base See base of logarithm.

See also base of

number system, basis.

B base of logarithm The number that forms the

base of the exponential to which the logarithm
is inverse. That is, a logarithm, base b, is the in-
verse of the exponential, base b. The logarithm
back substitution A technique connected is usually denoted by logb (unless the base is
with the Gaussian elimination method for solv- Euler’s constant e, when ln or log is used, log is
ing simultaneous linear equations. After the sys- also used for base 10 logarithm). A conversion
tem formula, from one base to another, is
a11 x1 + a12 x2 + · · · + a1n xn = b1 loga x = logb x loga b .
a21 x1 + a22 x2 + · · · + a2n xn = b2
.........
an1 x1 + an2 x2 + · · · + ann xn = bn base of number system The number which
is used as a base for successive powers, com-
is converted to triangular form binations of which are used to express all posi-
t11 x1 + t12 x2 + · · · + t1n xn = c1 tive integers and rational numbers. For example,
t22 x2 + · · · + t2n xn = c2 2543 in the base 7 system stands for the number
.........



2 73 + 5 72 + 4 71 + 3 .
tnn xn = cn .

One then solves for xn and then back substitutes Or, −524.37 in the base 8 system stands for the
this value for xn into the equation number





− 5 82 + 2 81 + 4 + 3 8−1 + 7 8−2 .
tn−1 n−1 xn−1 + tnn xn = cn−1
The base 10 number system is called the deci-
and solves for xn−1 . Continuing in this way, all
mal system. For base n, the term n-ary is used;
of the variables x1 , x2 , . . . , xn can be solved for.
for example, ternary, in base 3.
backward error analysis A technique for
base point The point in a set to which a bun-
estimating the error in evaluating f (x1 , . . . , xn ),
dle of (algebraic) objects is attached. For exam-
assuming one knows f (a1 , . . . , an ) = b and has
ple, a vector bundle V defined over a manifold
control of |xi − ai | for 1 ≤ i ≤ n.
M will have to each point b ∈ M an associated
vector space Vb . The point b is the base point
Baer’s sum For given R-modules A and C,
for the vectors in Vb .
the sum of two elements of the Abelian group
ExtR (C, A).
base term For a spectral sequence E =
{E r , d r } where d r : Ep,q
r r
→ Ep−r,q+r−1 ,r =
Bairstow method of solving algebraic equa- r
2, 3, . . . and Ep,q = 0 whenever p < 0 or
tions An iterative method for finding qua- r .
q < 0, a term of the form Ep,o
dratic factors of a polynomial. The goal being
to obtain complex roots that are conjugate pairs.
basic feasible solution A type of solution of
the linear equation Ax = b. If A is an m × n
Banach algebra An algebra over the com-
matrix with m ≤ n and b ∈ Rm , suppose A =
plex numbers with a norm  · , under which it
A1 + A2 when A1 is an m × m nonsingular
is a Banach space and such that
matrix. It is a solution of the form A(x1 , 0) =
xy ≤ xy A1 x1 + A2 0 = b where x1 ∈ Rm and x1 ≥ 0.

for all x, y ∈ B. If B is an algebra over the real basic form of linear programming problem
numbers, then B is called a real Banach algebra. The following form of a linear programming

c

2001 by CRC Press LLC
problem: Find a vector (x1 , x2 , . . . , xn ) which infinite (and convergent). If only finite sums are
minimizes the linear function permitted, a basis is referred to as a Hamel basis.

C = c1 x1 + c2 x2 + · · · + cn xn , Bernoulli method for finding roots An it-

erative method for finding a root of a polyno-
subject to the constraints
mial equation. If p(x) = a0 x n + an−1 x n−1 +
a11 x1 + a12 x2 + · · · + a1n xn = b1 · · · + an is a polynomial, then this method, ap-
a21 x1 + a22 x2 + · · · + a2n xn = b2 plied to p(x) = 0, consists of the following
... steps. First, choose some set of initial-values
am1 x1 + am2 x2 + · · · + amn xn = bm x0 , x−1 , . . . , x−n+1 . Second, define subsequent
values xm by the recurrence relation
and x1 ≥ 0, x2 ≥ 0, . . . , xn ≥ 0.
Here aij , bi , and cj are all real constants and a1 xm−1 + a2 xm−2 + · · · + an xm−n
xm = −
m < n. a0

basic invariants For a commutative ring K for m ≥ 1. Third, form the sequence of quo-
with identity and a ring R containing K and G tients rm = xm+1 /xm for m ≥ 1. If the polyno-
a subgroup of AutK (R), a minimal set of gener- mial has a single root, r, of largest magnitude,
ators of the ring R G . then the sequence {rm } will converge to r.

basic optimal solution A solution of a linear Bernoulli number Consider numbers Bn∗ de-
programming problem that minimizes the ob- fined by the functional equation
jective function (cost function) and is basic in ∞
the sense that, in the linear constraints x x  (−1)n−1 B ∗ x 2n
n
+ − 1 =
ex − 1 2 (2n)!
a11 x1 + a12 x2 + · · · + a1n xn = b1 n=1
a21 x1 + a22 x2 + · · · + a2n xn = b2 B1∗ x 2 B2∗ x 4 B3∗ x 6
......... = − + − +···
2! 4! 6!
am1 x1 + am2 x2 + · · · + amn xn = bm
or, alternatively, by the equation
for the problem, n − m values of the n variables
x1 , x2 , . . . , xn are zero. x
x  ∞
Bn∗ x 2n
1− cot =
2 2 (2n)!
basic variable A variable that has value zero n=1
in a linear programming problem. The basic B1∗ x 2 B2∗ x 4 B3∗ x 6
variables lie on the boundaries of the convex re- = + + + ··· .
2! 4! 6!
gions determined by the constraints in the prob-
lem. The numbers Bn∗ are called the Bernoulli num-
bers. The definition given here is the classical
basis A subset B of a vector space V which one. There are several alternative, and more
has the property that every vector v ∈ V can modern, definitions. Bernoulli numbers arise
be expressed uniquely as a finite linear com- in the theory of special functions, in the study
bination of elements of B. That is, if V is a of hypergeometric functions, and as the coeffi-
vector space over the field F , then for a given cients of the Taylor expansions of many classical
v ∈ V , there exists a unique, finite, collec- transcendental functions.
tion of vectors x1 , x2 , . . . , xn ∈ B and scalars
α1 , α2 , . . . , αn ∈ F such that x = α1 x1 +α2 x2 + Betti numbers The nth Betti number Bn ,
· · · + αn xn . of a manifold M, is the dimension of the nth
By definition, V is finite dimensional if it has cohomology group, H n (M, R). [The group H n
a finite basis. In an infinite dimensional vec- (M, R) is the quotient group consisting of equiv-
tor space, if there is a topology on V , the sum alence classes of the closed n forms modulo the
representing a vector x may be allowed to be differentials of (n − 1) forms.]

c

2001 by CRC Press LLC
Bezout’s Theorem If p1 (x) and p2 (x) are for all x, y, z ∈ V and α, β ∈ F . See also
two polynomials of degrees n1 and n2 , respec- quadratic form.
tively, having no common zeros, then there are
two unique polynomials q1 (x) and q2 (x) of de- bilinear function See multilinear function.
grees n1 − 1 and n2 − 1, respectively, such that
bilinear mapping A mapping b : V × V →
p1 (x)q1 (x) + p2 (x)q2 (x) = 1 . W , where V and W are vector spaces over the
field F , which satisfies

biadditive mapping For A-modules M, N b(αx + βy, z) = αb(x, z) + βb(y, z)

and L, the mapping f : M × N → L such that
    and
f x + x  , y = f (x, y) + f x  , y , and
    b(x, αy + βz) = αb(x, y) + βb(x, z)
f x, y + y  = f (x, y) + f x, y  ,
for all x, y, z ∈ V and α, β ∈ F .
x, x  ∈ M, y, y  ∈ L.
bilinear programming The area dealing with
bialgebra A vector space A over a field k that finding the extrema of functions
is both an algebra and a coalgebra over k. That
is, (A, µ, η, *, ε) is a bialgebra if (A, µ, η) is f (x1 , x2 ) = C1t x1 + C2t x2 + x1t Qx2
an algebra over k and (A, *, ε) is a coalgebra
over k, µ : A ⊗k A → A (multiplication). over
η : k → A (unit), * : A → A ⊗k A (co- 
multiplication). ε : A → k (counit) and these X = (x1 , x2 ) ∈ Rn1 × Rn2 : A1 x1 ≤ b1 ,
maps satisfy
A2 x2 ≤ b2 , x1 ≥ 0, x2 ≥ 0}
µ ◦ (µ ⊗ IA ) = µ ◦ (IA ⊗ µ) , where Q is an n1 × n2 real matrix, A1 is an
n1 × n1 , real matrix and A2 is an n2 × n2 real
µ ◦ (η ⊗ IA ) = µ ◦ (IA ⊗ η) , matrix.
(* ⊗ IA ) ◦ * = (IA ⊗ *) ◦ * ,
binary Diophantine equation A Diophan-
(ε ⊗ IA ) ◦ * = (IA ⊗ ε) ◦ *. tine equation in two unknowns. See Diophan-
tine equation.
bialgebra-homomorphism For (A, µ,
η, *, ε) and (A , µ , η , * , ε ) bialgebras over binary operation A mapping from the Carte-
a field k, a linear mapping f : A → A where sian product of a set with itself into the set.
f ◦η = η , f ◦µ = µ ◦(f ⊗f ), (f ⊗f )◦* = That is, if the set is denoted by S, a mapping
* ◦ f, ε = ε  ◦ f . See bialgebra. b : S × S → S. A notation, such as 3, is usually
adopted for the operation, so that b(x, y) = x3y.
biideal A linear subspace I of A, where (A,
µ, η, *, ε) is a bialgebra over k, such that µ(A binomial A sum of two monomials. For ex-
⊗k I ) = I and *(I ) ⊂ A ⊗k I + I ⊗k A. ample, if x and y are variables and α and β are
constants, then αx p y q +βx r y s , where p, q, r, s
bilinear form A mapping b : V × V → F , are integers, is a binomial expression.
where V is a vector space over the field F , which
satisfies binomial coefficients The numbers, often
denoted by (nk ), where n and k are nonnegative
b(αx + βy, z) = αb(x, z) + β(y, z) integers, with n ≥ k, given by

and n
= n!
k
b(x, αy + βz) = αb(x, y) + βb(x, z) k!(n − k)!

c

2001 by CRC Press LLC
where m! = m(m − 1) · · · (2)(1) and 0! = 1 and blowing up A process in algebraic geometry
1! = 1. The binomial coefficients appear in the whereby a point in a variety is replaced by the set
Binomial Theorem expansion of (x +y)n where of lines through that point. This idea of Zariski
n is a positive integer. See Binomial Theorem. turns a singular point of a given manifold into a
smooth point. It is used decisively in Hironaka’s
binomial equation An equation of the form celebrated “resolution of singularities” theorem.
x n − a = 0.
blowing up Let N be an n-dimensional com-
binomial The series (1 + x)α = pact, complex manifold (n ≥ 2), and p ∈ N .
∞ α seriesn
n=0 (n )x . It converges for all |x| < 1.
Let {z = (zi )} be a local coordinate system, in
a neighborhood U , centered at p and define
Binomial Theorem For  any nonnegative in- 
tegers b and n, (a + b)n = nj=o (nj )a j bn−j . Ũ = (z, l) ∈ U × P n−1 : z ∈ l ,

birational isomorphism A k-morphism ϕ : where P n−1 is regarded as a set of lines l in

G → G , where G and G are algebraic groups Cn . Let π : Ũ → U denote the projection
defined over k, that is a group isomorphism, π(z, l) = z. Identify π −1 (p) with P n−1 and
whose inverse is a k-morphism. Ũ \π −1 (p) with U \{p}, via the map π and set

Ñ = (N \{p}) ∪ Ũ , Bp (N ) = Ñ / ∼ ,
birational mapping For V and W irreducible
algebraic varieties defined over k, a closed irre- where z ∼ w if z ∈ N \{p} and w = (z, l) ∈ Ũ .
ducible subset T of V × W where the closure of The blowing up of N at p is π : Bp (N ) → N .
the projection T → V is V , the closure of the
projection T → W is W , and k(V ) = k(T ) = BN-pair A pair of subgroups (B, N ) of a
k(W ). Also called birational transformation. group G such that:
(i.) B and N generate G;
birational transformation See birational (ii.) B ∩ N = H *N ; and
mapping. (iii.) the group W = N/H has a set of genera-
tors R such that for any r ∈ R and any w ∈ W
Birch-Swinnerton-Dyer conjecture The (a) rBw ⊂ BwB ∩ BrwB,
rank of the group of rational points of an el- (b) rBr  = B.
liptic curve E is equal to the order of the 0 of
L(s, E) at s = 1. Consider the elliptic curve Bochner’s Theorem A function, defined on
E : y 2 = x 3 − ax − b where a and b are inte- R, is a Fourier-Stieltjes transform if and only if it
gers. If E(Q) = E ∩ (Q × Q), by Mordell’s is continuous and positive definite. [A function
Theorem E(Q) is a finitely generated Abelian f , defined on R, is defined to be positive definite
group. Let N be the conductor of E, and if if 
p  | N, let ap + p be the number of solutions of f (y)f (x − y)dy > 0
y 2 = (x 3 − ax − b) (mod p). The L-function R
of E, for all x-values.]
 
L(s, E) = 1 − εp p −s Borel subalgebra A maximal solvable sub-
p|N algebra of a reductive Lie algebra defined over

 an algebraically closed field of characteristic 0.
1 − ap p −s + p 1−2s
p|N Borel subgroup A maximal solvable sub-
group of a complex, connected, reductive Lie
where εp = 0 or ± 1. group.

block A term used in reference to vector bun- Borel-Weil Theorem If Gc is the complex-
dles, permutation groups, and representations. ification of a compact connected group G, any

c

2001 by CRC Press LLC
irreducible holomorphic representation of Gc is bounded variation Let I = [a, b] ⊆ R be
holomorphically induced from a one-dimension- a closed interval and f : I → R a function.
al holomorphic representation of a Borel sub- Suppose there is a constant C > 0 such that, for
group of Gc . any partition a = a0 ≤ a1 ≤ · · · ≤ ak = b it
holds that
boundary (1) (Topology.) The intersection
k

of the complements of the interior and exterior
|f (aj ) − f (aj −1 )| ≤ C .
of a set is called the boundary of the set. Or,
j =1
equivalently, a set’s boundary is the intersection
of its closure and the closure of its complement. Then f is said to be of bounded variation on the
(2) (Algebraic Topology.) A boundary in interval I .
a differential group C (an Abelian group with
homomorphism ∂ : C → C satisfying ∂∂ = 0) bracket product If a and b are elements of
is an element in the range of ∂. a ring R, then the bracket product is defined as
[a, b] = ab − ba. The bracket product satisfies
boundary group The group Im∂, which is a the distributive law.
subgroup of a differential group C consisting of
the image of the boundary operator ∂ : C → C. branch and bound integers programming
At step j of branch and bound integers pro-
boundary operator A homomorphism gramming for a problem list P a subproblem Pj
∂ : C → C of an Abelian group C that satisfies is selected and a lower bound is estimated for its
∂∂ = 0. Used in the field of algebraic topology. optimal objective function. If the lower bound
See also boundary, boundary group. is worse than that calculated at the previous step,
then Pj is discarded; otherwise Pj is separated
bounded homogeneous domain A bounded into two subproblems (the branch step) and the
domain with a transitive group of auto- process is repeated until P is empty.
morphisms. In more detail, a domain is a con-

nected open subset of complex N space CN . branch divisor The divisor iX X, where
A domain is homogeneous if it has a transi- iX is the differential index at a point X on a
tive group of analytic (holomorphic) automor- nonsingular curve.
phisms. This means that any pair of points z
and w can be interchanged, i.e., φ(z) = w, by Brauer group The Abelian group formed by
an invertible analytic map φ carrying the do- the tensor multiplication of algebras on the set of
main onto itself. For example, the unit ball in equivalence classes of finite dimensional central
complex N space, {z = (z1 , . . . , zN ) : |z1 |2 + simple algebras.
· · · + |zN |2 < 1}, is homogeneous. A domain
is bounded if it is contained in a ball of finite Brauer’s Theorem Let G be a finite group
radius. A bounded homogeneous domain is a and let χ be
 any character of G. Then χ can be
bounded domain which is also homogeneous. written as nk χψk , where nk is an integer and
Thus, the unit ball in CN is a bounded homo- each χψk is an induced character from a certain
geneous domain. There are many others. See linear character ψk of an elementary subgroup
also Siegel domain, Siegel domain of the sec- of G.
ond kind.
Bravais class An arithmetic crystal class de-
bounded matrix A continuous linear map termined by (L, B(L)), where L is a lattice and
K : ?2 (N) ⊗ ?2 (N) → ?1 (N) where N is the set B(L) is the Bravis group of L. See Bravais
of natural numbers. group.

bounded torsion group A torsion group T Bravais group The group of all orthogonal
where there is an integer n ≥ 0 such that t n = 1 transformations that leave invariant a given lat-
for all t ∈ T . tice L.

c

2001 by CRC Press LLC
Bravais lattice A representative of a Bravais apartments of C) such that every two simplices
type. See Bravais type. of C belong to an apartment and if A, B are in
S, then there exists an isomorphism of A onto
Bravais type An equivalence class of arith- B that fixes A ∩ B elementwise.
metically equivalent lattices. See arithmetical
equivalence. building of Euclidian type A building is of
Euclidean type if it could be used like a sim-
Brill-Noether number The quantity g − plical decomposition of a Euclidean space. See
(k + 1)(g − k + m), where g is the genus of building.
a nonsingular curve C and k and m are posi-
tive integers with k ≤ g. This quantity acts as a building of spherical type A building that
lower
 bound for the dimension of
 the subscheme has finitely many chambers. See building.
ϕ(D) : l(D) > m, deg D = k of the Jacobian
variety of C, where ϕ is the canonical function Burnside Conjecture A finite group of odd
from C to this variety. order is solvable.

Bruhat decomposition A decomposition of Burnside problem (1) The original Burn-

a connected semisimple algebraic group G, as a side problem can be stated as follows: If every
union of double cosets of a Borel subgroup B, element of a group G is of finite order and G
with respect to representatives chosen from the is finitely generated, then is G a finite group?
classes that comprise the Weyl group W of G. Golod has shown that the answer for p-groups
For each w ∈ W , let gw be a representative in is negative.
the normalizer N (B ∩ B − ) in G of the maximal (2) Another form of the Burnside problem is:
torus B ∩ B − formed from B and its opposite If a group G is finitely generated and the orders
subgroup B − . Then G is the disjoint union of of the elements of G divide an integer n, then is
the double cosets Bgw B as w ranges over W . G finite?

building A thick chamber complex C with

a system S of Coxeter subcomplexes (called the

c

2001 by CRC Press LLC
 canonical coordinates of the first kind For
each basis B1 , . . . , Bn of a Lie algebra L of
the Lie group G, there exists a positive
real

C number r with the property that {exp( bi Bi ) :

|bi | < r (i = 1, . . . , n)} is an open neigh-
borhood
of the identity element in G such that
exp( bi Bi )→(b1 , . . . , bn )(|bi | < r, i = 1,
Ci -field Let F be a field and let i, j be in- . . . , n) is a local coordinate system. These local
tegers such that i ≥ 0 and j ≥ 1. Also, let P coordinates are called the canonical coordinates
be a homogeneous polynomial of m variables of the first kind associated with the basis (Bi ) of
of degree j with coefficients in F . If the equa- this Lie algebra L.
tion P = 0 has a solution (s1 , s2 , . . . , sm )  =
(0, 0, . . . , 0) in F for any P such that m > j i , canonical coordinates of the second kind
then F is called a Ci (j ) field. If, for any j ≥ 1, For each basis B1 , . . . , Bn of a Lie algebra L
F is a Ci (j ) field, then F is called a Ci -field. of the Lie group  G, we have a local coordi-
nate system exp(bi Bi ) →(b1 , . . . , bn ) (i =
Calkin algebra Let H be a separable infinite 1, 2, . . . , n) in a neighborhood of the identity
dimensional Hilbert space, B(H ) the algebra of element in G. These b1 , . . . , bn are called the
bounded linear operators on H , and I (H ) be the canonical coordinates of the second kind asso-
ideal of H consisting of all compact operators. ciated with the basis (Bi ) of this Lie algebra L.
Then, the quotient C ∗ -algebra B(H )/I (H ) is
called the Calkin algebra. canonical divisor Any one of the linearly
equivalent divisors in the sheaf of relative dif-
Campbell-Hausdorff formula A long for- ferentials of a (nonsingular) curve.
mula for computation of z = ln(ex ey ) in the al-
gebra of formal power series in x and y with the canonical function A rational mapping φ :
assumption that x and y are associative but not X → J , from a nonsingular curve X to its Jaco-
commutative. It was first studied by Campbell. bian variety J , defined by φ(P ) = !(P − P0 ),
Then Hausdorff showed that z can be written in where P is a generic point of X and P0 is a fixed
terms of the commutators of x and y. rational point, ! : G0 (X)/Gl (X) → J is the
associated isomorphism, G(X) is the group of
cancellation Let x, y, and z be elements of divisors, G0 (X) is the subgroup of divisors of
a set S, with a binary operation ∗. The acts of degree 0 and Gl (X) the subgroup of divisors of
eliminating z in x ∗ z = y ∗ z or z ∗ x = z ∗ y functions. Such a φ is determined uniquely by
to obtain x = y is called cancellation. ! up to translation of J .

canonical homology basis A one-dimension-

cancellation law An axiom that allows can-
al homology basis {βi , βk+i }ki=1 such that
cellation.
(βi , βj ) = (βk+i , βk+j ) = 0, (βi , βk+i ) = 1,
and (βi , βk+j ) = 0 (i  = j ), (i, j = 1, 2, . . . , k).
canonical class A specified divisor class of
an algebraic curve.
canonical homomorphism (1) Let R be a
commutative ring with identity and let L, M
canonical cohomology class The 2-cocycle
be algebras over R. Then, the tensor product
in
L ⊗R M of R-modules is an algebra over R.
H 2 (Gal(K/k), IK ) ∼
= Z/nZ The mappings l → l ⊗ 1 (l ∈ L) and m →
m ⊗ 1 (m ∈ M) give algebra homorphisms
in the Galois cohomology of the Galois exten- L → L ⊗R M and M → L ⊗R M. Each
sion K/k of degree n with respect to the idéle one of these homomorphisms is called a canon-
class group IK that corresponds to 1 in Z/nZ ical homomorphism (on tensor products of al-
under the above isomorphism. gebras).

c

2001 by CRC Press LLC
 
(2) Let the ring R = i∈I Ri be the direct cap product (1) In a lattice or Boolean alge-
product of rings Ri . The mapping φi : R → Ri bra, the fundamental operation a ∧ b, also called
that assigns to each element r of R its ith com- the meet or product, of elements a and b.
ponent ri is called a canonical homomorphism (2) In cohomology theory, where Hr (X, Y ;
(of direct product of rings). G) and H s (X, Y ; G) are the homology and co-
homology groups of the pair (X, Y ) with coef-
canonical injection For a subgroup H of a ficients in the group G, the operation that as-
group G, the injective homomorphism θ :H → sociates to the pair (f, g), f ∈ Hr+s (X, Y ∪
G, defined by θ (h) = h for all h ∈ H . (θ is also Z; G1 ), g ∈ H r (X, Y ; G2 ) the element f ∪ g ∈
called the natural injection.) Hs (X, Y ∪ Z; G3 ) determined by the composi-
tion

canonically bounded complex Let F 0 (C) Hr+s (X, Y ∪ Z; G1 )

and F m+1 (C) (m an integer) be subcomplexes −→ Hr+s ((X, Y ) × (X, Z); G1 )
of a complex C such that F 0 (C) = C, and  
F m+1 (C m ) = 0, then the complex C is called a −→ Hom H r (X, Y ; G2 ) , Hs (X, Z; G3 )
canonically bounded complex.
where the first map is induced by the diagonal
map 0 : (X, Y ∪ Z) → (X, Y ) × (X, Z).
canonically polarized Jacobian variety A
pair, (J, P ), where J is a Jacobian variety whose Cardano’s formula A formula for the roots
polarization P is determined by a theta divisor. of the general cubic equation over the complex
numbers. Given the cubic equation ax 3 +bx 2 +
cx + d = 0, let A = 9abc − 2b3 − 27a 2 d
canonical projection Let S/ ∼ denote the
and B = b2 − 3ac. Also, let y1 and y2 be
set of equivalence classes of a set S, with re-
solutions of the quadratic equation Y 2 − AY +
spect to an equivalence relation ∼. The mapping
B 3 = 0. If ω is any cube root of 1, then (−b + ω
µ:S → S/ ∼ that carries s ∈ S to the equiva- √ √
3 y +ω 2 3 y )/3a is a root of the original cubic
lence class of s is called the canonical projection 1 2
equation.
(or quotient map).
cardinality A measure of the size of, or num-
canonical surjection (1) Let H be a normal ber of elements in, a set. Two sets S and T are
subgroup of a group G. For the factor group said to have the same cardinality if there is a
G/H , the surjective homomorphism θ:G → function f : S → T that is one-to-one and
G/H such that g ∈ θ (g), for all g ∈ G, is called onto. See also countable, uncountable.
the canonical surjection (or natural surjection)
to the factor group. Cartan integer Let R be the root system of
a Lie algebra L and let F = {x1 , x2 , . . . , xn } be
(2) Let G = G1 × G2 × . . . × Gn be the di-
a fundamental root system of R. Each of the
rect product of the groups G1 , G2 , . . . , Gn . The
n2 integers xij = −2(xi , xj )/(xj , xj ) (1 ≤ i,
mapping (g1 , g2 , . . . , gn ) → gi (i = 1, 2, . . . ,
j ≤ n) is called a Cartan integer of L, relative
n) from G to Gi is a surjective homomorphism,
to the fundamental root system F .
called the canonical surjection on the direct prod-
uct of groups. Cartan invariants Let G be a finite group
and let n be the number of p-regular classes
capacity of prime ideal Let A be a separable of G. Then, there are exactly n nonsimilar,
algebra of finite degree over the field of quotients absolutely irreducible, modular representations,
of a Dedekind domain. Let P be a prime ideal M1 , M2 , . . . , Mn , of G. Also, there are n non-
of A and let M be a fixed maximal order of similar, indecomposable components, denoted
A. Then, M/P is the matrix algebra of degree by R1 , R2 , . . . , Rn , of the regular representation
d over a division algebra. This d is called the R of G. These can be numbered in a such a way
capacity of the prime ideal P . that Mn appears in Rn as both its top and bottom

c

2001 by CRC Press LLC
component. If the degree of Mn is mn and the Cartan-Weyl Theorem A theorem that as-
degree of Rn is rn , then Rn appears mn times sists in the characterization of irreducible repre-
in R and Mn appears rn times in R. The multi- sentations of complex semisimple Lie algebras.
plicities µnt of Mt in Rn are called the Cartan Let G be a complex semisimple Lie algebra, H a
invariants of G. Cartan subalgebra,
8 the root system of G rela-
tive to H, α = σ ∈8 rσ σ , rσ ∈ R, a complex-
Cartan involution Let G be a connected valued linear functional on H, and ρ : G →
semisimple Lie group with finite center and let GLn (C) a representation of G. The functional α
M be a maximal compact subgroup of G. Then is a weight of the representation if the space of
there exists a unique involutive automorphism of vectors v ∈ Cn that satisfy ρ(h)v = α(h)v for
G whose fixed point set coincides with M. This all h ∈ H is nontrivial; Cn decomposes as a di-
automorphism is called a Cartan involution of rect sum of such spaces associated with weights
the Lie group G. α1 , . . . , αk . If we place a lexicographic linear
order ≤ on the set of functionals α, the Cartan-
Weyl Theorem asserts that there exists an irre-
Cartan-Mal’tsev-Iwasawa Theorem Let M
ducible representation ρ of G having α as its
be a maximal compact subgroup of a connected
highest weight (with respect to the order ≤) if
Lie group G. Then M is also connected and G is ]
and only if 2[α,σ[σ,σ ] is an integer for every σ ∈ 8,
homeomorphic to the direct product of M with
and w(α) ≤ α for every permutation w in the
a Euclidean space Rn .
Weyl group of G relative to H.

Cartan’s criterion of semisimplicity A Lie Carter subgroup Any finite solvable group
algebra L is semisimple if and only if the Killing contains a self-normalizing, nilpotent subgroup,
form K of L is nondegenerate. called a Carter subgroup.

Cartan’s criterion of solvability Let gl(n, Cartesian product If X and Y are sets, then
K) be the general linear Lie algebra of degree the Cartesian product of X and Y , denoted X ×
n over a field K and let L be a subalgebra of Y , is the set of all ordered pairs (x, y) with x ∈ X
gl(n, K). Then L is solvable if and only if and y ∈ Y .
tr(AB) = 0 (tr(AB) = trace of AB), for every
A ∈ L and B ∈ [L, L]. Cartier divisor A divisor which is linearly
equivalent to the divisor 0 on a neighborhood of
each point of an irreducible variety V .
Cartan’s Theorem (1) E. Cartan’s Theo-
rem. Let W1 and W2 be the highest weights of
Casimir element Let β1 , . . . , βn be a basis of
irreducible representations w1 , w2 of the Lie al-
the semisimple Lie algebra L. Using the Killing
gebra L, respectively. Then w1 is equivalent to
form K of L, let mij = K(βi , βj ). Also, let
w2 if and only if W1 = W2 .
mij represent the inverse of the matrix (mij ) and
(2) H. Cartan’s Theorem. The sheaf of let c be an element of the quotient

associative
ideals defined by an analytic subset of a com- algebra Q(L), defined by c = mij βi βj . This
plex manifold is coherent. element c is called the Casimir element of the
semisimple Lie algebra L.
Cartan subalgebra A subalgebra A of a Lie
algebra L over a field K, such that A is nilpotent Casorati’s determinant The n × n determi-
and the normalizer of A in L is A itself. nant
D(c1 (x), . . . , cn (x)) =
Cartan subgroup A subgroup H of a group 
 c1 (x) c2 (x) ··· cn (x)


 
G such that H is a maximal nilpotent subgroup 



 c1 (x + 1) c2 (x + 1) ··· cn (x + 1) 
of G and, for every subgroup K of H of finite 


 ,
 
index in H , the normalizer of K in G is also of 

··· 

 
finite index in K.  c 
1 (x + n − 1) c2 (x + n − 1) ··· cn (x + n − 1)

c

2001 by CRC Press LLC




If A = an and B = bn , then cn = AB
where c1 (x), . . . , cn (x) are n solutions of the (if all three series converge).

The

Cauchy prod-
homogeneous linear difference equation uct series converges if an and bn converge
n
 and at least one of them converges absolutely
pk (x)y(x + k) = 0 . (Merten’s Theorem).
k=0
Cauchy sequence (1) A sequence of real
casting out nines A method of checking numbers, {rn }, satisfying the following condi-
base-ten multiplications and divisions. See ex- tion. For any B > 0 there exists a positive inte-
cess of nines. ger N such that |rm − rn | < B, for all m, n > N .
(2) A sequence {pn } of points in a metric
casus irreducibilis If the cubic equation space (X, ρ), satisfying the following condition:
ax 3 + bx 2 + cx + d = 0 is irreducible over
the extension Q(a, b, c, d) of the rational num- ρ (pn , pm ) → 0 as n, m → ∞ .
ber field Q, and if all the roots are real, then it
is still impossible to find the roots of this cubic Cauchy sequences are also called fundamen-
equation, by only rational operations with real tal sequences.
radicals, even if the roots of the cubic equation
are real. Cauchy transform The Cauchy transform,

µ̂, of a measure µ, is defined by µ̂(ζ ) = (z −
category A graph equipped with a notion ζ )−1 dµ(z). If K is a compact planar set with
of identity and of composition satisfying certain connected complement and A(K) is the algebra
standard domain and range properties. of complex functions analytic on the interior of
K, then the Cauchy transform is used to show
Cauchy inequality The inequality that every element of A(K) can be uniformly
 n 2 n n
approximated on K by polynomials.
  
ai bi ≤ ai2 bi2 ,
Cayley algebra Let F be a field of charac-
i=1 i=1 i=1
teristic zero and let Q be a quaternion algebra
for real numbers a1 , . . . , an , b1 , . . . , bn . Equal- over F . A general Cayley algebra is a two-
ity holds if and only if ai = cbi , where c is a dimensional Q-module Q + Qe with the mul-
constant. tiplication (x + ye)(z + ue) = (xz + vu y) +
(xu + yz )e, where x, y, z, u ∈ Q, v ∈ F and
Cauchy problem Given an nth order partial z , u are the conjugate quarternions of z and
differential equation (PDE) in z with two inde- u, respectively. A Cayley algebra is the special
pendent variables, x and y, and a curve @ in the case of a general Cayley algebra where Q is the
xy-plane, a Cauchy problem for the PDE con- quaternian field, F is the real number field, and
sists of finding a solution z = φ(x, y) which v = −1.
meets prescribed conditions
∂ j +k z Cayley-Hamilton Theorem See Hamilton-
= fj k Cayley Theorem.
∂x j ∂y k
j + k ≤ n − 1, j, k = 0, 1, . . . , n − 1 on @. Cayley number The elements of a general
Cauchy problems can be defined for systems Cayley algebra. See Cayley algebra.
of partial differential equations and for ordinary
differential equations (then they are called ini- Cayley projective plane Let H be the set of
tial value problems). all 3 × 3 Hermitian matrices M over the Cayley
algebra such that M 2 = M and tr M = 1. The
Cauchy
product
The Cauchy
product of two
set H , with the structure of a projective plane, is
series ∞
n=1 an and
∞
n=1 bn is
∞
n=1 cn where called the Cayley projective plane. See Cayley
cn = a1 bn + a2 bn−1 + · · · + an b1 . algebra.

c

2001 by CRC Press LLC
Cayley’s Theorem Every group is isomor- chain equivalent Let C1 and C2 be chain
phic to a group of permutations. complexes. If there are chain mappings α: C1 →
C2 and β: C2 → C1 such that α ◦ β = 1C2 and
Cayley transformation The mapping be- β ◦ α = 1C1 , then we say that C1 is chain equiv-
tween n × n matrices N and M, given by M = alent to C2 . See chain complex, chain mapping.
(I − N )(I + N )−1 , which acts as its own in-
verse. The Cayley transformation demonstrates chain homotopy Let C1 and C2 be chain
a one-to-one correspondence between the real complexes. Let α, β: C1 → C2 be two chain
alternating matrices N and proper orthogonal mappings, and let R be a ring with identity. If
matrices M with eigenvalues different from −1. there is an R-homomorphism γ : C1 → C2 of
degree 1, such that α−β = γ ◦α  +β  ◦γ , where
CCR algebra A C*-algebra A, which is (C1 , α  ) and (C2 , β  ) are chain complexes over
mapped to the algebra of compact operators un- R. Then γ is called a chain homotopy of α to
der any irreducible ∗-representation. Also called β. See chain complex, chain mapping.
liminal C*-algebra.
chain mapping Let (C1 , α) and (C2 , β) be
center (1) Center of symmetry in Euclidean chain complexes over a ring R with identity. An
geometry. The midpoint of a line, center of a R-homomorphism γ : C1 → C2 of degree 0
triangle, circle, ellipse, regular polygon, sphere, that satisfies β ◦ γ = γ ◦ α is called a chain
ellipsoid, etc. mapping of C1 to C2 . See chain complex.
(2) Center of a group, ring, or Lie algebra X.
The set of all elements of X that commute with chain subcomplex Let R be a ring with iden-
every element of X. tity and
let (C, α) be a chain complex over R. If
(3) Center of a lattice L. The set of all central H = n Hn is a homogeneous R-submodule
elements of L. of C such that α(H ) ⊂ H , then H is called a
chain subcomplex of C. See chain complex.
central extension Let G, H , and K be groups
such that G is an extension of K by H . If H is Chain Theorem Let A, B, and C be alge-
contained in the center of G, then G is called a braic number fields such that C ⊂ B ⊂ A and
central extension of H . let 0A/C , 0A/B , and 0B/C denote the relative
difference of A over C, A over B, and B over C,
centralizer Let X be a group (or a ring) and respectively. Then 0A/C = 0A/B 0B/C . See
let S ⊂ X. The set of all elements of X that different.
commute with every element of S is called the
centralizer of S. chamber In a finite dimensional real affine
space A, any connected component of the com-
central separable algebra An R-algebra plement of a locally finite union of hyperplanes.
which is central and separable. Here a central See locally finite.
R-algebra A which is projective as a two-sided
A-module, where R is a commutative ring. chamber complex A complex with the prop-
erty that every element is contained in a chamber
central simple algebra A simple algebra A and, for two given chambers C, C  , there exists
over a field F , such that the center of A coincides a finite sequence of chambers C = C0 , C1 , . . . ,
with F . (Also called normal simple algebra.) Cr = C  in such a way that codimCk−1 (Ck ∩
Ck−1 ) = codimCk (Ck ∩ Ck−1 ) ≤ 1, for k =
chain complex Let R be a ring with iden- 1, 2, . . . , r. See chamber.
tity and let C be a unitary R-module. By a
chain complex (C,
α) over R we mean a graded character A character X of an Abelian group
R-module C = n Cn together with an R- G is a function that assigns to each element x of
homomorphism α: C → C of degree −1, where G a complex number X (x) of absolute value 1
α ◦ α = 0. such that X (xy) = X (x)X (y) for all x and y in

c

2001 by CRC Press LLC
G. If G is a topological Abelian group, then X with constant coefficients ai (i = n−1, . . . , n−
must be continuous. k) and then divide by λn−k , we obtain

character group The set of all characters λk + an−1 λk−1 + · · · + an−k+1 λ + an−k = 0 ,
of a group G, with addition defined by (X1 +
which is again called the characteristic equation
X2 )(x) = X1 (x)·X2 (x). The character group is
associated with this given difference equation.
Abelian and is sometimes called the dual group
(3) The above two definitions can be extended
of G. See character.
for a system of linear differential (difference)
characteristic Let F be a field with identity equations.
1. If there is a natural number c such that c1 = (4) Moreover, if M = (mij ) is a square ma-
1 + · · · + 1 (c 1s) = 0, then the smallest such c trix of degree n over a field F , then the algebraic
is a prime number p, called the characteristic of equation |λI − M| = 0 is also called the char-
the field F . If there is no natural number c such acteristic equation of M.
that c1 = 0, then we say that the characteristic
characteristic linear system Let S be a non-
of the field F is 0.
singular surface and let A be an irreducible al-
characteristic class (1) Of an R-module gebraic family of positive divisors of dimension
extension 0 → N → X → M → 0, d on S such that a generic member M of A is an
the element 00 (idN ) in the extension module irreducible non-singular curve. Then, the char-
Ext1R (M, N ), where idN is the identity map on acteristic set forms a (d − 1)-dimensional linear
system and contains TrM |M| (the trace of |M| on
N in HomR (N, N ) ∼ = Ext0R (N, N ) and 00 is
M) as a subfamily. This linear system is called
the connecting homomorphism Ext0R (N, N ) →
the characteristic linear system of A.
Ext1R (M, N ) obtained from the extension se-
quence. See connecting homomorphism. characteristic multiplier Let Y (t) be a fun-
(2) Of a vector bundle over base space X, damental matrix for the differential equation
any of a number of constructions of a particu-
lar cohomology class of X, chosen so that the y  = A(t)y . (∗)
bundle induced by a map f : Y → X is the
image of the characteristic class of the bundle Let ω be a period for the matrix A(t). Suppose
over X under the associated cohomological map that H is a constant matrix that satisfies
f ∗ : H ∗ (X) → H ∗ (Y ). See Chern class, Euler
Y (t + ω) = Y (t)H , t ∈ (−∞, ∞) .
class, Pontrjagin class, Stiefel-Whitney class,
Thom class. Then an eigenvalue µ for H of index k and mul-
tiplicity m is called a characteristic multiplier
characteristic equation (1) If we substitute for (∗), or for the periodic matrix A(t), of index
y = eλx in the general nth order linear differen- k and multiplicity m.
tial equation
y (n) (x) + an−1 y (n−1) (x) + . . . characteristic multiplier Let Y (t) be a fun-
damental matrix for the differential equation
+a1 y  (x) + a0 y(x) = 0
y  = A(t)y . (∗)
with constant coefficients ai (i = n − 1, . . . , 0)
and then divide by eλx , we obtain Let ω be a period for the matrix A(t). Suppose
n
λ + an−1 λ n−1
+ · · · + a1 λ + a 0 = 0 , that H is a constant matrix that satisfies

which is called the characteristic equation asso- Y (t + ω) = Y (t)H , t ∈ (−∞, ∞) .

ciated with the given differential equation.
Then an eigenvalue µ for H of index k and mul-
(2) If we substitute yn = λn in the general
tiplicity m is called a characteristic multiplier,
kth order difference equation
of index k and multiplicity m, for (∗), or for the
yn + an−1 yn−1 + · · · + an−k yn−k = 0 periodic matrix A(t).

c

2001 by CRC Press LLC
characteristic of logarithm The integral part Chebotarev Density Theorem Let F be an
of the common logarithm. algebraic number field with a subfield f , F /f
be a Galois extension, C be a conjugate class of
characteristic polynomial The polynomial the Galois group G of F /f , and I (C) be the set
on the left side of a characteristic equation. See of all prime ideals P of k such that the Frobenius
characteristic equation. automorphism of each prime factor Fi of P in
F is in C. Then the density of I (C) is |C|/|G|.
characteristic series Let G be a group. If we
take the group Aut(G) (the group of automor- Chern class The ith Chern class is an ele-
phisms of G) as an operator domain of G, then ment of H 2i (M; R), where M is a complex man-
a composition series is called a characteristic ifold. The Chern class measures certain proper-
series. See composition series. ties of vector bundles over M. It is used in the
Riemann-Roch Theorem.
characteristic set A one-dimensional set of
positive divisors D of a nonsingular curve of
Chevalley complexification Let G be a com-
dimension n so that, with respect to one such
pact Lie group, r(G) the representative ring of
generic divisor D0 of the curve, the degree of
G, A the group of all automorphisms of r(G),
the specialization of the intersection D · D0 over
and G the centralizer of a subgroup of A in A.
the specialization of D over D0 is a divisor of
If G is the closure of G relative to the Zariski
degree equal to that of D · D0 .
topology of G , then G is called the Chevalley
complexification of G.
character module Let G be an algebraic
group, with the sum of two characters X1 and X2
of G defined as (X1 + X2 )(x) = X1 (x) · X2 (x), Chevalley decomposition Let G be an alge-
for all x ∈ G. The set of all characters of G braic group, defined over a field F and Ru the
forms an additive group, called the character unipotent radical of G. If F is of characteristic
module of G. See character of group, algebraic zero, then there exists a reductive, closed sub-
group. group C of G such that G can be written as a
semidirect product of C and Ru . See algebraic
character of a linear representation For the group.
representation ρ : A → GLn (k) of the algebra
A over a field k, the function χρ on A given by Chevalley group Let F be a field, f an ele-
χρ (a) = tr(ρ(a)). ment of F , LF a Lie algebra over F , B a basis of
LF over F and tθ (f ) the linear transformation
character of group A rational homomor- of LF with respect to B, where θ ranges over the
phism α of an algebraic group G into GL(1), root system of LF . Then, the group generated
where GL(1) is a one-dimensional connected by the tθ (f ), for each root θ and each element
algebraic group over the prime field. See alge- f , is called the Chevalley group of type over F .
braic group.
Chevalley’s canonical basis Of a complex,
character system For the quadratic field k, semisimple Lie algebra G with Cartan subalge-
with discriminant d and ideal class group I ∼ = bra H and
root system 8,  a basis for G consisting
F /H (F the group of fractional ideals and H the of a basis H1 , . .
. , Hs of H and, for each root
subgroup of principal ideals
generated by posi- σ ∈ 8, a basis Xσ of its root subspace Gσ
tive elements), a collection χp (N (A)) (p|d) of that satisfy: (i.) σ (Hi ) is an integer for every
numbers, indexed by the prime factors of d, in σ ∈ 8 and each Hi ; (ii.) β(Xσ , X−σ ) = (σ,σ 2
)
which χp is the Legendre symbol mod p and for every σ ∈ 8, where ( , ) represents the inner
A is any representative ideal in its ideal class product on the roots induced by the Killing form
mod H . The character system is independent of β on G; (iii.) if σ , τ , and σ + τ are all roots and
the choice of representative and uniquely deter- [Xσ , Xτ ] = nσ,τ Xσ +τ , then the numbers nσ,τ
mines each class in I . are integers that satisfy n−σ,−τ = −nσ,τ .

c

2001 by CRC Press LLC
Chevalley’s Theorem Let G be a connected difficulties as Russell’s paradox, concerning the
algebraic group, defined over a field F , and let set of all sets that do not belong to themselves.
N be a (F -closed) largest, linear, connected,
closed, normal subgroup of G. If C is a closed, class field Let F be an algebraic number field
normal subgroup of G, then the factor group and E be a Galois extension of F . Then, E
G/C is complete if and only if N ⊂ C. is said to be a class field over F , for the ideal
group I (G), if the following condition is met: a
Cholesky method of factorization A method prime ideal P of F of absolute degree 1 which is
of factoring a positive definite matrix A as a relatively prime to G is decomposed in E as the
product A = LLT where L is a lower triangular product of prime ideals of E of absolute degree
matrix. Then the solution x of Ax = b is found 1 if and only if P is in I (G).
by solving Ly = b, LT x = y.
class field theory A theory created by E.
Choquet boundary Let X be a compact Artin and others to determine whether certain
Hausdorff space and let A be a function algebra primes are represented by the principal form.
on X. The Choquet boundary is c(A) = {x ∈
X : the evaluation at x has a unique representing class field tower problem Let F be a given
measure}. algebraic number field, and let F = F0 ⊂ F1 ⊂
F2 ⊂ · · · be a sequence of fields such that Fn
Chow coordinates Of a projective variety is the absolute class field over Fn−1 , and F∞ is
X, the coefficients of the associated form of the the union of all Fn . Now we ask, is F∞ a finite
variety, viewed as homogeneous coordinates of extension of F ? The answer is positive if and
points on X. See associated form. only if Fk is of class number 1 for some k. See
absolute class field.
Chow ring Of a nonsingular, irreducible,
class formation An axiomatic structure for
projective variety X, the graded ring whose ob-
class field theory, developed by Artin and Tate.
jects are rational equivalence classes of cycles
A class formation consists of
on X, with addition given by addition of cycles
(1) a group G, the Galois group

of the for-
and multiplication induced by the diagonal map
mation, together with a family GK : K ∈ Z
0 : X → X × X. The ring is graded by codi-
of subgroups of G indexed by a collection 8 of
mension of cycles.
fields K so that
Chow variety Let V be a projective variety. (i.) each GK has finite index in G;
The set of Chow coordinates of positive cycles
that are contained in V is a projective variety (ii.) if H is a subgroup of G containing some
called a Chow variety. GK , then H = GK  for some K  ;

circulant See cyclic determinant. (iii.) the family {GK } is closed under intersec-
tion and conjugation;
circular units The collection of units of the 
s (iv.) 8 GK is the trivial subgroup of G;
form 1−ζ
1−ζ t , where ζ is a p n th root of unity, p is
a prime, and s  ≡ t (modp) (and p  |s, t). (2) a G-module A, the formation module,
such that A is the union of its submodules A(GK )
class (1) (Algebra.) A synonym of set that that are fixed by GK ;
is used when the members are closely related, (3) cohomology groups H r (L/K), defined
like an equivalence class or the class of residues as H r (GK /GL , A(GK ) ), for which H 1 (L/K)
modulo m. = 0 whenever GL is normal in GK ;
(2) (Logic.) A generalization of set, includ- (4) for each field K, there is an isomorphism
ing objects that are “too big” to be sets. Con- A  → invK A of the Brauer group H 2 (∗ /K)
sideration of classes allows one to avoid such into Q/Z such that

c

2001 by CRC Press LLC
 (i.) if GL is normal in GK of index n, clearing of fractions An equation is cleared
 of fractions if both sides are multiplied by a com-
1 mon denominator of all fractions appearing in
invK H 2 (L/K) = Z /Z
n the equation.
and Clebsch-Gordon coefficient One of the co-
efficients, denoted
(ii.) even when GL is not normal in GK ,   
j1 m1 j2 m2 j1 j2 j m
invL ◦ resK,L = n invK
in the formula
where resK,L is the natural restriction map  
H 2 (∗/K) → H 2 (∗/E). ψ(j m) = (j1 m1 j2 m2 j1 j2 j m)
−j ≤m1 ,m2 ≤j
× ψ(j1 m1 )ψ(j2 m2 )
classical compact real simple Lie algebra
A compact real simple Lie algebra of the type which relates the basis elements of the represen-
An , Bn , Cn , or Dn , where An , Bn , Cn , and Dn tation space C2 ⊗ · · · ⊗ C2 of 2j copies of C2
are the Lie algebras of the compact Lie groups for a representation of SO(3) ∼ = SU(2)/ {±I }.
SU(n + 1), SO(2n + 1), Sp(n), and SO(2n), The coefficients are determined by the formula
respectively.   
j1 m1 j2 m2 j1 j2 j m = δm1 +m2 ,m
classical compact simple Lie group Any of

the connected compact Lie groups SU(n + 1), (2j + 1)(j1 + j2 − j )!(j + j1 − j2 )!(j + j2 − j1 )!
×
SO(nl +1), Sp(n), or SO(2n), with correspond- (j1 + j2 + j + 1)!

ing compact real simple Lie algebra An (n ≥ 1), 


(j1 + m1 )!(j1 − m1 )!(j2 + m2 )!
Bn (n ≥ 2), Cn (n ≥ 3), or Dn (n ≥ 4) as its × (−1)ν
ν ν!(j1 + j2 − j − ν)!(j1 − m1 − ν)!
Lie algebra. 
(j2 − m2 )!(j + m)!(j − m)!
× .
classical complex simple Lie algebra Let (j2 + m2 − ν)!(j − j2 + m1 + ν)!(j − j1 − m2 + ν)!

An , Bn , Cn , and Dn be the Lie algebras of the

complex Lie groups SL(n + 1, C), SO(2n +
1, C), Sp(n, C), and SO(2n, C). Then An (n ≥ Clifford algebra Let L be an n-dimensional
1), Bn (n ≥ 2), Cn (n ≥ 3), and Dn (n ≥ 4) are linear space over a field F , Q a quadratic form on
called classical complex simple Lie algebras. L, A(L) the tensor algebra over L, I (Q) the two-
sided ideal on A(L) generated by the elements
classical group Groups such as the general l ⊗ l − Q(l) · 1 (l ∈ L), where ⊗ denotes tensor
linear groups, orthogonal groups, symplectic multiplication. The quotient associative algebra
groups, and unitary groups. A(L)/I (Q) is called the Clifford algebra over
Q.
classification Let R be an equivalence rela-
tion on a set S. The partition of S into disjoint Clifford group Let L be an n-dimensional
union of equivalence classes is called the clas- linear space over a field F , Q a quadratic form
sification of S with respect to R. on L, C(Q) the Clifford algebra over Q, G the
set of all invertible elements g in C(Q) such that
class number The order of the ideal class gLg −1 = L. Then, G is a group with respect
group of an algebraic number field F . Similarly, to the multiplication of C(Q) and is called the
the order of the ideal class group of a Dedekind Clifford group of the Quadratic form Q. See
domain D is called the class number of D. Clifford algebra.

class of curve The degree of the tangential Clifford numbers The elements of the Clif-
equation of a curve. ford algebra. See Clifford algebra.

c

2001 by CRC Press LLC
closed boundary If X is a compact Haus- (k), M) is bijective for any algebraically closed
dorff space, then a closed boundary is a bound- field k, and
ary closed in X. (ii.) for any scheme N and any natural transfor-
mation ψ : M → Hom(−, N ) there is a unique
closed image Let µ:V → V  be a morphism natural transformation λ : Hom(−, M) →
of varieties. If V is not complete, then µ(V ) Hom(−, N ) such that ψ = λ ◦ ϕ.
may not be closed. The closure of µ(V ) which
is in V  , is called the closed image of V . coarser classification For R, S ⊂ X×X two
equivalence relations on X, S is coarser than R
closed subalgebra A subalgebra B1 of a Ba- if R ⊂ S.
nach algebra B that is closed in the norm topol-
ogy. B1 is then a Banach algebra, with respect coboundary See cochain complex.
to the original algebraic operations and norm of
B. coboundary operator See cochain complex.

closed subgroup A subgroup H of a group cochain See cochain complex.

G such that x ∈ H , whenever some nontrivial
power of x lies in H . cochain complex A cochain complex C =
{C i , δ i : i ∈ Z} is a sequence of modules {C i :
closed subsystem Let M be a character mod- i ∈ Z}, together with, for each i, a module
ule. Let s be a subset of a root system r, and let homomorphism δ i : C i → C i+1 such that δ i−1 ◦
s  be the submodule of M generated by s. If δ i = 0. Diagramatically, we have
s  ∩ r = s, then s is called a closed subsystem.
δ i−2 δ i−1 δi δ i+1
See character module. . . . −→ C i−1 −→ C i −→ C i+1 −→ . . . ,

closed under operation A set S, with a bi- where the composition of any two successive
nary operation ∗ such that a ∗ b ∈ S for all δ j is zero. In this context, the elements of C i
a, b ∈ S. are called i-cochains, the elements of the ker-
nel of δ i , i-cocycles, the elements of the im-
closure property The property of a set of be- age of δ i−1 , i-coboundaries, the mapping δ i ,
ing closed under a binary operation. See closed the ith coboundary operator, the factor module
under operation. H i (C) = ker δ i /im δ i−1 , the ith cohomology
module and the set H (C) = {H i (C) : i ∈ Z},
coalgebra Let ρ and ρ  be linear mappings the cohomology module of C. If c is an i-cocycle
defined as ρ : V → V ⊗F V , and ρ  : V → (i.e., an element of ker δ i ), the corresponding el-
F , where F is a field, V is vector space over ement c + im δ i−1 of H i (C) is called the coho-
F , and ⊗ denotes tensor product. Then, the mology class of c. Two i-cocycles belong to the
triple (V , ρ, ρ  ) is said to be a coalgebra over same cohomology class if and only if they dif-
F , provided (1V ⊗ ρ) ◦ ρ = (ρ ⊗ 1V ) ◦ ρ and fer by an i-coboundary; such cocycles are called
(1V ⊗ ρ  ) ◦ ρ = (ρ  ⊗ 1V ) ◦ ρ = 1V . cohomologous.

coalgebra homomorphism A k-linear map cochain equivalence Two cochain com-

f : C → C  , where (C, 0, ε) and (C  , 0 , ε ) plexes C and C̃ are said to be equivalent if
are coalgebras over a field k, such that (f ⊗ f ) ◦ there exist cochain mappings φ : C → C̃ and
0 = 0 ◦ f and ε = ε  ◦ f . ψ : C̃ → C such that φψ and ψφ are homo-
topic to the identity mappings on C and C̃, re-
coarse moduli scheme A scheme M and a spectively.
natural transformation ϕ : M → Hom(−, M)
where M is a contravariant functor from cochain homotopy Let C = {C i , δ i : i ∈ Z}
schemes to sets such that and C̃ = {C̃ i , B i : i ∈ Z} be two cochain com-
(i.) ϕ(Spec(k)) : M(Spec(k)) → Hom (Spec plexes and let φ, ψ : C → C̃ be two cochain

c

2001 by CRC Press LLC
mappings. A homotopy ζ : φ → ψ is a se- coefficient of equation A number or constant
quence of mappings {ζ i : i ∈ Z} such that, for appearing in an equation. (For example, in the
each i, ζ i is a homomorphism from C i to C̃ i and equation 2 tan x = 3x + 4, the coefficients are
2, 3, and 4.)
φ i − ψ i = δ i ζ i+1 + ζ i B i−1 .
coefficient of linear representation When a
When such a homotopy exists, φ and ψ are said representation χ (of a group or ring) is isomor-
to be homotopic and H (φ) = H (ψ), where phic to a direct sum of linear representations and
H (φ) and H (ψ) are the morphisms from the other irreducible representations, the number of
cohomology module H (C) to the cohomology times that a particular linear or irreducible rep-
module H (C̃) induced by φ and ψ. resentation φ occurs in the direct sum is called
the coefficient of φ in χ .
cochain mapping Let C = {C i , δ i : i ∈
i i
Z} and C̃ = {C̃ , B : i ∈ Z} be two cochain coefficient of polynomial term In a polyno-
complexes. A cochain mapping φ : C → C̃ is mial
a sequence of mappings {φ i : i ∈ Z} such that,
for each i, φ i is a homomorphism from C i to C̃ i a 0 + a1 x + a 2 x 2 + . . . + a n x n ,
and
φ i B i = δ i φ i+1 . the constants a0 , a1 , . . . , an are called the coef-
The mapping φ i induces a homomorphism from ficients of the polynomial. More specifically, a0
the ith cohomology module H i (C) to the ith is called the constant term, a1 the coefficient of
cohomology module H i (C̃). Hence φ can be x, a2 the coefficient of x 2 , . . . , an the coefficient
regarded as inducing a morphism φ ∗ from the of x n ; an is also called the leading coefficient.
cohomology module H (C) to the cohomology
module H (C̃). The morphism φ ∗ is also de- coefficient ring Consider the set of numbers
noted H (φ). This enables H (·) to be regarded as or constants that are being used as coefficients in
a functor from the category of chain complexes some algebraic expressions. If that set happens
to a category of graded modules; it is called the to be a ring (such as the set of integers), it is
cohomological functor. called the coefficient ring. Likewise, if that set
happens to be a field (such as the set of real
cocommutative algebra A coalgebra C, in numbers) or a module, it is called the coefficient
which the comultiplication 0 is cocommutative, field or coefficient module, respectively.
i.e., has the property that δτ = δ, where τ is the
flip mapping, i.e., the mapping from C ⊗ C to cofactor Let A be an n×n matrix. The (k, P)
itself that interchanges the two C factors. cofactor of A written AkP is (−1)k+P times the
determinant of the n − 1 × n − 1 matrix obtained
cocycle See cochain complex. by deleting the kth row and j th column of A.

codimension Complementary to the dimen- cofunction The trigonmetric function that is

sion. For example, if X is a subspace of a vector the function of the complementary angle. For
space V and V is the direct sum of subspaces X example, cotan is the cofunction of tan.
and X  , then the dimension of X  is the codi-
mension of X. cogenerator An element A of category C
such that the functor Hom(−, A) : C → A is
coefficient A number or constant appearing faithful where A is the category of sets. Also
in an algebraic expression. (For example, in called a coseparator.
3 + 4x + 5x 2 , the coefficients are 3, 4, and 5.)
Cohen’s Theorem A ring is Noetherian if
coefficient field See coefficient ring. and only if every prime ideal has a finite basis.
There are several other theorems which may be
coefficient module See coefficient ring. called Cohen’s Theorem.

c

2001 by CRC Press LLC
coherent algebraic sheaf A sheaf F of OV - coideal Let C be a coalgebra with comulti-
modules on an algebraic variety V , such that if, plication 0 and counit B. A subspace I of C is
for every x ∈ V , there is an open neighborhood called a coideal if 0(I ) is contained in
U of x and positive integers p and q such that
the OV |U -sequence I ⊗C+C⊗I

p q
OV |U → OV |U → F|U → 0 and B(I ) is zero.

is exact. coimage Let φ : M → N be a homomor-

phism between two modules M and N. Then
coherent sheaf of rings A sheaf of rings A the factor module M/ ker φ (where ker φ de-
on a topological space X that is coherent as a notes the kernel of φ) is called the coimage of
sheaf of A-modules. φ, denoted coim φ. By the First Isomorphism
Theorem, the coimage of φ is isomorphic to the
cohomological dimension For a group G, image of φ, as a consequence of which the term
the number m, such that the cohomology group coimage is not often used.
H i (G, A) is zero for every i > m and every G-
cokernel Let φ : M → N be a homomor-
module A, and H m (G, A) is non-zero for some
phism between two modules M and N . Then the
G-module A.
factor module N/ im φ (where im φ denotes the
image of φ) is called the cokernel of φ, denoted
cohomology The name given to the subject
coker φ.
area that comprises cohomology modules, co-
homology groups, and related topics.
collecting terms The name given to the proc-
ess of rearranging an expression so as to com-
cohomology class See cochain complex.
bine or group together terms of a similar nature.
For example, if we rewrite
cohomology functor See cochain mapping.
x 2 + 2x + 1 + 3x 2 + 5x as 4x 2 + 7x + 1
cohomology group (1) Because any Z-
module is an Abelian group, a cohomology we have collected together the x 2 terms and the
module H (C), where C is a cochain complex, is x terms; if we rewrite
called a cohomology group in the case where all    
the modules under consideration are Z-modules. x 2 +y 2 +2x−2y as x 2 + 2x + y 2 − 2y
(2) Let G be a group and A a G-module. Then
H i (G, A), the ith cohomology group of G with we have collected together all the terms involv-
coefficients in A, is defined by ing x, and all the terms involving y.

H i (G, A) = ExtiG (Z, A) . color point group A pair of groups (K, K1 )

such that K = G/T , G is a space group, T is
(When interpreting the Ext functor in this con- the group of translations, and for some positive
text, Z should be interpreted as a trivial G- integers r, K1 is the group of conjugacy classes
module.) of all subgroups G1 of G with T ⊂ G1 and
[G : G1 ] = r.
cohomology module See cochain complex.
color symmetry group A pair of groups (G,
cohomology set The set of cohomology G ) where G is a space group and [G : G ] < ∞.
classes, when these classes do not possess a
group structure. column finite matrix An infinite matrix with
an infinite number of rows and columns, such
cohomology spectral sequence See spectral that no column has any non-zero entry beyond
sequence. the nth entry, for some finite n.

c

2001 by CRC Press LLC
column in matrix See matrix. Here (x −1)(x +1) is the common denominator.

column nullity For an m × n matrix A, the common divisor A number that divides all
number m − r(A), where r(A) is the rank of A. the numbers in a list. For example, 3 is a com-
mon divisor of 6, 9, and 12. See also greatest
column vector A matrix with only one col- common divisor.
umn, i.e., a matrix of the form
 
a1 common fraction A quotient of the form a/b
 a2  where a, b are integers and b  = 0.
 
 ..  .
 . 
common logarithm Logarithm to the base
am
10 (the logarithm that was most often used in
Also called column matrix. arithmetic calculations before electronic calcu-
lators were invented). See also logarithm.
combination of things When r objects are
selected from a collection of n objects, the r common multiple A number that is a multi-
selected objects are called a combination of r ple of all the numbers in a list. For example, 12
objects from the collection of n objects, pro- is a common multiple of 3, 4, and 6. See also
vided that the selected objects are all regarded least common multiple.
as having equal status and not as being in any
particular order. (If the r objects are put into a commutant If S is a subset of a ring R, the
particular order, they are then called a permu- commutant of S is the set S  = {a ∈ R : ax =
tation, not a combination, of r objects from n.) xa for all x ∈ S}. Also called: commutor.
The number of different combinations ofr ob-
n n
jects from n is denoted by Cr or and commutative algebra (1) The name given to
r
the subject area that considers rings and modules
n!
equals . in which multiplication obeys the commutative
r!(n − r)! law, i.e., xy = yx for all elements x, y.
commensurable Two non-zero numbers a (2) An algebra in which multiplication obeys
and b such that a = nb,√for some rational
√ num- the commutative law. See algebra.
ber n. For example, 2√ 2 and 3√2 are com-
mensurable because 2 2 = 23 (3 2). All ra- commutative field A field in which multi-
tional numbers are commensurable with each plication is commutative, i.e., xy = yx, for all
other. No irrational number is commensurable elements x, y. (The axioms for a field require
with any rational number. that addition is always commutative. Some ver-
sions of the axioms insist that multiplication has
common denominator When two or more to be commutative too. When these versions of
fractions are about to be added, it is helpful to re- the axioms are in use, a commutative field be-
express them first, so that they have the same de- comes simply a field, and the term division ring
nominator, called a common denominator. This is used for structures that obey all the field ax-
process is also described as putting the fractions ioms except commutativity of multiplication.)
over a common denominator. For example, to See also field.
3
simplify x−1 − x+1
2
we might write
commutative group See Abelian group.
3 2
−
x−1 x+1
commutative law The requirement that a bi-
3(x + 1) 2(x − 1) nary operation ∗, on a set X, satisfy x ∗y = y ∗x,
= −
(x − 1)(x + 1) (x − 1)(x + 1) for all x, y ∈ X. Addition and multiplication of
x+5 real numbers both obey the commutative law;
= . matrix multiplication does not.
(x − 1)(x + 1)

c

2001 by CRC Press LLC
commutative ring A ring in which multi- compact simple Lie group A Lie group that
plication is commutative, i.e., xy = yx, for all is compact as a topological space and whose Lie
elements x, y. (The axioms for a ring ensure algebra is simple (i.e., not Abelian and having
that addition is always commutative.) The real no proper invariant subalgebra).
numbers are a commutative ring. Rings of ma-
trices are generally not commutative. compact topological space A topological
space X with the following property: whenever
commutator (1) An element of the form x −1 O = {Oα }α∈A is an open covering of X, then
y −1 xy or xyx −1 y −1 in a group. Such an el- there exists a finite subcovering O1 , O2 , . . . , Ok .
ement is usually denoted [x, y] and it has the
property that it equals the identity element if companion matrix Given the monic poly-
and only if xy = yx, i.e., if and only if x and y nomial over the complex field
commute.
(2) An element of the form xy − yx in a ring. p(t) = t n + an−1 t n−1 + · · · + a1 t + a0 ,
Such an element is denoted [x.y] and is also
called the Lie product of x and y. It equals zero the n × n matrix
if and only if xy = yx.  
0 0 ... 0 −a0
In a ring with an involution ∗ , the element 1 0 ... 0 −a1 
x x − xx ∗ is often called the self-commutator
∗  
 .. .. .. 
of x. A= 0 1 . . . 

 .. . . .. 
. . . 0 −an−2 
commutator group See commutator sub-
group. 0 ... 0 1 −an−1

is called the companion matrix of p(t). The

commutator subgroup The subgroup G of
characteristic polynomial and the minimal poly-
a given group G, generated by all the commu-
nomial of A are known to coincide with p(t). In
tators of G, i.e., by all the elements of the form
fact, a matrix is similar to the companion ma-
x −1 y −1 xy (where x, y ∈ G). G consists pre-
trix of its characteristic polynomial if and only if
cisely of those elements of G that are expressible
the minimal and the characteristic polynomials
as a product of a finite number of commutators.
coincide.
So G may contain elements that are not them-
selves commutators. G is a characteristic sub-
complementary degree Let F be a filtra-
group of G. G/G is Abelian and in fact G is
tion of a differential Z-graded module A. Then
the unique smallest normal subgroup of G with
{Fp An } is a Z-bigraded module and the module
the property that the factor group of G by it is
Fp An has complementary degree q = n − p.
Abelian. G is also called the commutator group
or the derived group of G. complementary law of reciprocity A reci-
procity law due to Hasse and superseded by
commutor See commutator.
Artin’s general law of reciprocity. Let p be
a prime number. If α ∈ k is such that α ≡
compact group A topological group that is
1 mod p(1 − τp ) where τp = exp 2π i/p ∈ k,
compact, as a topological space. A topologi-
then p 
cal group is a group G, with the structure of a c
topological space, such that the map d p = τp ,
 
(x, y)  → xy −1 from G × G to G c = Trk/Q αp(1−τp ) ;

is continuous. 1 − τp
= τpd ,
α p
compact real Lie algebra A real Lie alge- 
bra whose Lie group is a compact group. See α−1
d = −Trk/Q .
compact group. p(1 − τp )

c

2001 by CRC Press LLC
complementary series The irreducible, uni- complete intersection A variety V in P n (k)
tarizable, non-unitary, principal series represen- of dimension r, where I (V ) is generated by n−r
tations of a reductive group G. homogeneous polynomials. See variety.

complementary slackness This refers to complete linear system The set of all effec-
Tucker’s Theorem on Complementary Slackness, tive divisors linearly equivalent to a divisor.
which asserts that, for any real matrix A, the in-
equalities Ax = 0, x ≥ 0 and t uA ≥ 0 have complete local ring See local ring.
solutions x, u satisfying At u + x > 0.
completely positive mapping For von Neu-
mann algebras A and B, a linear mapping T :
complementary submodule Let N be a sub- A → B such that for all n ≥ 1 the induced
module of a module M. Then a complementary mappings
submodule to N in M is a submodule N  of M
such that M = N ⊕ N  . In ⊗ T : Mn (A) = Mn (C) ⊗ A → Mn (B)

complementary trigonometric functions are positive.

The functions cosine, cosecant, and cotangent
(cos θ , cscθ and cot θ ), so called because the completely reduced module An R module
cosine, cosecant, and cotangent of an acute an- (R a ring) which is the direct sum of irreducible
gle θ equal, respectively, the sine, secant, and R modules. An R module is irreducible if it has
tangent of the complementary angle to θ , i.e., no sub R-modules.
the acute angle φ such that θ and φ form two of
the angles in a right-angled triangle. (φ = π2 −θ completely reducible Let k be a commuta-
in radians or 90 − θ in degrees.) tive ring and E a module over k. Let R be a
k-algebra and let ϕ : R → Endk (E) be a repre-
sentation of R in E. We say that ϕ is completely
complete cohomology theory A cohomol-
reducible (or semi-simple) if E is an R-direct
ogy theory of the following form. Let π be a
sum of R-submodules Ei ,
finite multiplicative group and B = B(Z(π ))
the bar resolution. If A is a π -module define E = E1 ⊕ · · · ⊕ Em ,
H ∗ (π, A) = H ∗ (B, A). See cohomology.
with each Ei irreducible.
complete field A field F with the following
property: whenever p(x) = a0 +a1 x +· · · ak x k completely reducible representation A rep-
is a polynomial with coefficients in F then p has resentation σ such that the relevant R-module E
a root in F . is an R-direct sum of R-submodules Ei ,

E = E1 ⊕ E2 ⊕ · · · ⊕ Em ,
complete group A group G whose center
is trivial and all its automorphisms are inner.
such that each Ei is irreducible. See irreducible
Thus, G and the automorphism group of G are
representation.
canonically isomorphic.
Also called semi-simple.

complete integral closure Let O be an in- completely solvable group A group that is
tegral domain in a field K and M an O-module the direct product of simple groups.
contained in K. Let S be the set of all val-
uations on K that are nonnegative on O. If complete pivoting A process of solving an
υ ∈ S,let Rυ be the valuation ring of υ. Then n × n linear system of equations Ax = b. By
M  = v∈S Rn M is the completion of M. If O a succession of row and column operations, one
is the integral closure of O, M  is the complete may solve this equation once A has been trans-
integral closure of the O-module M  = OM. formed to an upper triangular matrix. Suppose

c

2001 by CRC Press LLC