Introduction to complex K theory

Junting Huang
May, 2010

Abstract
This paper gives an introduction on complex topological K theory.
It starts with basic definitions and constructions, and ends with Bott
periodicity and cohomology thoery.

1 Basics
1.1 Vector bundles
An n dimensional vector bundle is a map p : E → B (usually written as
{E, p, B}) together with a real vector space structure on p−1 (b) for each b ∈ B,
such that the following local triviality condition is satisfied: There is a cover
of B by open sets Uα for each of which there exists a homeomorphism hα :
p−1 Uα → Uα × Rn taking p−1 b to b × Rn by a vector space isomorphism for
each b ∈ Uα . Such an hα is called a local trivialization of the vector bundle.
The space B is called the base space, E is the total space, and the vector spaces
p−1 b are the fibers.
A section of a vector bundle p : E → B is a map s : B → E assigning to
each b ∈ B a vector s(b) in the fiber p−1 (b).
Given two vector bundles p1 : E1 → B and p2 : E2 → B over the same base
space B, one can define the direct sum of E1 and E2 as the space

E1 ⊕ E2 = {(v1 , v2 ) ∈ E1 × E2 |p1 (v1 ) = p2 (v2 )}. (1)

There is then a projection E1 × E2 → B sending (v1 , v2 ) to the point p1 (v1 ) =

p2 (v2 ). The fibers of this projection are the direct sums of the fibers of E1 and
E2 . the tensor product, E1 ⊗ E2 , as a set, can be defined as the disjoint union
of the vector spaces p−1 −1
1 (x) ⊗ p2 (x) for x ∈ B.
Given two spaces X and Y with a map f : X → Y and a vector bundle
E → Y we can form a pullback bundle f ∗ (E) using the definition f ∗ (E) =
{(x, v) : x ∈ X, v ∈ E, f (x) = p(v)}, i.e. the fibre over the point x in X is the
fibre over the point f (x) in Y .

1
1.2 clutching theorem
Theorem 1. Let (Ui ) be an open cover of a space X. Let ξi = {Ei , πi , Ui } be
a vector bundle over each Ui , and let gij : ξi |Ui ∩Uj → ξj |Ui ∩Uj be isomorphisms
0 0 0
which satisfy the compatibility condition gki |Ui ∩Uj ∩Uk = gkj · gji where gkj =
0
gkj |Ui ∩Uj ∩Uk and gji = gji |Ui ∩Uj ∩Uk . Then there exists a vector bundle ξ over
X and isomorphisms gi : ξi → ξ|Ui , such that the diagram

is commutative.
For the proof, please see Ref.[2]. As an example, we can consider constructing
vector bundles E → S k with base space a sphere. Write S k as the union of its
k k k k
upper and lower hemispheres D+ and D− , with D+ ∩ D− = S k−1 . Given a
map f : S k−1 → GLn (R), let Ef be the quotient of the disjoint union D+ k
×
Rn t D− k
× Rn obtained by identifying (x, v) ∈ ∂D− k
× Rn with (x, f (x)(v)) ∈
k
∂D+ × Rn . There is then a natural projection Ef → S k and this is an n
dimensional vector bundle. The map f is called a clutching function for Ef .
The same construction works equally well with C in place of R, so from a map
f : S k−1 → GLn (C) one obtains a complex vector bundle Ef → S k .

1.3 Others
The wedge product X ∨ Y of two topological spaces X and Y is the space
formed by taking the disjoint union of X and Y and identifying one point on X
with a point on Y : X ∨ Y = X t Y /(x0 ∼ y0 ).
The smash product X ∧ Y of two topological spaces X and Y is the space
formed by taking the product of X and Y and then quotienting by the wedge
product of X and Y , X ∧ Y = X × Y /X ∨ Y = X × Y /(x0 × Y ) t (X × y0 ).
For example, S 1 ∧ S 1 is homeomorphic to S 2 . S 1 ∨ S 1 is an inner circle of torus
with a circle around torus. These two intersect at one point. Identifying the
inner circle to a point, X × Y is a circle rotated about a point, and then take a
circle around the torus with zero inner radius, S 1 × S 1 becomes homeomorphic
to S 2 . In fact S m ∧ S n is homeomorphic to S m+n , for all m,n.
The cone CX over X is the space formed by taking the direct product
of X and the interval I = [0, 1] and collapsing one end to a point: CX =
X × I/(X × {1}). Note that CX is contractible to the vertex point.
The suspension SX of a space X is the space formed by taking the union
of two copies of the cone over X, or equivalently the space formed by attach-
ing I both ’above’ and ’below’ X and then collapsing two ends to two points
respectively, which can also be written as SX = X × I/(X × {0}) t (X × {1}).
The reduced suspension ΣX is the suspension of X quotiented by {x0 }×I
for some x0 , ΣX = X × I/(X × {0} t X × {1} t {x0 } × I). The reduced

2
suspension of a wedge product is the wedge product of the reduced suspensions
of the two spaces involved: Σ(X ∧ Y ) = ΣX ∧ ΣY . This follows as (X ∧ Y ) × I =
X ×I ∧Y ×I/(x0 ×I ∼ y0 ×I) and then quotienting by (X ∧Y ×{0})t(X ∧Y ×
{1})t((x0 , y0 ) × I) gives Σ(X ∧Y ) = ΣX ∧ΣY . Similarly, S(X ∧Y ) = SX ∧SY .
An important property to be used in this paper is ΣX = X ∧ S 1 . To see this
we view S 1 as an interval with the endpoints identified, S 1 = I/({0} ∼ {1}),
and write X ∧ S 1 = X ∧ I/X × {0} ∼ X × {1} = X × I/(X × {0} ∼ X × {1}) t
({x0 } × I t X × {0}) = X × I/(X × {0}) t (X × {1}) t ({x0 } × I) = ΣX where
x0 was the point on X.
n-fold suspension and reduced suspension can be defined as S n X =
· · S} X and Σn X = |Σ ·{z
|S ·{z · · Σ} X, respectively.
n n

2 Grothendieck group
Let S be an abelian semigroup. The Grothendieck group of S is K(S) =
S × S/ ∼ , where ∼ is the equivalence relation: (s, t) ∼ (u, v) if there exists
r ∈ S such that s + v + r = t + u + r . This is indeed an abelian group
with zero element (s, s) (∀s ∈ S), inverse −(s, t) = (t, s) and addition given by
(s, t) + (u, v) = (s + u, t + v). The Grothendieck group construction is a functor
from the category of abelian semigroups to the category of abelian groups. A
morphism f : S → T induces a morphism K(f ) : K(S) → K(T ) which sends
an element (s+ , s− ) ∈ K(S) to (f (s+ ), f (s− )) ∈ K(T ).

Example 1. Let (N, +) be the semigroup of natural numbers with composition

given by addition. Then, K(N, +) = Z. For instance, 5, 6, 7, 8 ∈ N and 5 −
6, 8 − 7 ∈ Z, (5, 6) ∼ (8, 7) since n + (5 + 7) = n + (6 + 8) for any n ∈ N.
Example 2. Let (Z \ {0}, ×) be the semigroup of non-zero integers with com-
position given by multiplication. Then, K(Z \ {0}, ×) = (Q \ {0}, ×). For
instance, 5, 6, 10, 12 ∈ (Z \ {0}, ×), 5/6, 10/12 ∈ (Q \ {0}, ×), (5, 6) ∼ (10, 12)
since a × 5 × 12 = a × 6 × 10 for any a ∈ (Z \ {0}, ×).
Example 3. Let G be an abelian group, then K(G) = ∼ G via the identification
(g, h) ↔ g − h (or (g, h) ↔ gh−1 if G is multiplicative).

3 Group K(X) and K(X)

e
We assume that all the bases of vector bundles are compact Haudsdoff.
Starting from the Grothendieck group construction, let (E, E 0 ) ≡ E − E 0 be
the difference of two vector bundles E and E 0 . We can define an equivalence
relation E1 − E10 ∼ E2 − E20 if E1 ⊕ E20 ⊕ n ' E2 ⊕ E10 ⊕ n where n is any trivial
bundle of X. Then the equivalence class E − E 0 form a group K(X) called K-
group, for which the addition is (E1 − E10 ) + (E2 − E20 ) = (E1 ⊕ E2 ) − (E10 ⊕ E20 ),
the identity is E − E for any E and the inverse for E − E 0 is −(E − E 0 ) = E 0 − E
since (E − E 0 ) + (E 0 − E) = (E ⊕ E 0 ) − (E ⊕ E 0 ) = E 00 − E 00 .

3
 To make Sthis clear, we can make an analogy with Example 1 above. Let
V ect(X) = k V ectk where V ectk is an isomorphism class of rank k vector
bundle. The set N corresponds to the collection of vector bundles V ect(X). +
of N corresponds to direct sum ⊕ of vector bundles. Z corresponds to K(X).
(a, b) = a − b for a, b ∈ N corresponds to (E1 , E2 ) = E1 − E2 ∈ K(X) for
E1 , E2 ∈ K(X). The inverse in Z is to add a minus sign and so is it for K(X).
Take x0 ∈ X, then K(x0 ) = Z since it is just a trivial bundle. Considering
the inclusion i : x0 → X and its pullback i∗ : K(x0 ) → K(X), the reduced
K-group K(K)
e e = ker(i∗ ). This means that for E1 − E2 ∈
can be defined as K
K(Z), rank(E1 − E2 ) ≡ rank(E1 ) − rank(E2 ) = 0, i.e. any element of K(X)
e e
rank(E)
can can be written in the form E −  . By the definition, we have a short
exact sequence
i∗
0 → K(X)
e → K(X) → Z → 0.
Since we can find a map p : Z → K(X) taking Z to trivial bundles in K(X)
such that p ◦ i∗ = idZ , by the splitting lemma,
Lemma 1. Given a short exact sequence with maps q and r:
q r
0→A→B→C→0

one writes the additional arrows t and u for maps that may not exist:
q r
0 → A −→ −→
←− B ←− C → 0
t u

Then the following are equivalent:

1. left split: there exists a map t : B → A such that tq is the identity on A,
2. right split: there exists a map u : C → B such that ru is the identity on
C,
3. direct sum: B is isomorphic to the direct sum of A and C, with q being
the natural injection of A and r being the natural projection onto C.
we have a splitting K(X) = Z ⊕ K(X).
e Additionally, the following is also a
short exact sequence
p q
0 → Z → K(X) → K(X)
e → 0,

where q : E 7→ E − rankE for E ∈ K(X). To see the kernel of q, let q(E) =

E − rank(E) = 0, E = rank(E) , i.e. ker(q) is trivial bundles in K(X), which is
also the image of p.
K(X) has a ring structure if we define the multiplication as ⊗, i.e. the tensor
product of two vector bundles. The identity is 1 , since it is just a trivial line
bundle. So for arbitrary elements of K(X) represented by differences of vector
bundles, their product in K(X) is defined by the formula

(E1 − E10 ) ⊗ (E2 − E20 ) = E1 ⊗ E2 − E1 ⊗ E20 − E10 ⊗ E2 + E10 ⊗ E20 (2)

4
If we denote n as n, then nE = n ⊗ E = ( n 1 ) ⊗ E = n E, since multi-
L L
plication is distributive over addition.
A map f : X −→ Y induces a map f ∗ : K(Y ) −→ K(X), sending E1 − E2 to
f (E1 − E2 ) = f ∗ (E1 ) − f ∗ (E2 ). This is a ring homomorphism, i.e. for E, E 0 ∈
∗

K(X), f ∗ (E + E 0 ) −→ f ∗ (E) + f ∗ (E 0 ) and f ∗ (E ⊗ E 0 ) −→ f ∗ (E) ⊗ f ∗ (E 0 ).

The external product µ : K(X) ⊗ K(Y ) = K(X × Y ) is defined by µ(a ⊗
b) = p∗1 (a) ⊗ p∗2 (b) where a ∈ K(X), b ∈ K(Y ) and p1 and p2 are the projections
from X × Y onto X and Y respectively, and the multiplication on the right-
hand side is the usual ring multiplication in K(X × Y ). We will also write
a ∗ b = µ(a ⊗ b).
Theorem 2. (External product theorem) The external product µ : K(X) ⊗
K(S 2 ) −→ K(X × S 2 ) is an isomorphism.
For the proof, please refer to Ref. [1]. First we need to understand K(S 2 ).
Note that the sphere is isomorphic to the complex projective line, S 2 ∼ = CP 1 ,
since a line can be identified with its slope z = z1 /z0 ∈ C and by Riemann
sphere we have C ∪ ∞ = S 2 . The canonical line bundle H over CP 1 satisfies
(H ⊗ H) ⊕ 1 ∼ = H ⊕ H for the following reason. The projective space CP 1 is the
2
subset of C consisting of all pairs (z0 , z1 ) of complex numbers, not both zero,
modulo the equivalence relation (z0 , z1 ) = (λz0 , λz1 ) for all nonzero complex
numbers λ. The complex plane C, with coordinate z, can be mapped into the
complex projective line by either z 7→ (z, 1) 7→ (z0 /z1 , 1) 7→ (z0 , z1 ) or z 7→
(1, 1/z) 7→ (1, z1 /z0 ) 7→ (z0 , z1 ). Considering a Riemann sphere, let the points
inside the circle S 1 on the complex plane C be expressed uniquely in the form
(z, 1) and the points outside the circle (1, 1/z), so when we pass through S 1 we
just need to multiply z. Therefore the clutching function is f : S 1 −→ GL1 (C)
defined by f (z) = (z). Thus the clutching function for H ⊕ H, which is the map
S 1 −→ GL2 (C), is given by
 
z 0
z 7−→ ,
0 z

where the upper left z corresponds to the first H and the lower right z cor-
responds to the second H. Similarly for (H ⊗ H) ⊕ 1, the clutching function
is  2 
z 0
z 7−→ .
0 1
Since GL2 (C) is path-connected, there is a homotopy between the two maps
from S 1 to GL2 (C). By the following proposition1 :
Proposition 1. The map Φ : [S k−1 , GLn (C)] −→ V ectnC (Sk ) which sends a
clutching function f to the vector bundle Ef is a bijection.
where [S k−1 , GLn (C)] is a homotopy class of S k−1 , GLn (C), we have the con-
clusion that (H ⊗ H) ⊕ 1 = H ⊕ H. In K(S 2 ) it can be written as H 2 + 1 = 2H
1 For proof see Ref. [1]

5
or (H −1)2 = 0. This gives us a natural ring homomorphism Z[H]/(H −1)2 −→
K(S 2 ) where Z[H]/(H − 1)2 is the ring of polynomials in H with integer coeffe-
cients, modulo the relation (H −1)2 = 0. This means we can deffine a homomor-
phism µ : K(X) ⊗ Z[H]/(H − 1)2 −→ K(X) ⊗ K(S 2 ) −→ K(X × S 2 ) with the
second map being the external product. It can be proved that this is a ring iso-
morphism (see Ref.[1]). If we take X = pt, we find that K(S 2 ) ∼
= Z[H]/(H −1)2
as a ring, which is generated by 1 and H or 1 and H − 1. Let (m, n), (p, q) de-
note m + n(H − 1) and p + q(H − 1) for m, n, p, q ∈ Z, respectively. Since
(m + n(H − 1)) (p + q(H − 1)) = mp + (mq + np)(H − 1) + nq(H − 1)2 =
mp+(mq+np)(H −1), we have (m, n)(p, q) = (mp, np+mq), i.e. Z(S e 2) ∼ = Z⊕Z.
At one point x0 , m + n(H − 1) = m + n(1 − 1) = m since H at one point
is a one dimensional trivial bundle. Since K(X) e is the kernel of the map
K(X) → K(x0 ) = Z, that is to say, the kernel of the restriction (m, n) → m, we
have K(X)
e ∼
= Z generated by H −1. Because of (H −1)2 = 0, the multiplication
2
in K(S
e ) is completely trivial: The product of any two elements is zero.

4 Bott periodicity
Bott periodicity Theorem allows us to compute K(X) for more complicated
spaces X, and in particular for all spheres.

Proposition 2. If X is compact Hausdorff and A ⊂ X is a closed subspace,

i q
then the inclusion and quotient maps A → X → X/A induce homomorphisms
∗ ∗
q i
K(X/A)
e → K(X)
e → K(A) for which the kernel of i∗ equals the image of q ∗ .
Proposition 3. If A is contractible, the quotient map q : X → X/A induces a
bijection q ∗ : V ectn (X/A) → V ectn (X) for all n.

Proofs can be found in Ref.[1]. We start with the inclusion A ,→ X and add
spaces by at each step forming the union of the preceding space with the cone
of the space two steps back. We then also quotient out by the most recently
attached cone, giving us the following sequence of inclusions (horizontal maps)
and quotients (vertical maps):

A ,→ X ,→X ∪ CA ,→ (X ∪ CA) ∪ CX ,→ ((X ∪ CA) ∪ CX) ∪ C(X ∪ CA) ,→ · · ·

↓ ↓ ↓
X/A SA SX

For example, for the first vertical arrow, X ∪ CA/CA = X/A. By the two
e ∪CA) ∼
propositions above, we have K(X = K(X/A),
e therefore an exact sequence
∗ ∗
q i
K(X/A) → K(X) → K(A). For the second vertical arrow, we have (X ∪ CA) ∪
e e e
CX/CX = SA and K((X e ∪ CA) ∪ CX/CX) ∼ = K(SA).
e The exact sequence
∗ ∗
q i
is K(SA)
e → K(X
e ∪ CA) → K(X)
e since X ∪ CA/X = SA. Combining these

6
two exact sequences by noting K(Xe ∪ CA) =∼ K(X/A),
e we have K(SA)
e →
K(X/A)
e → K(X)
e → K(A).
e It can be extended in the same fashion so we have
a long exact sequence of K:
e

· · · → K(SX)
e → K(SA)
e → K(X/A)
e → K(X)
e → K(A).
e (3)

If X is the wedge product A ∨ B then X/A = B. For example S 1 ∨ S 1 has a

shape of two circles with one common point, and collapsing one of them results
q∗ i∗ e
a circle. Considering the last three terms K(B) e → K(Ae ∨ B) → K(A), first
we need to show that it is a short exact sequence, i.e. i∗ is surjective and q ∗ is
injective. Let i0 be the map i0 : A ∨ B/B → A, so i0 i is an identity map of A.
It induces an identity map i∗ i0∗ of K(A),
e which indicates that i∗ is surjective.
Similarly, let q be the map q : B → A ∨ B, then qq 0 is the identity map of B.
0 0

The induced identity map of K(B)e q 0∗ q ∗ indicates that q ∗ is injective. By the

Splitting lemma, we have that K(A ∨ B) ∼
e = K(A)
e ⊕ K(B).
e In Eq. 3, Let X be
X × Y and A be X ∨ Y , we have

· · · → K(S(X
e × Y )) → K(S(X
e ∨ Y )) → K(X
e ∧ Y ) → K(X
e × Y ) → K(X
e ∨ Y ),

where K(X e ∨ Y) ∼ = K(X)

e ⊕ K(Y
e ) as we have seen, and K(S(X e ∼
∨ Y )) =
K(SX) ⊕ K(SY ) for a similar argument. Now it can be shown that the last
e e
q∗ i∗ e
three terms, K(X e ∧ Y ) → K(X e × Y ) → K(X ∨ Y ) is a short exact sequence.
Define q 0 : X ∨ Y → X, q 00 : X ∨ Y → Y , p1 : X × Y → X and p2 : X × Y → Y .
For i : X ∨ Y → X × Y , we have q 0 = p1 ◦ i and q 00 = p2 ◦ i. So q 0∗ ⊕ q 00∗ =
i∗ ◦ p∗1 ⊕ i∗ ◦ p∗2 = i∗ ◦ (p∗1 ⊕ p∗2 ). On the other hand, q 0∗ ⊕ q 00∗ : K(X)
e ⊕ K(Y
e )→
∗
K(X ∨ Y ) is an isomorphism, therefore i has to be surjective. Similarly, the
e
map K(S(X
e × Y )) → K(S(Xe ∨ Y )) is also a sujection, which implies that
the map K(S(Xe ∨ Y )) → K(X e ∧ Y )) takes K(S(X
e ∨ Y )) to the identity of
q∗
K(X
e ∧ Y )), thus K(X
e ∧ Y ) → K(Xe × Y ) is injective. At last we have a short
exact sequence

0 → K(X
e ∧ Y ) → K(X
e × Y ) → K(X
e ∨ Y ) → 0,

e ×Y ) ∼
which leads to a splitting: K(X e ∧Y )⊕ K(X)⊕
= K(X e K(Y
e ). Consequently
we have

K(X × Y ) = K(X
e ×Y)⊕Z
= K(X
e ∧ Y ) ⊕ K(X)
e ⊕ K(Y
e )⊕Z

On the other hand,

K(X) ⊗ K(Y ) = (K(X)

e ⊕ Z) ⊗ (K(Y
e ) ⊕ Z)
= (K(X)
e ⊗ K(Y
e )) ⊕ (K(X)
e ⊗ Z) ⊕ (K(Y
e ) ⊗ Z) ⊕ (Z ⊗ Z)
= (K(X)
e ⊗ K(Y
e )) ⊕ K(X)
e ⊕ K(Y
e )⊕Z

7
where K(X)
e ⊗ Z = K(X)
e was used for the following reason. For E ∈ K(X) e and
n ∈ Z, E⊗n = E⊗(1 ⊕1 ⊕· · · 1 ) = E⊕E⊕· · ·⊕E ∈ K(X) e and so K(X)⊗Z
e =
K(X). Remembering the external product K(X) ⊗ K(Y ) → K(X × Y ), we can
e
define a reduced external product K(X)⊗ e e ) → K(X
K(Y e ∧Y ). Let Y be S 2 ,
there is an isomorphism K(X)
e ⊗ K(S
e 2
)∼= K(X
e 2
∧ S ) induced by isomorphism
K(X) ⊗ K(S 2 ) ∼ = K(X × S 2
). Since Σ n
X = S n
∧ X is a quotient of S n X, we
e n ∧ X) ∼
have K(S e n X), so K(X)
= K(S e ⊗ K(Se 2) ∼ e 2 X). In fact the map
= K(S
2
K(X) → K(X) ⊗ K(S ) is also an isomorphism for the following reason. Since
e e e
e 2 ) by addition, take n(H − 1) ∈ K(S
(H − 1) generates K(S e 2 ) and a ∈ K(X),
e we
have a ⊗ n(H − 1) = a ⊗ ((H − 1) + (H − 1) + · · · + (H − 1)) = n(a ⊗ (H − 1)). It
seems that a is being mapped to {n(a ⊗ (H − 1))|n ∈ Z} and thus not injective,
but one should realize that ma for m ∈ Z lies also in K(X) e and also being
mapped to {n(a ⊗ (H − 1))|n ∈ Z}, so we have {ma|m ∈ Z} in K(X) e and
2
{n(a ⊗ (H − 1))|n ∈ Z} in K(X)
e ⊗ K(S
e ). One isomorphism naturally appears
as a 7→ a ⊗ (H − 1) for any a ∈ K(X).
e With this information, we arrive at the
Bott periodicity:
K(X)
e ∼ e 2 X).
= K(S
As an example, we can take a look at K(S e 0 ) and K(Se 2 ). As we have seen,
K(S e 2S0) ∼
e 2 ) = Z. S 0 is a pair of points, i.e. S 0 = {x0 , x1 }, and K(S e 2 ).
= K(S
A vector bundle over S consists of one copy of C over x0 and one copy of C p
0 m

over x1 . It follows that K(S 0 ) = {m − n, p − q} ∼

= Z ⊕ Z where m − n represents
the equivalence class of Cm − Cn . The ring structure is the usual multiplication
on Z on each factor. The reduced group K(S e 0 ) is the kernel of the restriction
K(S 0 ) → K(x0 ), i.e. the kernel of the map sending {Cm − Cn , Cp − Cq } to
{Cm − Cn }. The kernel consists of those elements with m = n, hence K(S e 0) =
p q ∼
{0, C − C } = {p − q} = Z, and the ring structure is the usual multiplication
on Z. Thus we can see that K(S e 0 ) = K(S
e 2 ).
With Bott periodicity, we can compute K(S e n ) for all n > 0. For even n,
e n) ∼
K(S e 2 ) = Z as S 2 S n = S n+2 . For odd n, we need to know K(S
= K(S e 1)
1
first. Since the upper and lower halves of S are contractible, then every vector
bundle over S 1 is determined by the map of S 0 into GL(C, n). Since clutching
functions determine the vector bundle up to homotopic maps into GL(C, n), the
path connectedness of GL(C, n) implies the every bundle over S 1 is isomorphic
to the line bundle, which has the shape of a closed belt and is trivial. Hence we
have for n odd K(Se n ) = 0.

5 Cohomology Theory
As we saw earlier the exact sequence of K
e groups is

e 2 X) → K(S
K(S e 2 A) → K(S(X/A))
e → K(SX)
e → K(SA)
e → K(X/A)
e
→ K(X)
e → K(A).
e

8
 e −n (X) = K(S
If we set K e −n (X, A) = K(S
e n X) and K e n (X/A)), this sequence
can be written as
e −2 (X) → K
K e −2 (A) → K
e −1 (X, A) → K
e −1 (X) → K
e −1 (A) → K
e 0 (X, A)
→K e 0 (X) → Ke 0 (A).

e 2 X) ∼
By Bott periodicity, we have K(S = K(X)
e = K e 2 A) ∼
e 0 (X) and K(S =
0
K(A)
e =K e (A). Then the six-term exact sequence can be written

e 0 (X, A) → K
K e 0 (X) → K
e 0 (A)
↑ ↓
Ke −1 e −1
(X, A) ← K e −1 (A)
(X) ← K

One can define Ke ∗ (X) = K e −1 (X) then the external product K(X)
e 0 (X) ⊕ K e ⊗
∗ ∗ ∗
K(Y ) → K(X ∧ Y ) induces K (X) ⊗ K (Y ) → K (X ∧ Y ). A product on
e e e e e
Ke ∗ (X) can be constructed by composing the external product with the map
induced by the diagonal mapping ∆ : X → X ∧ X, x 7→ (x, x), i.e.
∗
e ∗ (X) ⊗ K
K e ∗ (X ∧ X) ∆
e ∗ (X) → K →Ke ∗ (X).

This makes Ke ∗ (X) into a ring. Similarly for K ∗ (X, A), we have a map K
e ∗ (X, A)⊗
∗
e (Y, B) → K ∗
e (X/A ∧ Y /B). From diagonal map X/(A ∪ B) → X/A ∧ X/B
K
we get a product K e ∗ (X, A) ⊗ Ke ∗ (X, B) → Ke ∗ (X, A ∪ B).

References
[1] Allen Hatcher, Vector bundles and K theory (online)
[2] Max Karoubi, K theory, Springer-Verlag Berlin Heidelberg 1978
[3] Chris Blair, Some K-theory examples (online)
[4] Kasper Olsena, and Richard J. Szaboa, Constructing D-branes from K-
theory, arXiv: hep-th/9907140