The Riemann Curvature Tensor

Jennifer Cox

May 6, 2019

Project Advisor: Dr. Jonathan Walters

Abstract

A tensor is a mathematical object that has applications in areas including physics,

psychology, and artificial intelligence. The Riemann curvature tensor is a tool used
to describe the curvature of n-dimensional spaces such as Riemannian manifolds in
the field of differential geometry. The Riemann tensor plays an important role in
the theories of general relativity and gravity as well as the curvature of spacetime.
This paper will provide an overview of tensors and tensor operations. In particular,
properties of the Riemann tensor will be examined. Calculations of the Riemann tensor
for several two and three dimensional surfaces such as that of the sphere and torus will
be demonstrated. The relationship between the Riemann tensor for the 2-sphere and
3-sphere will be studied, and it will be shown that these tensors satisfy the general
equation of the Riemann tensor for an n-dimensional sphere. The connection between
the Gaussian curvature and the Riemann curvature tensor will also be shown using
Gauss’s Theorem Egregium.

Keywords: tensor, tensors, Riemann tensor, Riemann curvature tensor, curvature

1 Introduction

Coordinate systems are the basis of analytic geometry and are necessary to solve geomet-
ric problems using algebraic methods. The introduction of coordinate systems allowed for
the blending of algebraic and geometric methods that eventually led to the development of
calculus. Reliance on coordinate systems, however, can result in a loss of geometric insight
and an unnecessary increase in the complexity of relevant expressions. Tensor calculus is
an effective framework that will avoid the cons of relying on coordinate systems. Tensor

1
The Riemann Curvature Tensor 2
calculus acknowledges the need for coordinate systems but avoids implementing a coordi-
nate system until needed. This allows for equations that are valid in all coordinate systems
simultaneously [4].

Curvature describes how a geometric object such as a curve deviates from a straight line or
a surface from a flat plane. Curvature can be expressed simply as a scalar that represents
the magnitude of this deviation. Curvature can also be described as a vector that takes
into account the direction of the curve along with the magnitude. For more complex objects
such as surfaces or n-dimensional spaces, a more complex object is needed to describe the
curvature [3]. One such object, which will be the primary focus of this paper, is the Riemann
curvature tensor.

This paper will explore the use of tensor calculus in describing the curvature of manifolds.
Prior to this, the underlying ideas and definitions related to tensors will be discussed. The
focus of this paper is the calculations of the Riemann tensor for the surfaces of the 2-sphere,
2-torus, and 3-sphere. Results of these calculations are provided in the body of the paper
and more detailed calculations are provided in an appendix.

2 Background

Consider a physical quantity such as a vector. A vector can be described in terms of its mag-
nitude and direction and therefore does not require the application of a coordinate system.
If we want to perform calculus with vectors, we need to be able to describe their components
in terms of the surrounding space, which requires the use of a coordinate system. Because
this choice of coordinate system is arbitrary, the vector is not dependent upon this choice.
Once a coordinate system has been chosen, we can describe the vector in terms of the basis
vectors of the chosen system. Under a change of coordinates, this description would differ
although the vector itself is unchanged. Is there a way to describe the vector without using
a particular coordinate system? This is the aim of tensor calculus. In order to construct
expressions independent of the choice of coordinates, we must determine how individual el-
ements change under a change of coordinates [4].

A variant is an object that can be calculated using a similar rule in different coordinate
systems. As expected, the results vary when the same rule is applied to different coordinate
systems. Tensors are a special type of variant that transform according to certain rules under
a change of coordinates [8]. A tensor is a multi-dimensional array of mathematical objects,
generally numbers or functions. Tensors are classified based on their number and type of
components. The components of a tensor may be covariant or contravariant. Covariance and
contravariance describe how quantities change with a change of basis. Covariant components
change in the same way as changes to the basis vectors and are denoted with lower indices ai .
Contravariant components change inversely to changes in the basis vectors and are denoted
The Riemann Curvature Tensor 3
i
with upper indices a [9].

The rank of a tensor is the total number of covariant and contravariant components. Tensors
of rank 0 are scalars, tensors of rank 1 are vectors, and tensors of rank 2 are matrices. For
example, the metric tensor, which has rank two, is a matrix. Higher-order tensors are multi-
dimensional arrays. A visualization of a rank 3 tensor from [3] is shown in figure 1 below.

Figure 1: Visualization of a rank 3 tensor.

Intrinsic objects are those which can be obtained by measuring distances and computing the
derivatives of those distances in some space. The metric tensor is an example of an intrinsic
object. The metric tensor describes how to compute distances and lengths of curves in a
given space [7]. In three dimensional Euclidean space, the distance ds between two points
whose cartesian coordinates are (x, y, z) and (x + dx, y + dy, z + dz) is given by
ds2 = dx2 + dy 2 + dz 2 . (1)
Riemann generalized this idea and extended it to spaces of n-dimensions. Consider two
points whose coordinates in any system are xi and xi + dxi where i ranges from 1 to n. He
defined the infinitesimal distance between the two points to be
ds2 = gij dxi dxj (2)
where the coefficients gij are functions of the coordinates xi . These coefficients collectively
make up the metric tensor gij [10]. For example, the metric tensor for three dimensional
Euclidean space in cartesian coordinates is the collection of the coefficients from equation 1.
The metric tensor is therefore the 3x3 identity matrix.
 
1 0 0
gij = 0 1 0 (3)
0 0 1
The Riemann Curvature Tensor 4

Because the metric tensor is an intrinsic object, subsequent objects that can be described
in terms of the metric tensor and its derivatives are also intrinsic. One object that can be
derived from the metric tensor is the Christoffel symbol. The Christoffel symbol describes
the variation in basis vectors from one point to another in curvilinear coordinate systems.
The Christoffel symbols are the n3 partial derivatives of each basis vector differentiated
with respect to each coordinate.The Christoffel symbols measure the rate of change of the
covariant basis with respect to the coordinate variables [1]. Christoffel symbols of the first
kind (three lower indices) can be found directly from the metric tensor using the following
formula
1
Γijk = (−gijk + gjki + gkij ) (4)
2
where the partial derivative of a tensor with respect to xk is indicated by a final subscript k
[5]. Christoffel symbols of the second kind (one upper and two lower indices) are obtained
by taking the inner product (defined in methods section below) of the Christoffel symbols of
the first kind with the metric tensor
Γijk = g ir Γjkr (5)
where g ir is the contravariant metric tensor.

In general, partial derivatives of the components of a vector or a tensor are not components
of a tensor. This is the case for Christoffel symbols which are partial derivatives of the
metric tensor but are not tensors themselves. Because tensor properties, namely invariance,
are desirable, a new differential operator ∇i , the covariant derivative, arises. The covari-
ant derivative produces tensors from tensors. The resulting tensors are one covariant order
greater than the original tensor[4]. In affine coordinates, the covariant basis is the same at
all points. Subsequently, the covariant derivative is commutative. However, this is not the
case for curved surfaces.

The non-commutativity of the covariant surface derivative is measured with the Riemann
tensor. The Riemann tensor is a four-index tensor that provides an intrinsic way of describing
the curvature of a surface. The Riemann tensor of the second kind can be represented
independently from the formula

i
∂Γijm ∂Γijk
Rjkm = − + Γirk Γrjm − Γirm Γrjk (6)
∂xk ∂xm
The Riemann tensor of the first kind is represented similarly, using Christoffel symbols of
the first kind
∂Γjmi ∂Γjki
Rijkm = − + Γimr Γrjk − Γikr Γrjm (7)
∂xk ∂xm
Alternatively, the Riemann tensor of the first kind can be obtained by lowering the con-
travariant index. This is done by taking the inner product of the Riemann tensor of the
second kind with the metric tensor as follows:
r
Rijkm = gir Rjkm (8)
The Riemann Curvature Tensor 5
For a sphere of radius r, components of the Riemann tensor of the first kind can be calculated
directly from the metric, without first calculating the Christoffel symbols, using the following
equation from [10].
1
Rijkm = 2 (gik gjm − gim gjk ) (9)
r
Tensor calculations are significantly simplified when symmetric properties of tensors are used.
The Riemann tensor has several symmetric properties that will help to simplify calculations

first skew symmetry Rijkl = −Rjikl

second skew symmetry Rijkl = −Rijlk (10)
block symmetry Rijkl = Rklij
In general, the number of independent components C of the Riemann tensor in n-dimensions
is given by the equation
1
C = n2 (n2 − 1) (11)
12
from [5].

A manifold is a space that can be curved but is locally flat. That is, near each point the space
resembles Euclidean space. One example of a curved manifold is the surface of a sphere. The
Riemann tensor is the most common tool used to describe the curvature of a Riemannian
manifold. Contraction, a tensor operation defined later in the paper, of the Riemann tensor
produces the Ricci tensor. The Ricci tensor provides a way measure the degree to which
a space differs from Euclidean space. Contraction of the Ricci tensor produces the scalar
curvature or Ricci scalar. The Ricci scalar is the simplest curvature invariant of a manifold.[4]

The Riemann tensor, Ricci tensor, and Ricci scalar are all derived from the metric tensor
and are therefore intrinsic measures of curvature. There is another tensor, simply called the
curvature tensor, that can be used to express the curvature of a surface. This tensor depends
on the way in which the surface is embedded in the surrounding space. The curvature tensor
is therefore an extrinsic object because it is not innate to the surface itself. Gauss’s Theorem
Egregium relates the Riemann tensor and the curvature tensor. This is a powerful result
that links the intrinsic and extrinsic perspectives of curvature [4].

3 Methods

3.1 Summation Notation

Computations with tensors often involve the use of Einstein summation notation. This
notation uses repeated indices rather than the traditional sigma notation to represent a
sum:
ai xi = a1 x1 + a2 x2 + ... + an xn (12)
The Riemann Curvature Tensor 6
where 1 ≤ i ≤ n is the range for summation. Therefore the expressions aii xk and aik xk
represent a summation while aij xk does not. Because the choice of summation index is
arbitrary (aik xk and aim xm represent the same summation) this index is called a dummy
index. The non-repeated index has the same range as the dummy index and is called a free
index.

3.2 Tensor Contraction and Inner Product

Another operation that can be performed on tensors is contraction. The contraction of a

tensor is produced by summing over a pair of indices and is a generalization of the trace oper-
ation for matrices. Contraction can occur in a single mixed tensor or on a linear combination
or product of tensors. Contraction on a single tensor reduces the rank of a tensor by order
2. Contraction can be repeated until all indices are gone, resulting in a tensor of order 0,
which is an invariant. This is a powerful result because it allows for forming invariants using
linear combinations, products, and contraction of tensors. [6] The inner product between
two tensors is the contraction of the inner indices (the last index of the first tensor and the
first index of the last tensor). Essentially, the inner product is equating an upper index of
one tensor with a lower index of another tensor and summing over the repeated index [2].

3.3 Raising and Lowering of Indices

The tensor type can be changed by taking the inner product with the metric tensor. Covari-
ant tensors can be converted to contravariant tensors and vice versa using the covariant or
contravariant metric tensor. This process is referred to as the raising and lowering of indices.
This process does not change the rank of a tensor, but does change the type. In Cartesian
coordinates the metric is the identity matrix and therefore raising and lowering indices does
not change the value of the tensor components[6].

3.4 Metric Tensors

To calculate the Riemann tensor, we need only the metric tensor for the surface. The metric
tensors for the surfaces of the 2-sphere, 2-torus, and 3-sphere are given below.
2-Sphere
θ = 1, φ = 2
 2 
r 0
gij = (13)
0 r2 sin2 θ
2-Torus
θ = 1, φ = 2
The Riemann Curvature Tensor 7

 
(R + r cos φ)2 0
gij = (14)
0 r2

3-Sphere
ψ = 1, θ = 2, φ = 3

 2 
r 0 0
gij =  0 r2 sin2 ψ 0  (15)
2 2 2
0 0 r sin ψ sin θ

4 Results

4.1 2-Sphere

Riemann Curvature Tensor

First Kind Second Kind

R1212 = r2 sin2 θ 1
R212 = sin2 θ

R1221 = −r2 sin2 θ R221

1
= − sin2 θ
(16)
R2121 = r2 sin2 θ 2
R121 =1

R2112 = −r2 sin2 θ R112

2
= −1

Ricci Tensor
 
1 0
Rij = (17)
0 sin2 θ

Scalar Curvature
2
S= (18)
r2
The Riemann Curvature Tensor 8
4.2 2-Torus

Riemann Curvature Tensor

First Kind Second Kind

1 r cos φ
R1212 = r cos φ(R + r cos φ) R212 = R+r cos φ

1 r cos φ
R1221 = −r cos φ(R + r cos φ) R221 = − R+r cos φ (19)
R2121 = r cos φ(R + r cos φ) 2
R121 = 1r cos φ(R + r cos φ)
2
R2112 = −r cos φ(R + r cos φ) R112 = − 1r cos φ(R + r cos φ)

Ricci Tensor 1 
cos φ(R + r cos φ) 0
Rij = r r cos φ (20)
0 R+r cos φ

Scalar Curvature
2 cos φ
S= (21)
r(R + r cos φ)

4.3 3-Sphere

Riemann Curvature Tensor

First Kind Second Kind

R1212 = r2 sin2 ψ 1
R212 = sin2 ψ
(22)
R1313 = r2 sin2 ψ sin2 θ R313
1
= sin2 ψ sin2 θ

R2323 = r2 sin2 ψ sin2 θ R323

2
= sin2 ψ sin2 θ

Ricci Tensor  
2 0 0
Rij = 0 2 sin2 ψ 0  (23)
0 0 2 sin2 ψ sin2 θ

Scalar Curvature
6
S= (24)
r2
The Riemann Curvature Tensor 9
5 Conclusion

Tensors are intriguing mathematical objects that possess the desirable property of invariance.
The metric tensor is of particular importance, allowing for the concept of distances and
maintaing the invariance of distance in different coordinate systems. Calculations involving
tensors are systematic and are simplified by the use of summation notation. Tensors and
tensor calculus provide a powerful framework for solving problems in areas of physics such
as general relativity. Representing fundamental physical laws in a tensor form ensures that
they are invariant under a change of coordinates. Complex objects like tensors are needed
to represent characteristics, such as curvature, of higher-dimensional spaces. The Riemann
tensor is one such object used to express the curvature of manifolds.

6 Appendix A: Calculations

This appendix contains calculations of the Christoffel symbols, Riemann tensor, Ricci tensor,
and Ricci scalar for the surface of a 2-sphere. Quantities that evaluate to zero have been
excluded for conciseness. The calculations of these objects for the surfaces of the 2-torus and
3-sphere are similar and use each surface’s corresponding metric tensor as listed in section
3.4.
Christoffel Symbols of the Second Kind

g ir
Γijk = (gijk − gjki + gkij )
2

g 11
Γ122 = (g122 − g221 + g212 )
2

1
Γ122 = (0 − 2r2 sin θ cos θ + 0)
2r2

Γ122 = − sin θ cos θ (25)

g 22
Γ212 = Γ221 = (−g212 + g122 + g221 )
2

1
Γ212 = Γ221 = 2
2 (−0 + 0 + 2r sin θ cos θ)
2r2 sin θ

cos θ
Γ212 = Γ221 =
sin θ
The Riemann Curvature Tensor 10
Riemann Tensor of the Second Kind

i
∂Γijm ∂Γijk
Rjkm = − + Γink Γnjm − Γinm Γnjk
∂xk ∂xm

1 ∂Γ122 ∂Γ121
R212 = − + [Γ111 Γ122 + Γ121 Γ222 ] − [Γ112 Γ121 + Γ122 Γ221 ]
∂x1 ∂x2 (26)

1
R212 = (sin2 θ − cos2 θ) − (− cos2 θ)

1
R212 = sin2 θ

Riemann Tensor of the First Kind (using eq. 9)

1
Rijkm = (gik gjm − gim gjk )
r2

1
R1212 = (g11 g22 − g12 g21 )
r2
(27)
1
R1212 = 2 (r2 (r2 sin2 θ) − 0)
r

R1212 = r2 sin2 θ

Ricci Tensor
Rhk = g ab Rahbk

R11 = g 11 R1111 + g 22 R2121

1 2 2
R11 = 2 2 (R sin θ)
R sin θ

R11 = 1 (28)

R22 = g 11 R1212 + g 22 R2222

1
R22 = 2
(R2 sin2 θ)
R

R22 = sin2 θ
The Riemann Curvature Tensor 11
Ricci Scalar
S = g ij Rij

S = g 11 R11 + g 22 R22

1 1
S= 2
(1) + 2 2 (sin2 θ)
R R sin θ (29)

1 1
S= 2
+ 2
R R

2
S=
R2

References
[1] Edmund Bertschinger. Introduction to tensor calculus for general relativity, 1999.

[2] David A. Clarke. A primer on tensor calculus, 2011.

[3] Kees Dullemond and Kasper Peeters. Introduction to tensor calculus, 1991.

[4] Pavel Grinfeld. Introduction to Tensor Analysis and the Calculus of Moving Surfaces.
Springer, 2013.

[5] David C. Kay. Schaum’s Outline of Theory and Problems of Tensor Calculus. McGraw-
Hill, 1988.

[6] Joseph C. Kolecki. An introduction to tensors for students of physics and engineering,
2002.

[7] Wha-Suck Lee, Johann C. Engelbrecht, and Rita Moller. Tensor calculus: unlearning
vector calculus. International Journal of Mathematical Education, 2017.

[8] A. V. Smirnov. Introduction to tensor calculus, 2004.

[9] Taha Sochi. Introduction to tensor calculus, 2016.

[10] C. E. Weatherburn. An Introduction to Riemannian Geometry and the Tensor Calculus.

Cambridge University Press, 1938.