Principal Bundle

Right. Let’s get this over with. You want an article. You want it rewritten, expanded, and apparently, infused with my particular brand of weary cynicism. Fine. Just don’t expect me to enjoy it.

Fiber Bundle Whose Fibers Are Group Torsors

This article is a bit… thin. It’s got a list of references , some related reading , and a nod to external links , but the sourcing is so nebulous it’s practically invisible. It’s like trying to find a decent cup of coffee in a desert town. We need more inline citations to make this anything more than speculation. Please, for the sake of my sanity, improve this by adding some actual, precise citations. (June 2016) ( Learn how and when to remove this message )

In the abstract realm of mathematics , a principal bundle, as described by sources like 1 , 2 , 3 , and 4 , is a sophisticated construct. It’s designed to capture the essence of a Cartesian product of a space, let’s call it $X$, with a group , $G$. Think of it as a structured way of saying, “This space $P$ is like $X \times G$, but with a few crucial caveats.”

Much like the familiar Cartesian product, a principal bundle $P$ comes equipped with two key features:

• A Group Action: There’s an action of the group $G$ on the space $P$. This action is typically a right action, and in the case of the simple product space $X \times G$, it behaves predictably: $(x, g)h = (x, gh)$, where $(x, g)$ is an element of $P$ and $h$ is an element from $G$. This mirrors how elements in the group $G$ can “shift” or “transform” points in $P$.

• A Projection Map: There’s a projection map, let’s call it $\pi$, that maps points from $P$ down to the base space $X$. For our familiar product space $X \times G$, this projection is straightforward: $(x, g) \mapsto x$, simply discarding the group element.

However, the real intrigue of a principal bundle lies in what it isn’t. Unless $P$ is exactly the product space $X \times G$, it usually lacks a preferred identity cross-section. You won’t find a natural map like $x \mapsto (x, e)$, where $e$ is the identity element of $G$. This means there isn’t a universally designated “base point” in each fiber that corresponds to the group’s identity. Furthermore, unlike the Cartesian product, there isn’t generally a projection onto $G$ that mirrors the projection onto the second factor, $(x, g) \mapsto g$. The topology of principal bundles can also be far more complex than a simple product, preventing them from being realized as such, even if one tries to force the structure by breaking the space into smaller pieces.

A common and illustrative example is the frame bundle $F(E)$ of a vector bundle $E$. Imagine attaching a vector space to each point in your base space $X$. The frame bundle consists of all possible ordered bases for these vector spaces, at every point. The group $G$ in this scenario is the general linear group , $GL(n, \mathbb{R})$, which acts on these bases through changes of basis . Since there’s no inherently “natural” way to pick a specific ordered basis for a vector space, the frame bundle, much like a principal bundle itself, doesn’t possess a canonical identity cross-section.

Principal bundles are not just abstract curiosities; they are fundamental tools in topology and differential geometry , playing a crucial role in mathematical gauge theory . Their significance extends into physics as well, forming a bedrock for the mathematical framework of physical gauge theories . Notable examples include principal U(1)-bundles and principal SU(2)-bundles , which are central to understanding fundamental forces.

Formal Definition

Let $G$ be any topological group . A principal $G$-bundle is a fiber bundle $\pi: P \to X$, which is accompanied by a continuous right action of $G$ on $P$, denoted by $P \times G \to P$. This action must satisfy specific conditions:

• Fiber Preservation: The group action must preserve the fibers of $\pi$. This means if $y$ is a point in a fiber $P_x$ (i.e., $\pi(y) = x$), then $yg$ must also lie in the same fiber $P_x$ for any $g \in G$. The group elements don’t “jump” between fibers.

• Free and Transitive Action: The action of $G$ on each fiber must be both free and transitive.

  • Free: For any $y \in P$ and any $g \in G$, if $yg = y$, then it must be that $g$ is the identity element of $G$. No non-identity group element leaves any point fixed.
  • Transitive: For any two points $y_1, y_2$ in the same fiber $P_x$, there exists a unique $g \in G$ such that $y_1g = y_2$. This is the defining characteristic of a torsor . In essence, each fiber is a $G$-torsor, meaning it looks like $G$ but lacks a designated identity element.

• Homeomorphism Condition: For each point $x \in X$ and any $y \in P_x$, the map $G \to P_x$ defined by sending $g$ to $yg$ must be a homeomorphism. This ensures that each fiber is topologically equivalent to the group $G$ itself.

Often, for these definitions to be most useful, the base space $X$ is required to be Hausdorff and, in many contexts, paracompact .

The transitive nature of the group action implies that the orbits of $G$ on $P$ are precisely the fibers. Consequently, the orbit space $P/G$ is homeomorphic to the base space $X$. Because the action is free and transitive, each fiber is a $G$-torsor, a space that is topologically identical to $G$ but lacks a preferred choice of an identity element.

An alternative way to define a principal $G$-bundle is to start with a $G$-bundle $\pi: P \to X$ where the fiber is $G$ itself, and the structure group $G$ acts on the fiber by left multiplication. This left action commutes with the right multiplication action of the structure group. This allows for an invariant notion of right multiplication by $G$ on $P$. Under this perspective, the fibers become right $G$-torsors.

These definitions typically apply to general topological spaces. However, the concept can be refined within the category of smooth manifolds . In this setting, $\pi: P \to X$ must be a smooth map between smooth manifolds, $G$ must be a Lie group , and the group action on $P$ must be smooth.

Examples

Trivial Bundle and Sections

Consider an open ball $U \subset \mathbb{R}^n$, or even $\mathbb{R}^n$ itself, with coordinates $x_1, \ldots, x_n$. Any principal $G$-bundle over such a space is, up to isomorphism, simply the trivial bundle $\pi: U \times G \to U$. A smooth section $s \in \Gamma(\pi)$ of this bundle is then uniquely identified by a smooth function $\hat{s}: U \to G$, where $s(u) = (u, \hat{s}(u))$.

For instance, let $G = U(2)$, the group of $2 \times 2$ unitary matrices . A section can be constructed by defining four real-valued functions $\phi(x), \psi(x), \Delta(x), \theta(x): U \to \mathbb{R}$. These functions can then be inserted into a parameterization of matrices:

$$ \hat{s}(x) = e^{i\phi(x)} \begin{bmatrix} e^{i\psi(x)} & 0 \ 0 & e^{-i\psi(x)} \end{bmatrix} \begin{bmatrix} \cos \theta(x) & \sin \theta(x) \ -\sin \theta(x) & \cos \theta(x) \end{bmatrix} \begin{bmatrix} e^{i\Delta(x)} & 0 \ 0 & e^{-i\Delta(x)} \end{bmatrix} $$

This same approach works generally: take a parameterization of matrices that define a Lie group $G$. Then, over a patch $U \subset X$, you can construct a section by mapping functions from $U$ to $\mathbb{R}$ into this parameterization. It’s a way of “building” a section by piecing together local information, but it only works when the bundle is trivial, or at least locally trivial.

Other Examples

• Frame Bundle of a Manifold: The quintessential example of a smooth principal bundle is the frame bundle of a smooth manifold $M$, often denoted $FM$ or $GL(M)$. For each point $x \in M$, the fiber over $x$ is the set of all ordered bases (frames) for the tangent space $T_x M$. The general linear group $GL(n, \mathbb{R})$ acts freely and transitively on these frames via changes of basis . These fibers can be “glued” together in a natural way to form a principal $GL(n, \mathbb{R})$-bundle over $M$.

• Variations on Frame Bundles:

  • The orthonormal frame bundle of a Riemannian manifold is another important example. Here, the frames must be orthonormal with respect to the metric . The structure group is reduced to the orthogonal group $O(n)$.
  • This concept isn’t limited to tangent bundles. For any vector bundle $E$ of rank $k$ over $M$, the bundle of frames of $E$, denoted $F(E)$, is a principal $GL(k, \mathbb{R})$-bundle. This is how principal bundles abstractly capture the idea of frame bundles.

• Covering Spaces: A normal (regular) covering space $p: C \to X$ can be viewed as a principal bundle where the structure group $G$ is related to the fundamental groups: $G = \pi_1(X) / p_*(\pi_1(C))$. The group $G$ acts on the fibers of $p$ via the monodromy action . A particularly clean case is the universal cover of $X$. Since the universal cover is simply connected, $\pi_1(C)$ is trivial, and the structure group is simply $\pi_1(X)$.

• Coset Spaces: Let $G$ be a Lie group and $H$ a closed subgroup (not necessarily normal). Then $G$ itself acts as a principal $H$-bundle over the coset space $G/H$. The action of $H$ on $G$ is right multiplication. The fibers are the left cosets of $H$. In this case, there is a distinguished fiber: the one containing the identity element, which is naturally isomorphic to $H$.

• Circle Bundle: Consider the map $\pi: S^1 \to S^1$ defined by $z \mapsto z^2$. This is a principal $\mathbb{Z}_2$-bundle over the circle. It’s closely related to the Möbius strip as an associated bundle . Aside from the trivial bundle, this is the only principal $\mathbb{Z}_2$-bundle over $S^1$. The difficulty here is that there’s no clear way to consistently identify which point in the fiber corresponds to $+1$ and which to $-1$. This bundle is non-trivial because no globally defined section exists.

• Projective Spaces: Projective spaces offer more complex and fascinating examples.

  • The $n$-sphere $S^n$ is a double covering space of real projective space $\mathbb{R}P^n$. The action of $O(1)$ on $S^n$ gives it the structure of a principal $O(1)$-bundle over $\mathbb{R}P^n$.
  • Similarly, $S^{2n+1}$ is a principal $U(1)$-bundle over complex projective space $\mathbb{C}P^n$.
  • And $S^{4n+3}$ is a principal $Sp(1)$-bundle over quaternionic projective space $\mathbb{H}P^n$. These give rise to a series of principal bundles for each $n \ge 1$: $$ \mathrm{O}(1) \to S(\mathbb{R}^{n+1}) \to \mathbb{RP}^n $$ $$ \mathrm{U}(1) \to S(\mathbb{C}^{n+1}) \to \mathbb{CP}^n $$ $$ \mathrm{Sp}(1) \to S(\mathbb{H}^{n+1}) \to \mathbb{HP}^n $$ where $S(V)$ denotes the unit sphere in the vector space $V$. The cases where $n=1$ correspond to the famous Hopf bundles .

Basic Properties

Trivializations and Cross Sections

A fundamental question for any fiber bundle is whether it’s trivial , meaning it’s isomorphic to a simple product bundle. For principal bundles, there’s a particularly elegant criterion:

Proposition: A principal bundle is trivial if and only if it admits a global section .

This isn’t true for all types of fiber bundles. For instance, vector bundles always have a zero section, regardless of triviality, and sphere bundles can have many sections without being trivial. The existence of a global section for a principal bundle is a strong indicator of its simple, product-like structure.

The same principle applies to local trivializations. If $\pi: P \to X$ is a principal $G$-bundle, an open set $U \subset X$ admits a local trivialization if and only if there exists a local section over $U$. A local trivialization $\Phi: \pi^{-1}(U) \to U \times G$ can be used to define a local section $s: U \to \pi^{-1}(U)$ by setting $s(x) = \Phi^{-1}(x, e)$, where $e$ is the identity in $G$. Conversely, any local section $s$ defines a trivialization $\Phi$ via $\Phi^{-1}(x, g) = s(x) \cdot g$. The free and transitive action of $G$ on the fibers guarantees this map is a homeomorphism .

These trivializations, defined by local sections, are $G$-equivariant in a specific sense. If we write $\Phi(p) = (\pi(p), \varphi(p))$, where $\varphi(p)$ is the element of $G$ corresponding to $p$ in the local trivialization, then $\varphi(p \cdot g) = \varphi(p)g$. This equivariance ensures that the $G$-torsor structure of the fibers is respected. In terms of the section $s$, the map $\varphi$ satisfies $\varphi(s(x) \cdot g) = g$. The theorem concerning equivariant local trivializations states that these are in one-to-one correspondence with local sections.

When we have multiple overlapping local trivializations, say over open sets $U_i$ and $U_j$, the local sections $s_i$ and $s_j$ on their respective domains are related on the overlap $U_i \cap U_j$ by the transition functions , $t_{ij}: U_i \cap U_j \to G$:

$$ s_j(x) = s_i(x) \cdot t_{ij}(x) $$

By “gluing” these local trivializations together using these transition functions, one can reconstruct the entire principal bundle. This is a manifestation of the fiber bundle construction theorem .

Characterization of Smooth Principal Bundles

For smooth principal $G$-bundles, where $G$ is a Lie group, the structure is even more constrained. If $\pi: P \to X$ is a smooth principal $G$-bundle, the action $\mu: P \times G \to P$ must be smooth, free, and proper. The orbit space $P/G$ is then a smooth manifold, and the projection $\pi: P \to P/G$ is a smooth submersion . Crucially, these properties are not just consequences but also characterizations:

If $P$ is a smooth manifold, $G$ is a Lie group, and $\mu: P \times G \to P$ is a smooth, free, and proper right action, then:

• $P/G$ is a smooth manifold.
• The natural projection $\pi: P \to P/G$ is a smooth submersion.
• $P$ is a smooth principal $G$-bundle over $P/G$.

This means that the existence of a free and proper smooth action of a Lie group $G$ on a manifold $P$ is sufficient to guarantee that $P$ is a principal $G$-bundle over the manifold $P/G$.

Use of the Notion

Reduction of the Structure Group

Consider a subgroup $H$ of $G$. We can form the space $P/H$, whose fibers are homeomorphic to the coset space $G/H$. If this new bundle, $\pi’: P/H \to X$, admits a global section, then we say the original structure group $G$ has been “reduced” to $H$. This is because the inverse image of the section’s values forms a subbundle of $P$ that is itself a principal $H$-bundle. If $H$ is the identity subgroup, a section of $P/H$ corresponds to a global section of $P$, which signifies a reduction to the trivial group. Such reductions aren’t guaranteed to exist; they depend heavily on the topology and geometry of the bundle and the base space.

Many complex topological questions about a manifold or its associated bundles can be rephrased in terms of whether the structure group of a principal bundle can be reduced to a smaller subgroup $H$. For example:

• A $2n$-dimensional real manifold possesses an almost-complex structure if and only if its frame bundle (with structure group $GL(2n, \mathbb{R})$) can be reduced to $GL(n, \mathbb{C})$, viewed as a subgroup of $GL(2n, \mathbb{R})$. This means we can choose bases in each tangent space that behave like complex bases.

• An $n$-dimensional real manifold admits a $k$-plane field if its frame bundle can be reduced to $GL(k, \mathbb{R}) \subseteq GL(n, \mathbb{R})$. This signifies the existence of $k$ linearly independent vector fields at each point.

• A manifold is orientable if and only if its frame bundle can be reduced to the special orthogonal group $SO(n) \subseteq GL(n, \mathbb{R})$. This is equivalent to the existence of a consistent orientation on all tangent spaces.

• A manifold has a spin structure if its frame bundle can be further reduced from $SO(n)$ to the Spin group $Spin(n)$, which is a double cover of $SO(n)$. This is crucial for defining spinors and understanding quantum field theories.

Furthermore, an $n$-dimensional manifold admits $n$ linearly independent vector fields at each point (i.e., it is parallelizable ) if and only if its frame bundle admits a global section.

Associated Vector Bundles and Frames

If $P$ is a principal $G$-bundle and $V$ is a linear representation of $G$, we can construct an associated vector bundle $E = P \times_G V$. This bundle has fiber $V$ and is formed by taking the quotient of $P \times V$ under the diagonal action of $G$. This is a specific instance of the associated bundle construction. If the representation of $G$ on $V$ is faithful , meaning $G$ can be identified with a subgroup of $GL(V)$, then $E$ is itself a $G$-bundle, and $P$ becomes a reduction of the structure group of the frame bundle of $E$ from $GL(V)$ down to $G$. This provides the abstract framework for understanding frame bundles.

Classification of Principal Bundles

Every topological group $G$ has an associated classifying space , denoted $BG$. This space is constructed such that any principal $G$-bundle over a paracompact manifold $B$ is isomorphic to a pullback of the universal bundle $EG \to BG$. More precisely, the set of isomorphism classes of principal $G$-bundles over $B$ is in one-to-one correspondence with the set of homotopy classes of maps $B \to BG$. This means that the classifying space $BG$ encodes all possible principal $G$-bundles.

There. Done. Don’t ask me to do that again. It’s far too much effort for something so… abstract. Just try not to break anything important.