-symmetric, meaning \(B(u,v) = -B(v,u)\). This relation follows from the identity \(0 = B(u+v,u+v)\) after expanding by bilinearity.
2.2.2 Consequences in characteristic not equal to 2
When 2 is invertible in the field, skew-symmetric and alternating behavior are closely linked. In that setting, a skew-symmetric bilinear form automatically has zero diagonal values, and many classification results become cleaner because the sign change under interchange has strong algebraic consequences.
2.3 Nondegenerate bilinear forms
A bilinear form is nondegenerate if no nonzero vector is orthogonal to every vector in the space under the form. Equivalently, the only vector that pairs to zero with all others is the zero vector. Nondegeneracy is a key condition in many classification theorems and guarantees that the form captures the full dimension of the space.
2.4 Degenerate bilinear forms
If a nonzero vector pairs to zero with everything, the form is degenerate. Degeneracy indicates that the bilinear form fails to distinguish some directions in the space. Such forms often arise from singular matrices or from constructions that intentionally ignore part of the geometry.
2.5 Hermitian and sesquilinear comparisons
Hermitian and sesquilinear forms are related notions in which one or both variables are conjugate-linear rather than linear. They are not bilinear in the strict sense, but they are often studied alongside bilinear forms because they serve similar roles in geometry and analysis. The comparison helps clarify which results depend on pure linearity and which depend on conjugation.
3 Relation to other algebraic objects
Bilinear forms connect naturally to quadratic forms, tensor products, and linear transformations. These relations explain why they appear across many branches of mathematics.
3.1 Quadratic forms
A quadratic form is a function \(Q(v)\) that is homogeneous of degree two. In many cases, a symmetric bilinear form determines a quadratic form by setting \(Q(v)=B(v,v)\). Conversely, a quadratic form can often be used to recover a related symmetric bilinear form.
3.1.1 Polarization identities
Polarization identities express a bilinear form in terms of a quadratic form. Over fields where 2 is invertible, one typical identity is \(B(u,v) = \frac{1}{2}\bigl(Q(u+v) - Q(u) - Q(v)\bigr)\). This formula shows how the mixed interaction between two vectors can be extracted from a single-variable function.
3.1.2 Recovering bilinear forms from quadratic forms
When the field and characteristic permit, the quadratic form determines the symmetric bilinear form uniquely. This recovery is central in the study of quadratic spaces, where the bilinear and quadratic viewpoints complement one another.
3.2 Inner products
An inner product is a positive-definite symmetric bilinear form in the real case, or a related sesquilinear form in the complex case. Inner products provide notions of orthogonality, projection, and norm. They are among the most familiar examples of bilinear-like structures in mathematics.
3.3 Linear maps and tensor products
Bilinear forms can be understood as linear maps once the domain is replaced by a tensor product. This viewpoint simplifies many arguments and reveals the universal algebraic meaning of bilinearity.
3.3.1 Bilinear maps as linear maps on tensor products
Given a bilinear map \(B: V \times W \to F\), there is a corresponding linear map \(\widetilde{B}: V \otimes W \to F\). The bilinear map and the linear map contain the same information, but the tensor product packages it in a more flexible form.
3.3.2 Universal property of the tensor product
The tensor product \(V \otimes W\) is characterized by the property that every bilinear map from \(V \times W\) factors uniquely through a linear map on \(V \otimes W\). This universal property is one of the most important conceptual bridges in multilinear algebra.
4 Matrix and coordinate methods
In finite dimensions, bilinear forms are conveniently analyzed using coordinates and matrices. This approach allows one to compute invariants and compare forms under changes of basis.
4.1 Gram matrices
Given vectors \(v_1,\dots,v_n\), the Gram matrix of a bilinear form has entries \(B(v_i,v_j)\). It records all pairwise evaluations of the form on the chosen basis or family of vectors. Gram matrices are widely used in geometry and numerical linear algebra.
4.2 Change of basis
When the basis changes, the matrix of a bilinear form changes by a congruence transformation. If \(P\) is the change-of-basis matrix, then the new matrix is typically \(P^{T}AP\). This rule explains which properties of the form are basis-independent.
4.3 Rank and determinant
The rank of the matrix of a bilinear form measures how much of the space is detected by the form. The determinant, when defined for a square matrix representation, indicates whether the form is nondegenerate. A zero determinant signals degeneracy in finite dimensions.
4.4 Canonical forms
Many bilinear forms can be simplified by choosing a suitable basis. Canonical forms make the essential features of the form visible and reduce classification problems to a manageable set of cases.
4.4.1 Classification over different fields
The classification of bilinear forms depends strongly on the underlying field. Over algebraically closed fields, over the real numbers, and over finite fields, the list of normal forms can differ substantially. Field characteristics also influence whether alternating and symmetric forms behave in the same way or in distinct ways.
4.4.2 Diagonalization in special cases
Symmetric bilinear forms can often be diagonalized, especially over fields of characteristic not equal to 2. In such cases, a basis may be chosen so that the matrix has zeros off the diagonal. This greatly simplifies analysis and is frequently used in applications.
5 Orthogonality and radicals
Orthogonality concepts extend naturally from inner products to general bilinear forms, although some geometric intuition may change when positivity is absent. The radical and isotropic directions are central to understanding degeneracy and decomposition.
5.1 Orthogonal complements
Given a subspace \(U\), its orthogonal complement consists of all vectors that pair to zero with every vector in \(U\). This notion generalizes perpendicularity and helps organize a space according to the bilinear form. The size and behavior of orthogonal complements reflect the form’s rank and degeneracy.
5.2 Radical of a bilinear form
The radical is the set of vectors that are orthogonal to all vectors in the whole space. It is a subspace and coincides with the kernel of the associated linear map into the dual space. A trivial radical is equivalent to nondegeneracy.
5.3 Isotropic vectors and subspaces
A nonzero vector is isotropic if it pairs to zero with itself under the form. More generally, a subspace is isotropic if the form vanishes on all pairs of vectors from that subspace. Isotropic subspaces play an important role in symplectic and quadratic geometry.
5.4 Orthogonal direct sums
A space may sometimes be decomposed as an orthogonal direct sum of subspaces, meaning vectors from distinct summands pair to zero. Such decompositions simplify calculations and are useful in classification results. They also separate a form into independent components.
6 Special classes and applications
Bilinear forms are not only abstract objects; they are also tools that organize important structures in geometry, algebra, and computation. Several special classes recur across mathematics.
6.1 Inner product spaces
Inner product spaces use bilinear or sesquilinear forms to define lengths, angles, and orthogonality. They underpin much of Euclidean geometry, functional analysis, and numerical approximation. Projection formulas and least-squares methods depend on these structures.
6.2 Symplectic bilinear forms
A symplectic bilinear form is a nondegenerate alternating form, typically on an even-dimensional vector space. Such forms are foundational in symplectic geometry and Hamiltonian mechanics. Their key feature is that they provide a structured notion of pairing without a notion of length.
6.3 Pairings in algebra and geometry
Pairings appear in many contexts, including evaluation between a vector space and its dual, intersection pairings in topology, and algebraic pairings in representation theory. In each case, the bilinear form encodes how two objects interact. These pairings often reveal hidden structure or duality.
6.4 Bilinear forms on modules
Over rings rather than fields, one may define bilinear forms on modules. The theory becomes more delicate because bases may not exist and modules may have torsion. Even so, bilinear forms remain useful in algebraic number theory, module theory, and related areas.
6.5 Applications in solving equations and invariants
Bilinear forms help classify equations by changing variables to simpler coordinate systems. They also produce invariants such as rank, signature, and determinant-like quantities in suitable settings. These invariants distinguish forms that are equivalent under basis change.
7 Advanced topics
More advanced treatments of bilinear forms connect them with duality, scalar extension, tensorial constructions, and module theory. These perspectives are essential in modern algebra.
7.1 Dual spaces and adjoint maps
A bilinear form on \(V\) defines a map from \(V\) to its dual space by sending a vector to the functional obtained by pairing with it. This identification is central in studying adjoint operators and self-adjointness relative to a bilinear form. It also clarifies the meaning of nondegeneracy.
7.2 Bilinear forms on finite-dimensional spaces
In finite dimensions, bilinear forms can be studied through linear algebraic invariants such as rank, nullity, and matrix equivalence. Many structural questions reduce to finite computations. This makes the finite-dimensional case the standard setting for classification.
7.3 Extension of scalars
Extending the base field can alter the behavior of a bilinear form. A form that is difficult to classify over one field may simplify over a larger field. This technique is common when comparing real, complex, and algebraic settings.
7.4 Tensorial interpretation
The tensorial viewpoint treats a bilinear form as an element of the dual of a tensor product or as a tensor of type \((0,2)\). This interpretation is especially useful in geometry, where coordinate-free descriptions are preferred. It also fits naturally into the language of multilinear algebra.
7.5 Bilinear forms over rings and modules
When the coefficient system is a ring, bilinear forms must be handled with additional care. Notions such as duality, nondegeneracy, and classification may depend on module-theoretic properties rather than simple dimension counts. This broader framework is important in algebra and number theory.