📐 Math · Differential Geometry · Manifolds Ep. XX October 11, 2026

Contravariance and Invariance — Why Components Transform Opposite to the Basis

The most fundamental reason is just one:

The tangent vector $v$ is itself a geometric object, independent of any coordinate system, so it must not change.

When we change coordinates, the basis $\partial_i$ changes and the components $v^i$ change too. But they change in opposite ways, so that the final result $v=v^i\partial_i$ stays exactly the same.

1. The Requirement of Invariance

The tangent vector $v$ is an arrow that does not depend on the coordinate system. Therefore:

$$v = v^i\partial_i = \tilde{v}^i\tilde{\partial}_i$$

This equality must hold no matter which coordinates we use. It is the starting point for everything.

2. How the Basis Transforms (Chain Rule)

Given the relation between the new coordinates $\tilde{x}^i$ and the old coordinates $x^j$, the chain rule gives:

$$\tilde{\partial}_i = \frac{\partial}{\partial \tilde{x}^i} = \frac{\partial x^j}{\partial \tilde{x}^i}\,\frac{\partial}{\partial x^j}$$

That is:

Basis transformation$$\tilde{\partial}_i = \frac{\partial x^j}{\partial \tilde{x}^i}\,\partial_j$$

Example: the basis doubles. Change the coordinate to $\tilde{x}=x/2$:

$$\tilde{\partial} = \frac{\partial}{\partial \tilde{x}} = \frac{\partial x}{\partial \tilde{x}}\,\frac{\partial}{\partial x} = 2\partial$$

The new basis is twice as large.

3. How the Components Transform (Contravariance)

Express the same vector $v$ in the new basis:

$$v = \tilde{v}^i\tilde{\partial}_i = \tilde{v}^i\,\frac{\partial x^j}{\partial \tilde{x}^i}\,\partial_j$$

This must equal $v=v^j\partial_j$, so:

$$v^j = \tilde{v}^i\,\frac{\partial x^j}{\partial \tilde{x}^i}$$

Multiply both sides by $\dfrac{\partial \tilde{x}^k}{\partial x^j}$ and sum (using the inverse-matrix relation):

Component transformation$$\tilde{v}^i = \frac{\partial \tilde{x}^i}{\partial x^j}\,v^j$$

Example: the basis doubles.

$$\tilde{v} = \frac{\partial \tilde{x}}{\partial x}\,v = \frac{1}{2}\,v$$

The component is cut in half.

4. The Product Is Invariant

The basis doubles and the component is multiplied by $1/2$:

$$\tilde{v}\,\tilde{\partial} = \left(\tfrac{1}{2}v\right)(2\partial) = v\,\partial$$

We get back exactly what we started with. This is the key.

5. Why Must They Change in Opposite Ways? An Intuition

Imagine climbing a staircase to a fixed height.

  • Basis = the height of one step ($\partial_i$ in the old coordinates, $\tilde{\partial}_i$ in the new)
  • Components = the number of steps ($v^i$ in the old coordinates, $\tilde{v}^i$ in the new)
  • Vector = the total height climbed ($v = v^i\partial_i = \tilde{v}^i\tilde{\partial}_i$, the same in both)

Suppose you climb a total of 3 m. With steps of 10 cm, you need 30 steps. Double the step height to 20 cm, and you need only 15 steps.

  • The basis (step height) doubled, and the component (number of steps) became $1/2$ of what it was.
  • The total height climbed (the vector) is still 300 cm.

When the steps get taller, you take fewer of them, and when the steps get shorter, you take more. The two must move in opposite directions for the total height to be preserved. This is the same structure as our earlier example, $\tilde{\partial}=2\partial$ and $\tilde{v}=\tfrac12 v$.

6. Seeing It with Matrices

$$\tilde{\partial}_i = A_i{}^{j}\,\partial_j, \qquad A_i{}^{j} = \frac{\partial x^j}{\partial \tilde{x}^i}$$
$$\tilde{v}^i = B^i{}_{j}\,v^j, \qquad B^i{}_{j} = \frac{\partial \tilde{x}^i}{\partial x^j}$$

Here $B=A^{-1}$. In other words:

The basis transforms by $A$, and the components transform by $A^{-1}$.

This is where the name “contravariant” comes from. The components change contra (opposite) to the change of basis.

7. What If They Changed in the Same Way?

Suppose the components also transformed by $A$, like the basis:

$$\tilde{v}^i\tilde{\partial}_i = (Av)^i(A\partial)_i = A^2\,(v^i\partial_i)$$

The vector would be multiplied by $A^2$. That would be a strange theory in which the size of a vector changes every time we change coordinates.

To keep the vector invariant, one must transform by $A$ and the other by $A^{-1}$. That is why the basis and the components change in opposite ways.

8. In the Dual Space (Covectors), It Is Reversed

For a covector $\alpha=\alpha_i\,dx^i$:

  • The dual basis $dx^i$ transforms by $A^{-1}$.
  • The components $\alpha_i$ transform by $A$.

So vectors and covectors obey exactly opposite transformation laws. This is what makes

$$v^i\alpha_i = \text{invariant (scalar)}$$

possible. One has an upper index (contravariant), and the other has a lower index (covariant).

9. Summary

ObjectTransformation lawReason
Basis $\partial_i$$\dfrac{\partial x^j}{\partial \tilde{x}^i}$ (chain rule)Definition of the coordinate derivative
Components $v^i$$\dfrac{\partial \tilde{x}^i}{\partial x^j}$ (contravariant)Preserves the invariance of the vector
Their productInvariant$A\cdot A^{-1}=I$

10. One-Line Summary

The basis and the components change in opposite ways for a single reason: the tangent vector $v=v^i\partial_i$ must be independent of the coordinate system. Just as taller steps mean fewer steps, the basis and the components must move in opposite directions for their product, the actual vector, to be preserved.

11. A Hint of What Comes Next

We have seen that the basis $\partial_i$ changes when we change coordinates. But the basis also varies from point to point within a single coordinate system. For example, in polar coordinates, $\partial_\theta$ points in a different direction at every point and gets longer as $r$ increases. In other words, the tangent space is a different space at each point.

So how can we compare vectors at different points, or differentiate a vector field?

Coming up. We will answer this in Foundations 5, with Christoffel symbols. First, we build a few tools: index notation, the basis of the tangent space, and covectors.
manifolds metric tensor covariant & contravariant general relativity