Bilinear form#Reflexivity and orthogonality

Let {{math|V}} be an {{mvar|n}}-dimensional vector space with basis {{math|{e₁, …, e_n}}}.

The {{math|n × n}} matrix A, defined by {{math|1=A_ij = B(e_i, e_j)}} is called the matrix of the bilinear form on the basis {{math|{e₁, …, e_n}}}.

If the {{math|n × 1}} matrix {{math|x}} represents a vector {{math|x}} with respect to this basis, and similarly, the {{math|n × 1}} matrix {{math|y}} represents another vector {{math|y}}, then:

$$B(\backslash mathbf\{x\},\; \backslash mathbf\{y\})\; =\; \backslash mathbf\{x\}^\backslash textsf\{T\}\; A\backslash mathbf\{y\}\; =\; \backslash sum\_\{i,j=1\}^n\; x\_i\; A\_\{ij\}\; y\_j.$$

A bilinear form has different matrices on different bases. However, the matrices of a bilinear form on different bases are all congruent. More precisely, if {{math|{f₁, …, f_n}}} is another basis of {{mvar|V}}, then

$$\backslash mathbf\{f\}\_j=\backslash sum\_\{i=1\}^n\; S\_\{i,j\}\backslash mathbf\{e\}\_i,$$

where the $S\_\{i,j\}$ form an invertible matrix {{mvar|S}}. Then, the matrix of the bilinear form on the new basis is {{math|S^TAS}}.

{{further|Degenerate bilinear form}}

Every bilinear form {{math|B}} on {{mvar|V}} defines a pair of linear maps from {{mvar|V}} to its dual space {{math|V^∗}}. Define {{math|B₁, B₂: V → V^∗}} by

{{block indent|left=1.6|text={{math|1=B₁(v)(w) = B(v, w)}}}}

{{block indent|left=1.6|text={{math|1=B₂(v)(w) = B(w, v)}}}}

This is often denoted as

{{block indent|left=1.6|text={{math|1=B₁(v) = B(v, ⋅)}}}}

{{block indent|left=1.6|text={{math|1=B₂(v) = B(⋅, v)}}}}

where the dot ( ⋅ ) indicates the slot into which the argument for the resulting linear functional is to be placed (see Currying).

For a finite-dimensional vector space {{mvar|V}}, if either of {{math|B₁}} or {{math|B₂}} is an isomorphism, then both are, and the bilinear form {{math|B}} is said to be nondegenerate. More concretely, for a finite-dimensional vector space, non-degenerate means that every non-zero element pairs non-trivially with some other element:

:$B(x,y)=0$ for all $y\; \backslash in\; V$ implies that {{math|1=x = 0}} and

:$B(x,y)=0$ for all $x\; \backslash in\; V$ implies that {{math|1=y = 0}}.

The corresponding notion for a module over a commutative ring is that a bilinear form is {{visible anchor|unimodular}} if {{math|V → V^∗}} is an isomorphism. Given a finitely generated module over a commutative ring, the pairing may be injective (hence "nondegenerate" in the above sense) but not unimodular. For example, over the integers, the pairing {{math|1=B(x, y) = 2xy}} is nondegenerate but not unimodular, as the induced map from {{math|1=V = Z}} to {{math|1=V^∗ = Z}} is multiplication by 2.

If {{mvar|V}} is finite-dimensional then one can identify {{mvar|V}} with its double dual {{math|V^∗∗}}. One can then show that {{math|B₂}} is the transpose of the linear map {{math|B₁}} (if {{mvar|V}} is infinite-dimensional then {{math|B₂}} is the transpose of {{math|B₁}} restricted to the image of {{mvar|V}} in {{math|1=V^∗∗}}). Given {{math|B}} one can define the transpose of {{math|B}} to be the bilinear form given by

{{block indent|left=1.6|text=^tB(v, w) = B(w, v).}}

The left radical and right radical of the form {{math|B}} are the kernels of {{math|B₁}} and {{math|B₂}} respectively;{{sfn|Jacobson|2009|page=346}} they are the vectors orthogonal to the whole space on the left and on the right.{{sfn|Zhelobenko|2006|page=11}}

If {{mvar|V}} is finite-dimensional then the rank of {{math|B₁}} is equal to the rank of {{math|B₂}}. If this number is equal to {{math|dim(V)}} then {{math|B₁}} and {{math|B₂}} are linear isomorphisms from {{mvar|V}} to {{math|V^∗}}. In this case {{math|B}} is nondegenerate. By the rank–nullity theorem, this is equivalent to the condition that the left and equivalently right radicals be trivial. For finite-dimensional spaces, this is often taken as the definition of nondegeneracy:

{{block indent|left=1.6|text= Definition: B is nondegenerate if {{math|1=B(v, w) = 0}} for all w implies {{math|1=v = 0}}.}}

Given any linear map {{math|1=A : V → V^∗}} one can obtain a bilinear form B on V via

{{block indent|left=1.6|text=B(v, w) = A(v)(w).}}

This form will be nondegenerate if and only if {{math|A}} is an isomorphism.

If {{mvar|V}} is finite-dimensional then, relative to some basis for {{mvar|V}}, a bilinear form is degenerate if and only if the determinant of the associated matrix is zero. Likewise, a nondegenerate form is one for which the determinant of the associated matrix is non-zero (the matrix is non-singular). These statements are independent of the chosen basis. For a module over a commutative ring, a unimodular form is one for which the determinant of the associate matrix is a unit (for example 1), hence the term; note that a form whose matrix determinant is non-zero but not a unit will be nondegenerate but not unimodular, for example {{math|1=B(x, y) = 2xy}} over the integers.

We define a bilinear form to be

• symmetric if {{math|1=B(v, w) = B(w, v)}} for all {{math|v}}, {{math|w}} in {{mvar|V}};
• alternating if {{math|1= B(v, v) = 0}} for all {{math|v}} in {{mvar|V}};
• {{visible anchor|skew-symmetric bilinear form|text=skew-symmetric}} or {{visible anchor|antisymmetric bilinear form|text=antisymmetric}} if {{math|1=B(v, w) = −B(w, v)}} for all {{math|v}}, {{math|w}} in {{mvar|V}};
• ; Proposition: Every alternating form is skew-symmetric.
• ; Proof: This can be seen by expanding {{math|B(v + w, v + w)}}.

If the characteristic of {{mvar|K}} is not 2 then the converse is also true: every skew-symmetric form is alternating. However, if {{math|1=char(K) = 2}} then a skew-symmetric form is the same as a symmetric form and there exist symmetric/skew-symmetric forms that are not alternating.

A bilinear form is symmetric (respectively skew-symmetric) if and only if its coordinate matrix (relative to any basis) is symmetric (respectively skew-symmetric). A bilinear form is alternating if and only if its coordinate matrix is skew-symmetric and the diagonal entries are all zero (which follows from skew-symmetry when {{math|char(K) ≠ 2}}).

A bilinear form is symmetric if and only if the maps {{math|B₁, B₂: V → V^∗}} are equal, and skew-symmetric if and only if they are negatives of one another. If {{math|char(K) ≠ 2}} then one can decompose a bilinear form into a symmetric and a skew-symmetric part as follows

$$B^\{+\}\; =\; \backslash tfrac\{1\}\{2\}\; (B\; +\; \{\}^\{\backslash text\{t\}\}B)\; \backslash qquad\; B^\{-\}\; =\; \backslash tfrac\{1\}\{2\}\; (B\; -\; \{\}^\{\backslash text\{t\}\}B)\; ,$$

where {{math|^tB}} is the transpose of {{math|B}} (defined above).

{{block indent|left=1| Definition: A bilinear form {{math|B : V × V → K}} is called reflexive if {{math|1=B(v, w) = 0}} implies {{math|1=B(w, v) = 0}} for all v, w in V.}}

{{block indent|left=1| Definition: Let {{math|B : V × V → K}} be a reflexive bilinear form. v, w in V are orthogonal with respect to B if {{math|1=B(v, w) = 0}}.}}

A bilinear form {{math|B}} is reflexive if and only if it is either symmetric or alternating.{{sfn|Grove|1997}} In the absence of reflexivity we have to distinguish left and right orthogonality. In a reflexive space the left and right radicals agree and are termed the kernel or the radical of the bilinear form: the subspace of all vectors orthogonal with every other vector. A vector {{math|v}}, with matrix representation {{math|x}}, is in the radical of a bilinear form with matrix representation {{math|A}}, if and only if {{math|1=Ax = 0 ⇔ x^TA = 0}}. The radical is always a subspace of {{math|V}}. It is trivial if and only if the matrix {{math|A}} is nonsingular, and thus if and only if the bilinear form is nondegenerate.

Suppose {{mvar|W}} is a subspace. Define the orthogonal complement{{sfn|Adkins|Weintraub|1992|page=359}}

$$W^\{\backslash perp\}\; =\; \backslash left\backslash \{\backslash mathbf\{v\}\; \backslash mid\; B(\backslash mathbf\{v\},\; \backslash mathbf\{w\})\; =\; 0\; \backslash text\{\; for\; all\; \}\; \backslash mathbf\{w\}\; \backslash in\; W\backslash right\backslash \}\; .$$

For a non-degenerate form on a finite-dimensional space, the map {{math|V/W → W^⊥}} is bijective, and the dimension of {{math|W^⊥}} is {{math|dim(V) − dim(W)}}.

Definition: A bilinear form on a normed vector space {{math|(V, ‖⋅‖)}} is bounded, if there is a constant {{math|C}} such that for all {{math|u, v ∈ V}},

$$B\; (\; \backslash mathbf\{u\}\; ,\; \backslash mathbf\{v\})\; \backslash le\; C\; \backslash left\backslash |\; \backslash mathbf\{u\}\; \backslash right\backslash |\; \backslash left\backslash |\backslash mathbf\{v\}\; \backslash right\backslash |\; .$$

Definition: A bilinear form on a normed vector space {{math|(V, ‖⋅‖)}} is elliptic, or coercive, if there is a constant {{math|c > 0}} such that for all {{math|u ∈ V}},

$$B\; (\; \backslash mathbf\{u\}\; ,\; \backslash mathbf\{u\})\; \backslash ge\; c\; \backslash left\backslash |\; \backslash mathbf\{u\}\; \backslash right\backslash |\; ^2\; .$$

{{further|Quadratic form#Definitions}}

For any bilinear form {{math|B : V × V → K}}, there exists an associated quadratic form {{math|Q : V → K}} defined by {{math|Q : V → K : v ↦ B(v, v)}}.

When {{math|char(K) ≠ 2}}, the quadratic form Q is determined by the symmetric part of the bilinear form B and is independent of the antisymmetric part. In this case there is a one-to-one correspondence between the symmetric part of the bilinear form and the quadratic form, and it makes sense to speak of the symmetric bilinear form associated with a quadratic form.

When {{math|1=char(K) = 2}} and {{math|dim V > 1}}, this correspondence between quadratic forms and symmetric bilinear forms breaks down.

By the universal property of the tensor product, there is a canonical correspondence between bilinear forms on {{mvar|V}} and linear maps {{math|V ⊗ V → K}}. If {{math|B}} is a bilinear form on {{mvar|V}} the corresponding linear map is given by

{{block indent|left=1.6|text= {{math|v ⊗ w ↦ B(v, w)}}}}

In the other direction, if {{math|F : V ⊗ V → K}} is a linear map the corresponding bilinear form is given by composing F with the bilinear map {{math|V × V → V ⊗ V}} that sends {{math|(v, w)}} to {{math|v⊗w}}.

The set of all linear maps {{math|V ⊗ V → K}} is the dual space of {{math|V ⊗ V}}, so bilinear forms may be thought of as elements of {{math|(V ⊗ V)^∗}} which (when {{mvar|V}} is finite-dimensional) is canonically isomorphic to {{math|V^∗ ⊗ V^∗}}.

Likewise, symmetric bilinear forms may be thought of as elements of {{math|(Sym²V)^*}} (dual of the second symmetric power of {{math|V}}) and alternating bilinear forms as elements of {{math|(Λ²V)^∗ ≃ Λ²V^∗}} (the second exterior power of {{math|V^∗}}). If {{math|char(K) ≠ 2}}, {{math|(Sym²V)^* ≃ Sym²(V^∗)}}.

Much of the theory is available for a bilinear mapping from two vector spaces over the same base field to that field

{{block indent|left=1.6|text={{math|B : V × W → K}}.}}

Here we still have induced linear mappings from {{mvar|V}} to {{math|W^∗}}, and from {{mvar|W}} to {{math|V^∗}}. It may happen that these mappings are isomorphisms; assuming finite dimensions, if one is an isomorphism, the other must be. When this occurs, B is said to be a perfect pairing.

In finite dimensions, this is equivalent to the pairing being nondegenerate (the spaces necessarily having the same dimensions). For modules (instead of vector spaces), just as how a nondegenerate form is weaker than a unimodular form, a nondegenerate pairing is a weaker notion than a perfect pairing. A pairing can be nondegenerate without being a perfect pairing, for instance {{math|Z × Z → Z}} via {{math|(x, y) ↦ 2xy}} is nondegenerate, but induces multiplication by 2 on the map {{math|Z → Z^∗}}.

Terminology varies in coverage of bilinear forms. For example, F. Reese Harvey discusses "eight types of inner product".{{sfn|Harvey|1990|page=22}} To define them he uses diagonal matrices A_ij having only +1 or −1 for non-zero elements. Some of the "inner products" are symplectic forms and some are sesquilinear forms or Hermitian forms. Rather than a general field {{mvar|K}}, the instances with real numbers {{math|R}}, complex numbers {{math|C}}, and quaternions {{math|H}} are spelled out. The bilinear form

$$\backslash sum\_\{k=1\}^p\; x\_k\; y\_k\; -\; \backslash sum\_\{k=p+1\}^n\; x\_k\; y\_k$$

is called the real symmetric case and labeled {{math|R(p, q)}}, where {{math|1=p + q = n}}. Then he articulates the connection to traditional terminology:{{sfn|Harvey|1990|page=23}}

{{quote|

Some of the real symmetric cases are very important. The positive definite case {{nowrap|R(n, 0)}} is called Euclidean space, while the case of a single minus, {{nowrap|R(n−1, 1)}} is called Lorentzian space. If {{nowrap|1=n = 4}}, then Lorentzian space is also called Minkowski space or Minkowski spacetime. The special case {{nowrap|R(p, p)}} will be referred to as the split-case.

}}

Given a ring {{mvar|R}} and a right Module (mathematics) {{math|M}} and its dual module {{math|M^∗}}, a mapping {{math|B : M^∗ × M → R}} is called a bilinear form if

{{block indent|left=1.6|text={{math|1=B(u + v, x) = B(u, x) + B(v, x)}}}}

{{block indent|left=1.6|text={{math|1=B(u, x + y) = B(u, x) + B(u, y)}}}}

{{block indent|left=1.6|text={{math|1=B(αu, xβ) = αB(u, x)β}}}}

for all {{math|u, v ∈ M^∗}}, all {{math|x, y ∈ M}} and all {{math|α, β ∈ R}}.

The mapping {{math|⟨⋅,⋅⟩ : M^∗ × M → R : (u, x) ↦ u(x)}} is known as the natural pairing, also called the canonical bilinear form on {{math|M^∗ × M}}.{{sfn|Bourbaki|1970|page=233}}

A linear map {{math|S : M^∗ → M^∗ : u ↦ S(u)}} induces the bilinear form {{math|B : M^∗ × M → R : (u, x) ↦ ⟨S(u), x⟩}}, and a linear map {{math|T : M → M : x ↦ T(x)}} induces the bilinear form {{math|B : M^∗ × M → R : (u, x) ↦ ⟨u, T(x)⟩}}.

Conversely, a bilinear form {{math|B : M^∗ × M → R}} induces the R-linear maps {{math|S : M^∗ → M^∗ : u ↦ (x ↦ B(u, x))}} and {{math|T′ : M → M^∗∗ : x ↦ (u ↦ B(u, x))}}. Here, {{math|M^∗∗}} denotes the double dual of {{math|M}}.

{{cmn|colwidth=20em|

• Bilinear map
• :Category:Bilinear maps
• Inner product space
• Linear form
• Multilinear form
• Polar space
• Quadratic form
• Sesquilinear form
• System of bilinear equations
• Metric tensor

}}

{{reflist}}

{{refbegin}}

• {{citation | last1=Adkins | first1=William A. | last2=Weintraub | first2=Steven H. | year=1992 | title=Algebra: An Approach via Module Theory | series=Graduate Texts in Mathematics | volume=136 | publisher=Springer-Verlag | isbn=3-540-97839-9 | zbl=0768.00003 }}
• {{citation | last=Bourbaki | first=N. | year=1970 | title=Algebra | publisher=Springer |author-link=Nicolas Bourbaki}}
• {{citation | last=Cooperstein | first=Bruce | year=2010 | title=Advanced Linear Algebra | chapter=Ch 8: Bilinear Forms and Maps | pages=249–88 | publisher=CRC Press | isbn=978-1-4398-2966-0 }}
• {{citation | last=Grove | first=Larry C. | year=1997 | title=Groups and characters | publisher=Wiley-Interscience | isbn=978-0-471-16340-4}}
• {{citation | last1=Halmos | first1=Paul R. | author1-link=Paul R. Halmos | year=1974 | title=Finite-dimensional vector spaces | series=Undergraduate Texts in Mathematics | publisher=Springer-Verlag | location=Berlin, New York | isbn=978-0-387-90093-3 | zbl=0288.15002 }}
• {{citation | last1=Harvey | first1=F. Reese | year=1990 | title=Spinors and calibrations | chapter=Chapter 2: The Eight Types of Inner Product Spaces | pages=19–40 | publisher=Academic Press | isbn=0-12-329650-1 }}
• {{citation | last = Popov | first = V. L. | editor=Hazewinkel, M. | year=1987 | title = Bilinear form | url= https://www.encyclopediaofmath.org/index.php/Bilinear_form |encyclopedia=Encyclopedia of Mathematics | volume=1 | pages=390–392 | publisher=Kluwer Academic Publishers}}. Also: {{Google books|RVr5BwAAQBAJ|page=390}}
• {{citation | last=Jacobson | first=Nathan | year=2009 | title=Basic Algebra | volume=I | publisher=Courier Corporation | edition=2nd | isbn=978-0-486-47189-1 }}
• {{citation | last1=Milnor | first1=J. | author1-link=John Milnor| first2=D. | last2=Husemoller | year=1973 | title=Symmetric Bilinear Forms | series=Ergebnisse der Mathematik und ihrer Grenzgebiete | volume=73 | publisher=Springer-Verlag | isbn=3-540-06009-X | zbl=0292.10016 }}
• {{citation | last=Porteous | first=Ian R. | author-link=Ian R. Porteous | year=1995 | title=Clifford Algebras and the Classical Groups | series=Cambridge Studies in Advanced Mathematics | volume=50 | publisher=Cambridge University Press | isbn=978-0-521-55177-9 }}
• {{citation | last=Shafarevich | first=I. R. | author-link=Igor Shafarevich | author2=A. O. Remizov | year=2012 | title = Linear Algebra and Geometry | publisher=Springer | url=https://www.springer.com/mathematics/algebra/book/978-3-642-30993-9 | isbn=978-3-642-30993-9}}
• {{citation | last=Shilov | first=Georgi E. | title=Linear Algebra | editor-last=Silverman | editor-first=Richard A. | year=1977 | publisher=Dover | isbn=0-486-63518-X}}
• {{citation | last=Zhelobenko | first=Dmitriĭ Petrovich | year=2006 | title=Principal Structures and Methods of Representation Theory | series=Translations of Mathematical Monographs | publisher=American Mathematical Society | isbn=0-8218-3731-1 }}

{{refend}}

{{Commonscat|Bilinear forms}}

• {{springer|title=Bilinear form|id=p/b016250}}
• {{planetmath reference|urlname=BilinearForm|title=Bilinear form}}

{{Functional Analysis}}

{{PlanetMath attribution|id=7553|title=Unimodular}}

Category:Abstract algebra

Category:Linear algebra

Category:Multilinear algebra