Cotangent Spaces and One-Forms

Welcome to your third lesson in our journey through the mathematical foundations of General Relativity.

In the last lesson, we developed a rigorous definition of a vector at a point $p$ on a manifold $M$ . We defined it as a derivation, an operator that acts on smooth functions to produce a directional derivative. The set of all such vectors at a point forms the tangent space $T_{p} M$ , and we saw that the partial derivative operators ${\partial / \partial x^{μ}}$ form a natural basis for this space.

Today, we will introduce the dual concept: the cotangent space. For every vector space, linear algebra provides a natural "shadow" space called the dual space. The dual to the tangent space is the cotangent space, and its elements are known as covectors or, more commonly in geometry, one-forms. These are the second fundamental building block we need to construct the general tensors used in relativity.

Our learning objective is to define the cotangent space (dual space) and construct one-forms. We will accomplish this by:

Defining a one-form as a linear map from the tangent space to the real numbers.
Building a powerful visual intuition for one-forms as "stacks of surfaces."
Constructing the most important example of a one-form: the differential of a function, $df$ .
Defining the basis for the cotangent space, ${d x^{μ}}$ , and showing how it is dual to the vector basis ${\partial_{ν}}$ .

1. The Cotangent Space: The Dual of the Tangent Space

In the previous lesson, we saw that a tangent vector $X \in T_{p} M$ is a map that takes a smooth function $f \in C^{\infty} (M)$ and gives a real number, $X (f) \in R$ . We can also view this relationship from a different perspective. Instead of the vector "acting on" the function, what if we think of an object associated with the function "acting on" the vector?

This is the core idea of duality. Given a vector space $V$ , its dual space, denoted $V^{*}$ , is the set of all linear functions from $V$ to the real numbers $R$ . These linear functions are called functionals.

Applying this to our context:

Definition: Cotangent Space and One-Forms
The cotangent space to a manifold $M$ at a point $p$ , denoted $T_{p}^{*} M$ , is the dual vector space to the tangent space $T_{p} M$ . An element $ω \in T_{p}^{*} M$ is called a covector or a one-form. It is a linear map that takes a tangent vector $X \in T_{p} M$ as input and returns a real number:
$ω : T_{p} M \to R$
The action of a one-form $ω$ on a vector $X$ is often written like a scalar product: $ω (X)$ or $⟨ ω, X ⟩$ .

The set of one-forms $T_{p}^{*} M$ itself forms a vector space. We can add two one-forms $(ω + σ)$ and scale them by a constant $(aω)$ by defining their action on an arbitrary vector $X$ :

$(ω + σ) (X) := ω (X) + σ (X)$
$(aω) (X) := a (ω (X))$

Introduction to Tensor Calculus for General Relativity

The following text introduces the formal definition of a one-form and the cotangent space. As a physics student, you will appreciate the analogy it draws to the bra-ket notation in quantum mechanics, where the 'bras' \langle\phi| are elements of the dual Hilbert space and 'kets' (|\psi angle) are elements of the Hilbert space.

Please read Section 2.2, 'One-forms and dual vector space' (page 4). Focus on the definition of a one-form as a linear scalar function of a vector and the definition of the dual vector space.

2. Visualizing One-Forms

While the algebraic definition is precise, it can feel abstract. Fortunately, there is a very powerful way to visualize one-forms. Since a one-form assigns a number to each direction (each vector), we can visualize it as a set of surfaces (or lines in 2D). The value of the one-form acting on a vector is then the number of surfaces the vector "pierces".

A dense packing of surfaces corresponds to a "stronger" one-form in that direction, as a vector of a given length will pierce more surfaces and thus yield a larger number.

Geometric Interpretation of a One-Form on $\mathbb{R}^2$ — A visual representation of a one-form $\alpha$ as a set of contour lines. The value $\alpha(v_p)$ is the number of lines pierced by the vector $v_p$. In this case, the vector pierces about 6 lines, so $\alpha(v_p) \approx 6$.

The following video provides an excellent and detailed development of this visualization.

Tensors for Beginners 4: What are Covectors?

This video from the 'Tensors for Beginners' series by eigenchris provides a superb intuitive and visual explanation of what covectors (one-forms) are.

Watch the video from the beginning to 14:09. Please focus on these key ideas: Covectors as linear functions (00:18 - 04:01): Understand that a covector is fundamentally a linear function that maps a vector to a scalar. Visualization as stacks of lines (04:01 - 08:56): Grasp the analogy with topographic maps and how the value \omega(X) is found by counting the number of lines pierced by the vector X. Scaling and adding covectors (08:56 - 11:53): See how scaling a covector changes the density of the lines, and adding covectors corresponds to combining their densities. The dual space V^* (11:53 - 14:09): See how these rules confirm that the set of all covectors forms its own vector space.

3. The Differential of a Function: A Prototypical One-Form

The most important and natural example of a one-form is the differential of a smooth function. Given a smooth function $f : M \to R$ , its differential at a point $p$ , denoted $(df)_{p}$ , is a one-form. By definition, it must be a machine that eats a vector $X_{p} \in T_{p} M$ and spits out a number. What number? The directional derivative of $f$ along $X_{p}$ .

Definition: The Differential $df$
The differential of a function $f \in C^{\infty} (M)$ at a point $p$ is the one-form $(df)_{p} \in T_{p}^{*} M$ whose action on a vector $X_{p} \in T_{p} M$ is defined as:
$(df)_{p} (X_{p}) := X_{p} (f)$

This definition beautifully crystallizes the concept of duality. The expression $X_{p} (f)$ can be read in two ways:

The vector $X_{p}$ acts on the function $f$ .
The one-form $(df)_{p}$ acts on the vector $X_{p}$ .

Both interpretations yield the same number. This is the heart of the relationship between tangent and cotangent spaces.

4. The Basis of the Cotangent Space

We established that the tangent space $T_{p} M$ is an $n$ -dimensional vector space with a coordinate basis ${\partial_{μ} \equiv \frac{\partial}{\partial x ^{μ}}}$ . Since the dual of an $n$ -dimensional vector space is also an $n$ -dimensional vector space, $T_{p}^{*} M$ must also have a basis of $n$ elements.

This basis is constructed from the differentials of the coordinate functions themselves, ${x^{μ}}$ . The basis for the cotangent space $T_{p}^{*} M$ is ${(d x^{1})_{p}, \dots, (d x^{n})_{p}}$ .

These two bases, ${\partial_{μ}}$ for $T_{p} M$ and ${d x^{ν}}$ for $T_{p}^{*} M$ , are said to be dual to each other. This means their pairing gives the Kronecker delta. Let's compute the action of the basis one-form $d x^{μ}$ on the basis vector $\partial_{ν}$ :
$⟨ d x^{μ}, \partial_{ν} ⟩ \equiv (d x^{μ})_{p} (\partial_{ν}) = \partial_{ν} (x^{μ}) = \frac{\partial x ^{μ}}{\partial x ^{ν}} = δ_{ν}^{μ}$
This fundamental relationship is the cornerstone of calculations with components.

Just as any vector can be expanded in the vector basis, $X = X^{μ} \partial_{μ}$ , any one-form $ω$ can be expanded in the one-form basis:
$ω = ω_{μ} d x^{μ}$
The coefficients $ω_{μ}$ are the components of the one-form $ω$ . How do we find them? By acting with $ω$ on the basis vectors:
$ω (\partial_{ν}) = (ω_{μ} d x^{μ}) (\partial_{ν}) = ω_{μ} ⟨ d x^{μ}, \partial_{ν} ⟩ = ω_{μ} δ_{ν}^{μ} = ω_{ν}$
So, the component $ω_{ν}$ is simply the number you get when the one-form $ω$ acts on the $ν$ -th basis vector.

Visualizing the Action of a One-Form (dy) — A concrete example of the basis one-form `dy`. The horizontal lines are surfaces of constant `y`. The one-form `dy` acting on a vector gives the change in the `y` coordinate along that vector, which is precisely the number of lines pierced.

Di erential and Riemannian Geometry for General Relativity

The following text from 'Differential and Riemannian Geometry for General Relativity' provides the formal definitions for one-forms, the cotangent space basis, and the duality condition. It is concise and uses the notation standard in GR literature.

Please read Section 1.5, 'One-Forms' (pages 11-12). Focus on how an arbitrary one-form \omega is defined and decomposed in the basis \{dx^\mu\} and the explicit calculation of (\langle\omega|X angle = \omega_\mu X^\mu).

Test your understanding!

Let $ω$ be a one-form in $R^{2}$ with Cartesian coordinates $(x, y)$ . Its components are given by $ω_{x} = 2 y$ and $ω_{y} = - x$ , so $ω = 2 y d x - x d y$ .
Let $X$ be the vector field $X = x^{2} \partial_{x} + y \partial_{y}$ .

Calculate the value of $ω (X)$ at the point $p = (1, 3)$ .

Show answer

First, we write out the action of the one-form on the vector:
$ω (X) = (2 y d x - x d y) (x^{2} \partial_{x} + y \partial_{y})$
Using linearity, we can expand this:
$ω (X) = 2 y x^{2} ⟨ d x, \partial_{x} ⟩ + 2 y^{2} ⟨ d x, \partial_{y} ⟩ - x \cdot x^{2} ⟨ d y, \partial_{x} ⟩ - x \cdot y ⟨ d y, \partial_{y} ⟩$
Now we use the duality relation $⟨ d x^{i}, \partial_{j} ⟩ = δ_{j}^{i}$ :
$⟨ d x, \partial_{x} ⟩ = 1, ⟨ d x, \partial_{y} ⟩ = 0, ⟨ d y, \partial_{x} ⟩ = 0, ⟨ d y, \partial_{y} ⟩ = 1$
Substituting these in gives:
$ω (X) = 2 y x^{2} (1) + 0 - 0 - x y (1) = 2 x^{2} y - x y$
This is a scalar field, representing the value of $ω (X)$ at any point $(x, y)$ .
Finally, we evaluate this at the point $p = (1, 3)$ :
$ω_{p} (X_{p}) = 2 (1)^{2} (3) - (1) (3) = 6 - 3 = 3$
The result is a scalar, as expected.

5. One-Form Fields and the Total Differential

So far, we have mostly focused on a single point $p$ . A one-form field is a smooth assignment of a one-form from $T_{p}^{*} M$ to every point $p$ on the manifold. The differential $df$ of a global function $f$ is a prime example of a one-form field.

Let's find the components of the one-form field $df$ . The components, which we'll call $(df)_{μ}$ , are found by acting on the basis vectors $\partial_{μ}$ :
$(df)_{μ} = (df) (\partial_{μ}) = \partial_{μ} (f) = \frac{\partial f}{\partial x ^{μ}}$
Therefore, we can expand the one-form field $df$ in the basis ${d x^{μ}}$ as:
$df = \frac{\partial f}{\partial x ^{μ}} d x^{μ}$
You should recognize this immediately from multivariable calculus as the "total differential" of $f$ . What differential geometry provides is a profound reinterpretation: this is not just a formula for infinitesimal changes. It is the component expansion of a geometric object—the one-form field $df$ —in the coordinate basis ${d x^{μ}}$ .

The following video explores this connection in detail, showing how this familiar formula arises naturally from the geometric framework we've been building.

Tensor Calculus 7: Covector Field Components

This video, also by eigenchris, makes the crucial connection between the formal basis of one-forms \{dx^\mu\} and the familiar total differential formula from calculus. It will solidify your understanding of what the components of a one-form field represent.

Please watch the indicated segments: Basis Duality (01:23 - 04:54): This section establishes the duality relation dC^i(\partial_j) = \delta^i_j in a very visual way, showing how the basis one-forms \{dx, dy\} act on the basis vectors \{\partial_x, \partial_y\}. Deriving the Component Formula (05:37 - 07:50): Follow the elegant derivation of the formula df = (\partial f/\partial x)dx + (\partial f/\partial y)dy, starting from the chain rule. This is the rigorous justification for our reinterpretation of the total differential. Summary and Reinterpretation (12:10 - 14:07): This segment summarizes the key takeaway: the differentials we used in calculus are actually basis one-form fields, and the familiar formula is a component expansion.

Conclusion

In this lesson, we have introduced the duals to tangent vectors: one-forms.

Key Takeaways:

The cotangent space $T_{p}^{*} M$ is the vector space of all linear maps from the tangent space $T_{p} M$ to $R$ . Its elements are called one-forms or covectors.
A one-form $ω$ can be visualized as a stack of surfaces, and its action on a vector $X$ , $ω (X)$ , is the number of surfaces pierced by $X$ .
The differential of a function, $df$ , is a canonical one-form defined by its action on a vector $X$ : $(df) (X) = X (f)$ .
The differentials of the coordinate functions, ${d x^{μ}}$ , form a basis for the cotangent space. This basis is dual to the vector basis ${\partial_{ν}}$ , satisfying $⟨ d x^{μ}, \partial_{ν} ⟩ = δ_{ν}^{μ}$ .
Any one-form can be expanded as $ω = ω_{μ} d x^{μ}$ , and the familiar formula for the total differential, $df = (\partial f / \partial x^{μ}) d x^{μ}$ , is precisely the component expansion of the one-form $df$ .

We now have two types of fundamental objects: vectors (elements of $T_{p} M$ ) and covectors (elements of $T_{p}^{*} M$ ). These are, in fact, the simplest types of tensors. A vector is a tensor of type (1,0) and a one-form is a tensor of type (0,1). In our next lesson, we will see how to combine them to build the entire hierarchy of general (r,s) tensors, which are the mathematical language of General Relativity.

Recommended resources