nLab basic complex line bundle on the 2-sphere

Redirected from "basic line bundle on the 2-sphere".

Context

Bundles

bundles

Contents

Idea

The basic line bundle on the 2-sphere is the complex line bundle on the 2-sphere whose first Chern class is a generator ±1∈ℤ≃H 2(S 2,ℤ)\pm 1 \in \mathbb{Z} \,\simeq\, H^2(S^2, \mathbb{Z}), equivalently the tautological line bundle on the Riemann sphere regarded as complex projective 1-space.

This is the pullback bundle of the map S 2→BU(1)≃B 2ℤS^2 \to B U(1) \simeq B^2 \mathbb{Z} to the classifying space/Eilenberg-MacLane space which itself represents a generator of the homotopy group π 2(S 2)≃ℤ\pi_2(S^2) \simeq \mathbb{Z}.

Beware that this basic line bundle is sometimes called the “canonical line bundle on the 2-sphere”, but it is not isomorphic to what in complex geometry is called the canonical bundle of the 2-sphere regarded as a Riemann surface. Instead it is “one half” of the latter, its theta characteristic. See also at geometric quantization of the 2-sphere.

The basic line bundle is the canonically associated bundle to basic principal U(1)-bundle: the complex Hopf fibration. Another name for it is the tautological line bundle for the complex projective line ℙ 1(ℂ)\mathbb{P}^1(\mathbb{C}) (the Riemann sphere), namely the map ℂ 2∖{(0,0)}→ℙ 1(ℂ)\mathbb{C}^2 \setminus \{(0, 0)\} \to \mathbb{P}^1(\mathbb{C}) mapping (x,y)(x, y) to [x;y][x; y].

Definition

The classifying space for principal U(1)-bundles, or equivalently (via forming associated bundles) that of complex line bundles is B U ( 1 ) B U(1) , which as a Grassmannian is the infinite complex projective space ℂP ∞\mathbb{C}P^\infty. The homotopy type of this space is that of the Eilenberg-MacLane space K(ℤ,2)K(\mathbb{Z},2). This means that K(ℤ,2)K(\mathbb{Z},2) is in particular path-connected and has second homotopy group the integers: π 2(K(ℤ,2))≃ℤ\pi_2(K(\mathbb{Z},2)) \simeq \mathbb{Z}.

It being the classifying space for complex line bundles means that

{isomorphism classes of complex line bundles onS 2}≃{continuous functions S 2⟶K(ℤ,2) up to homotopy}≃π 2(K(ℤ,2))≃ℤ. \left\{ \array{ \text{isomorphism classes of} \\ \text{complex line bundles} \\ \text{on}\,\, S^2 } \right\} \;\simeq\; \left\{ \array{ \text{continuous functions} \\ S^2 \longrightarrow K(\mathbb{Z},2) \\ \text{up to homotopy} } \right\} \;\simeq\; \pi_2(K(\mathbb{Z},2)) \;\simeq\; \mathbb{Z} \,.

The (isomorphism class) of the complex line bundle which corresponds to +1∈ℤ+1 \in \mathbb{Z} under this sequence of isomorphisms is called the basic complex line bundle on the 2-sphere.

Hence the basic complex line bundle on the 2-sphere is the pullback bundle of the universal complex line bundle on B U ( 1 ) B U(1) along the map S 2→BU(1)S^2 \to B U(1) which represents the element 1∈ℤ≃π 2(BU(1))1 \in \mathbb{Z} \simeq \pi_2(B U(1)). If the classifying space B U ( 1 ) B U(1) is represented by the infinite complex projective space ℂP ∞\mathbb{C}P^\infty with its canonical CW-complex structure (this prop.), then this map is represented by the canonical cell inclusion S 2↪ℂP ∞S^2 \hookrightarrow\mathbb{C}P^\infty.

Notice that there is a non-trivial automorphism of ℤ\mathbb{Z} as an abelian group given by n↦−nn \mapsto -n. This means that there is an ambiguity in the definition of the basic line bundle on the 2-sphere.

Properties

General

Lemma

(clutching construction of the basic line bundle)

Under the clutching construction of vector bundles on the 2-sphere, the basic complex line bundle on the 2-sphere is given by the transition function

ℂ⊃S 1⟶GL(1,ℂ)⊂ℂ \mathbb{C} \supset \, S^1 \longrightarrow GL(1,\mathbb{C}) \, \subset \mathbb{C}

from the Euclidean circle S 1⊂ℝ 2≃ℂS^1 \subset \mathbb{R}^2 \simeq \mathbb{C} to the complex general linear group in 1-dimension, which is GL(1,ℂ)≃ℂ∖{0}GL(1,\mathbb{C}) \simeq \mathbb{C} \setminus \{0\} given simply by

z↦z, z \mapsto z \,,

Alternatively, due to the sign ambiguity in the definition of the basic bundle, its clutching transition function is given by

z↦−z. z \mapsto - z \,.
Proof

Under the clutching construction the isomorphism class of a complex line bundle corresponds to the homotopy class of its clutching transition function

S 1→GL(1,ℂ)≃ℂ∖{0} S^1 \to GL(1, \mathbb{C}) \simeq \mathbb{C} \setminus \{0\}

hence to an element of the fundamental group π 1(ℂ∖{0})≃ℤ\pi_1(\mathbb{C} \setminus \{0\}) \simeq \mathbb{Z}. Hence by definition, the basic bundle has clutching transition function corresponding to ±1∈ℤ≃[S 1,GL(1,ℤ)]\pm 1 \in \mathbb{Z} \simeq [S^1, GL(1,\mathbb{Z})] and this element is represented by the function z↦±zz \mapsto \pm z.

Proposition

(fundamental tensor/sum relation of the basic complex line bundle)

Under direct sum of vector bundles ⊕ S 2\oplus_{S^2} and tensor product of vector bundles ⊗ S 2\otimes_{S^2}, the basic line bundle on the 2-sphere H→S 2H \to S^2 satisfies the following relation

H⊕ S 2H≃(H⊗ S 2H)⊕ S 21 S 2 H \oplus_{S^2} H \;\simeq\; \left( H \otimes_{S^2} H \right) \oplus_{S^2} 1_{S^2}

(where 1 S 21_{S^2} denotes the trivial vector bundle complex line bundle on the 2-sphere).

(e.g (Hatcher, Example 1.13))

Proof

Via the clutching construction there is a single transition function of the form

S 1⟶GL(n,ℂ) S^1 \longrightarrow GL(n,\mathbb{C})

that characterizes all the bundles involved. With S 1↪ℂS^1 \hookrightarrow \mathbb{C} identified with the topological subspace of complex numbers of unit absolute value, the standard choice for these functions is

  • for the trivial line bundle 1 S 21_{S^2} we may choose

    f 1:z↦(1)f_1 \colon z \mapsto \left( 1 \right);

  • for the basic line bundle we may choose (by lemma )

    f H:z↦(z)f_H \colon z \mapsto \left( z\right)

This yields

  • for H⊗H⊕1 S 2H \otimes H \oplus 1_{S^2} the clutching function

    z↦(z 2 0 0 1)z \mapsto \left( \array{ z^2 & 0 \\ 0 & 1 }\right)

  • for H⊕HH \oplus H the clutching function

    z↦(z 0 0 z)z \mapsto \left( \array{ z & 0 \\ 0 & z } \right).

Since the complex general linear group Gl(n,ℂ)Gl(n,\mathbb{C}) is path-connected (by this prop.), there exists a continuous function

γ:[0,1]⟶GL(2,ℂ) \gamma \colon [0,1] \longrightarrow GL(2,\mathbb{C})

connecting the identity matrix on ℂ 2\mathbb{C}^2 with the one that swaps the two entries, i.e.

with γ(0)=(1 0 0 1)\gamma(0) = \left( \array{ 1 & 0 \\ 0 & 1 } \right)

and γ(1)=(0 1 1 0)\gamma(1) = \left( \array{ 0 & 1 \\ 1 & 0 } \right)

Therefore the function

S 1×[0,1] ⟶ GL(2,ℂ) (z,t) ⟶AA f H⊕1(z)⋅γ(t)⋅f 1⊕H(z)⋅γ(t) \array{ S^1 \times [0,1] &\overset{}{\longrightarrow}& GL(2,\mathbb{C}) \\ (z,t) &\overset{\phantom{AA}}{\longrightarrow}& f_{H \oplus 1}(z) \cdot \gamma(t) \cdot f_{1 \oplus H}(z) \cdot \gamma(t) }

(with matrix multiplication on the right) is a left homotopy from f H⊕Hf_{H \oplus H} to f H⊗H⊕1f_{H \otimes H \oplus 1}.

Remark

(fundamental product theorem in topological K-theory)

Under the map

Vect(S 2) /∼⟶K(X) Vect(S^2)_{/\sim} \longrightarrow K(X)

that sends complex vector bundles to their class in the topological K-theory ring K(X)K(X), the fundamental tensor/sum relation of prop. says that the K-theory class HH of the basic line bundle in K(X)K(X) satisfies the relation

(H−1) 2 =H 2+1−2H⏟=H 2+1 = 0 \begin{aligned} (H - 1)^2 & = H^2 + 1 - \underset{= H^2 + 1}{\underbrace{2 H}} \\ = & 0 \end{aligned}

in K(X)K(X).

(Notice that H−1H-1 is the image of [H][H] in the reduced K-theory K˜(X)\tilde K(X) of S 2S^2 under the splitting K(X)≃K˜(X)⊕ℤK(X) \simeq \tilde K(X) \oplus \mathbb{Z} (by this prop.).)

It follows that there is a ring homomorphism of the form

ℤ[h]/((h−1) 2) ⟶ K(S 2) h ↦AAA H \array{ \mathbb{Z}[h]/\left( (h-1)^2 \right) &\overset{}{\longrightarrow}& K(S^2) \\ h &\overset{\phantom{AAA}}{\mapsto}& H }

from the polynomial ring in one abstract generator, quotiented by this relation, to the topological K-theory ring.

It turns out that this homomorphism is in fact an isomorphism, hence that the relation (H−1) 2=0(H-1)^2 = 0 from prop. is the only relation satisfied by the basic complex line bundle in topological K-theory.

More generally, for XX a topological space, then there is a composite ring homomorphism

K(X)⊗ℤ[h]/((h−1) 2) ⟶ K(X)×K(S 2) ⟶ K(X×S 2) (E,h) ↦AAA (E,H) ↦AAA (π X *E)⋅(π S 2 *H) \array{ K(X) \otimes \mathbb{Z}[h]/((h-1)^2) & \longrightarrow & K(X) \times K(S^2) & \longrightarrow & K(X \times S^2) \\ (E, h) &\overset{\phantom{AAA} }{\mapsto}& (E,H) &\overset{\phantom{AAA}}{\mapsto}& (\pi_{X}^\ast E) \cdot (\pi_{S^2}^\ast H) }

to the topological K-theory ring of the product topological space X×S 2X \times S^2, where the second map is the external tensor product of vector bundles.

This composite is an isomorphism if XX is a compact Hausdorff space (for X=*X = \ast the point space this reduces to the previous statement).

This is called the fundamental product theorem in topological K-theory. It is the main ingredient in the proof of Bott periodicity in complex topological K-theory.

As an index bundle

We discuss how the basic complex line bundle on S 2S^2 is isomorphically the subbundle of +1 eigenspaces of a natural S 2S^2-parameterized Dirac-like linear operator acting on ℂ 2\mathbb{C}^2, which gives its natural realization as a Fredholm index bundle (cf. Rem. below). This kind of construction plays a key role in the discussion of Chern insulators (topological phases of matter with nontivial first Chern class of the valence bundle, cf. Rem. below).


To this end, consider:

  • ℍ\mathbb{H} the real star-algebra of quaternions, with norm given by |q| 2≔qq *{\vert q \vert}^2 \coloneqq q q^\ast,

  • ℍ im⊂ℍ\mathbb{H}_{im} \subset \mathbb{H} the subspace of imaginary quaternions (q∈ℍq \in \mathbb{H} such that q *=−qq^\ast = - q), in which we choose, as usual, an orthonormal linear basis i,j,k\mathbf{i}, \mathbf{j}, \mathbf{k} such that ij=k\mathbf{i} \mathbf{j} = \mathbf{k},

  • ℍ⟶End(ℂ 2)\mathbb{H} \longrightarrow End(\mathbb{C}^2) the star-algebra homomorphism which identifies the unit imaginary quaternions with Pauli matrices:

    ℍ ⟶−−σ−− End(ℂ 2) 1 ↦ [+1 +0 +0 +1] i ↦ i[+1 +0 +0 −1] j ↦ i[+0 +1 +1 +0] k ↦ i[+0 +i −i +0], \begin{array}{rcr} \mathbb{H} &\overset {\phantom{--} \sigma \phantom{--}} {\longrightarrow}& End(\mathbb{C}^2) \\ 1 &\mapsto& \left[\begin{matrix} \phantom{+}1 & \phantom{+}0 \\ \phantom{+}0 & \phantom{+}1 \end{matrix}\right] \\ \mathbf{i}\, &\mapsto& \mathrm{i} \left[\begin{matrix} \phantom{+}1 & \phantom{+}0 \\ \phantom{+}0 & -1 \end{matrix}\right] \\ \mathbf{j}\, &\mapsto& \mathrm{i} \left[\begin{matrix} \phantom{+}0 & \phantom{+}1 \\ \phantom{+}1 & \phantom{+}0 \end{matrix}\right] \\ \mathbf{k}\, &\mapsto& \mathrm{i} \left[\begin{matrix} \phantom{+}0 & \phantom{+}\mathrm{i} \\ -\mathrm{i} & \phantom{+}0 \end{matrix}\right] \mathrlap{\,,} \end{array}
  • S(ℍ)S(\mathbb{H}) the 3-sphere of unit norm quaternions, which by the above homomorphism is identified with the special unitary matrices,

    S(ℍ)⟶∼−−σ−−SU(2) S(\mathbb{H}) \underoverset {\sim} {\phantom{--} \sigma \phantom{--}} {\longrightarrow} SU(2)

    because

    det(x 0id+x 1σ i+x 2σ j+x 3σ k) =det[x 0+ix 1 ix 2−x 3 ix 2+x 3 x 0−ix 1] =(x 0) 2+(x 1) 2+(x 2) 2+(x 3) 2 \begin{array}{l} det\big( x_0 \mathrm{id} + x_1 \sigma_{\mathbf{i}} + x_2 \sigma_{\mathbf{j}} + x_3 \sigma_{\mathbf{k}} \big) \\ \;=\; det \left[ \begin{matrix} x_0 + \mathrm{i} x_1 & \mathrm{i}x_2 - x_3 \\ \mathrm{i} x_2 + x_3 & x_0 - \mathrm{i} x_1 \end{matrix} \right] \\ \;=\; (x_0)^2 + (x_1)^2 + (x_2)^2 + (x_3)^2 \end{array}

    and

    (x 0id+x 1σ i+x 2σ j+x 3σ k) †(x 0id+x 1σ i+x 2σ j+x 3σ k) =(x 0id−x 1σ i−x 2σ j−x 3σ k)(x 0id+x 1σ i+x 2σ j+x 3σ k) =(x 0) 2+(x 1) 2+(x 2) 2+(x 3) 2, \begin{array}{l} \big( x_0 \mathrm{id} + x_1 \sigma_{\mathbf{i}} + x_2 \sigma_{\mathbf{j}} + x_3 \sigma_{\mathbf{k}} \big)^\dagger \big( x_0 \mathrm{id} + x_1 \sigma_{\mathbf{i}} + x_2 \sigma_{\mathbf{j}} + x_3 \sigma_{\mathbf{k}} \big) \\ \;=\; \big( x_0 \mathrm{id} - x_1 \sigma_{\mathbf{i}} - x_2 \sigma_{\mathbf{j}} - x_3 \sigma_{\mathbf{k}} \big) \big( x_0 \mathrm{id} + x_1 \sigma_{\mathbf{i}} + x_2 \sigma_{\mathbf{j}} + x_3 \sigma_{\mathbf{k}} \big) \\ \;=\; (x_0)^2 + (x_1)^2 + (x_2)^2 + (x_3)^2 \,, \end{array}
  • the group action of SU(2)≃Spin(3)SU(2) \simeq Spin(3) through SO(3) on ℍ im≃ ℝℝ 3\mathbb{H}_{im} \simeq_{{}_{\mathbb{R}}} \mathbb{R}^3 given by

    S(ℍ)×ℍ im⏞Spin(3)×ℝ 3 ⟶ ℍ im⏞ℝ 3 (q,x) ↦ qxq *, \begin{array}{ccc} \overset{ Spin(3) \times \mathbb{R}^3 }{ \overbrace{ S(\mathbb{H}) \times \mathbb{H}_{im} } } &\longrightarrow& \overset{ \mathbb{R}^3 }{ \overbrace{ \mathbb{H}_{im} } } \\ (q, x) &\mapsto& q x q^\ast \mathrlap{\,,} \end{array}
  • S(ℍ im)S(\mathbb{H}_{im}) the 2-sphere of unit-norm imaginary quaternions.

Observing that the complex Hopf fibration is equivalently (see there):

S(ℍ) ⟶−−h ℂ−− S(ℍ im) q ↦ q⋅i⋅q * \begin{array}{c} S(\mathbb{H}) &\overset {\phantom{--} h_{\mathbb{C} \phantom{--}} } {\longrightarrow} & S(\mathbb{H}_{im}) \\ q &\mapsto& q \cdot \mathbf{i} \cdot q^\ast \end{array}

whose fiber over i\mathbf{i} is

S(ℂ)⏞U(1) ↪−−− S(ℍ)⏞SU(2) x+yi ↦ x+yi, \begin{array}{rcl} \overset{ \mathrm{U}(1) }{ \overbrace{ S(\mathbb{C}) } } &\xhookrightarrow{\phantom{---}}& \overset{ \mathrm{SU}(2) }{ \overbrace{ S(\mathbb{H}) } } \\ x + y \mathrm{i} &\mapsto& x + y \mathbf{i} \,, \end{array}

it follows that the basic complex line bundle on the 2-sphere is the associated complex line bundle

S 3× U(1)ℂ={(q,z)∈S(ℍ)×ℂ} /((q,z)∼(qα *,αz)∀ α∈S(ℂ)). S^3 \times_{\mathrm{U}(1)} \mathbb{C} \;=\; \Big\{ (q,z) \in S(\mathbb{H}) \times \mathbb{C} \Big\}_{ \big/\big( (q,z) \sim (q \alpha^\ast, \alpha z) \;\forall_{\alpha \in S(\mathbb{C})}\, \big) } \,.

As such, this is a subbundle of the trivial complex vector bundle of rank=2=2, via this map: This subbundle is evidently characterized as the bundle of +1 eigenspaces of this S 2S^2-parameterized linear operator:

(1)S(ℍ) ⟶−−−iσ−− End(ℂ 2) qiq * ↦ −iσ qiq *hence equivalently ofS 2 ⟶−−H−− End(ℂ 2) x→≡(x 1,x 2,x 3) ↦ −i(x 1σ i+x 2σ j+x 3σ k). \begin{array}{rrcl} S(\mathbb{H}) &\overset{\phantom{--} - \mathrm{i}\, \sigma \phantom{--}}{\longrightarrow}& End(\mathbb{C}^2) \\ q \mathbf{i} q^\ast &\mapsto& -\mathrm{i} \sigma_{ q \mathbf{i} q^\ast } \end{array} \;\;\;\; \text{hence equivalently of} \;\;\;\; \begin{array}{rrcl} S^2 &\overset{\phantom{--} H \phantom{--}}{\longrightarrow}& End(\mathbb{C}^2) \\ \vec x \equiv (x^1, x^2, x^3) &\mapsto& -\mathrm{i} \big( x^1 \sigma_{\mathbf{i}} + x^2 \sigma_{\mathbf{j}} + x^3 \sigma_{\mathbf{k}} \big) \,. \end{array}

(This statement, without proof, may be found for instance in Baum 2015 slide 32, Baum & Erp 2018 p 106-107).

Remark

(relation to Chern insulators) The normalized Hamiltonians (1) play a key role in solid state physics in the discussion of 2D 2-band crystals such as notably considered as models for Chern insulators (like the Haldane model, cf. Sticlet et al 2012 (2-3)).

It follows from the above discussion that

  1. any continuous family of such operators over a Brillouin torus T 2T^2 is given by a continuous map f:T 2⟶S 2f \colon T^2 \longrightarrow{} S^2,

  2. the +1 eigenspace bundle of the corresponding family of operators on T 2T^2 (the valence bundle in these models) has first Chern class equal to the pullback along ff of the generating class 1∈H 2(S 2;ℤ)1 \in H^2(S^2;\mathbb{Z}) of the basic line bundle on S 2S^2.

Similar arguments are ubiquituous in the literature on Chern insulators (cf. Sticlet et al 2012 p 2), though the concrete statement that all these constructions amount to pulling back the Bott generator along a map to the 2-sphere may not have been made explicit elsewhere.

Remark

(As a Fredholm index bundle)
With the basic line bundle realized as the kernel bundle of a ℂP 1\mathbb{C}P^1-parameterized projection operator P (−):ℂ 2→ℂ 2P_{(-)} \colon \mathbb{C}^2 \to \mathbb{C}^2, there is a a nice way to produce also a ℂP 1\mathbb{C}P^1-parameterized Fredholm operator

(2)S 2 ⟶F (−) Fred(ℋ), \begin{array}{rcl} S^2 &\overset{F_{(-)}}{\longrightarrow}& Fred(\mathscr{H}) \mathrlap{\,,} \end{array}

whose index bundle represents the class of the basic line bundle in topological K-theory under the Atiyah-Jänich theorem, like this:

This is a Fredholm operator with ker(F P)=ker(P)ker(F_P) = ker(P) and coker(F P)=0coker(F_P) = 0.

Remark

The first presentation in (1) makes manifest that the map S 2⟶End(ℂ 2)S^2 \longrightarrow End(\mathbb{C}^2) is Spin(3)Spin(3)-equivariant: with respect to the rotation action through SO(3)SO(3) on the left and the conjugation action by special unitary matrices in SU(2)≃Spin(3)SU(2) \simeq Spin(3) on the right.

From this it follows that also the corresponding parameterized Fredholm operator (2) is Spin(3)Spin(3)-equivariant, if we let Spin(3)≃SU(2)Spin(3) \simeq SU(2) act on

ℋ≃ℂ 2⊕ℓ 2(ℕ)≃⨁ ℕℂ 2 \mathscr{H} \,\simeq\, \mathbb{C}^2 \oplus \ell_2(\mathbb{N}) \,\simeq\, \bigoplus_{\mathbb{N}} \mathbb{C}^2

as the infinite direct sum of the defining representation of SU(2)SU(2).

The resulting Spin(3)Spin(3)-equivariant family of Fredholm operators should the the representation of the tautological equivariant line bundle in the equivariant K-theory of S 2S^2.

References

Last revised on May 24, 2026 at 08:29:33. See the history of this page for a list of all contributions to it.