MTH3008 Lecture 16

Quote

Last time we began Chapter 7 on tensor fields, setting up vector fields, local bases, and the machinery of covariant differentiation - deriving the differential of a vector in Cartesian, fixed general, and local basis coordinate systems. This lecture picks up where we left off, introducing Christoffel symbols to package the geometry of the local basis, then deriving their connection to the metric tensor and extending covariant differentiation to tensors of arbitrary rank.

Covariant Derivatives - Recap

Recall from Lecture 15 that the Covariant Differentiation of the covariant and contravariant components of a vector are

Each expression splits into a partial derivative of the component plus a correction term that accounts for how the basis vectors themselves change from point to point. Those correction terms are crying out for a name.

Christoffel Symbols

Definition - Second Kind

We define the Christoffel Symbols by

These are the contravariant components of expanded in the basis . In other words, are the Christoffel Symbols of the rate of change of the basis vectors:

This follows directly from the definition: dotting both sides of the expansion with picks out via the reciprocity relation .

Definition - First Kind

We can equally expand with respect to the dual basis . The coefficients of that expansion are the Christoffel Symbols :

so that

Note

The two kinds differ only in which basis is used for the dot product: (upper index, second kind) versus (lower index, first kind). The vector being differentiated is always .

Covariant Derivatives in Terms of Christoffel Symbols

Substituting the definitions into the covariant derivative expressions gives the clean forms

Warning

Watch the signs and index placement carefully. The covariant component picks up a minus sign with , while the contravariant component picks up a plus sign with . The free index on the Christoffel symbol matches the position (up or down) of the component being differentiated.

Fixed Basis as a Sanity Check

If the basis is fixed (constant everywhere), then and all Christoffel symbols vanish:

The covariant derivatives then collapse to ordinary partial derivatives:

This is exactly what we expect - in a Cartesian system, differentiation of components is just partial differentiation.

Christoffel Symbols and the Metric Tensor

Raising and Lowering with the Metric

The two kinds of Christoffel symbol are related by the metric tensor in the same way that covariant and contravariant components of any object are related - by raising or lowering indices:

This is a direct consequence of the definitions and the metric relations and .

Deriving the First Kind from the Metric Tensor

The Christoffel symbols of the first kind enjoy a symmetry in their last two indices: . This follows because

and the local basis vectors satisfy , so , giving .

Using this symmetry, we can write and expand via the definition:

The trick is to use the product rule to introduce metric tensor components. Since , we have

so each dot product can be rewritten as . Applying this systematically:

The two leftover dot products recombine via the product rule as , yielding the final result:

This is a major result: the Christoffel symbols of the first kind are determined entirely by the metric tensor and its partial derivatives.

Formula for the Second Kind

Combining the first-kind formula with the index-raising relation gives

So both kinds of Christoffel symbol are fully determined by the metric. No additional geometric data is needed.

Worked Example - Computing Christoffel Symbols

Example

Suppose a coordinate system has metric tensor components , , , and for . Find all Christoffel symbols of both kinds.

Step 1 - Identify Non-Zero Derivatives

All are constant except (depends on ) and (depends on and ). So

The non-zero derivatives are:

Step 2 - First Kind Christoffel Symbols

The only possibly non-zero are those whose indices are some permutation of , , or . Computing each:

By symmetry in the last two indices, .

All other .

Step 3 - Second Kind Christoffel Symbols

Use with (since the metric is diagonal):

All other .

Note

The task at the end of the slides asks you to repeat this procedure for spherical coordinates with , , - see 8.2. Christoffel Symbols in Spherical Coordinates.

Covariant Differentiation of Tensors

The covariant derivative machinery extends naturally from vectors to tensors of any rank. Recall the vector formulas:

The pattern is: each covariant (lower) index generates a term, and each contravariant (upper) index generates a term.

Second-Rank Tensors

For a Covariant and Contravariant Components of rank 2:

Each lower index contributes its own correction, with the dummy index contracting against the corresponding slot.

For a Covariant and Contravariant Components of rank 2:

Each upper index contributes a correction.

For a Mixed Components of rank 2:

The upper index gives and the lower index gives .

Warning

When writing covariant derivatives of mixed tensors, the dot in matters - it marks which slot is up and which is down. Without it, is ambiguous between and , which are generally different tensors.

Important

The rule generalises to any rank: take the partial derivative, then add one Christoffel correction per index - for each upper index, for each lower index. Each correction replaces that index with a dummy and contracts the Christoffel symbol appropriately. These covariant derivatives transform as tensors (proof left as an exercise in the lecture slides).


Pre-Lecture Notes from mth3008 lecture notes 16.pdf

  • Christoffel symbols of the second kind: - expansion coefficients of in the basis
  • Christoffel symbols of the first kind: - expansion coefficients of in the dual basis
  • Related via the metric: and
  • For a fixed basis, all Christoffel symbols vanish and covariant derivatives reduce to partials
  • First kind formula from the metric:
  • Second kind formula:
  • Worked example computing all Christoffel symbols for a diagonal metric with ,
  • Covariant differentiation of rank-2 tensors: each lower index gives , each upper index gives
  • Next lecture: Ricci’s theorem