How Einstein Got His Field Equations

S. Walters

In commemoration of General Relativitys centennial

arXiv:1608.05752v1 [physics.hist-ph] 19 Aug 2016

A BSTRACT. We study the pages in Albert Einsteins 1916 landmark paper in the Annalen der
Physik where he derived his field equations for gravity. Einstein made two heuristic and
physically insightful steps. The first was to obtain the field equations in vacuum in a rather
geometric fashion. The second step was obtaining the field equations in the presence of matter
from the field equations in vacuum. (This transition is an essential principle in physics,
much as the principle of local gauge invariance in quantum field theory.) To this end, we go
over some quick differential geometric background related to curvilinear coordinates, vectors,
tensors, metric tensor, Christoffel symbols, Riemann curvature tensor, Ricci tensor, and see
how Einstein used geometry to model gravity.

This paper is a more detailed version of my talk given at the Math-Physics

Symposium at UNBC on February 25, 2016. It is in reference to Einsteins
paper:
A. Einstein, The Foundation of the General Theory of Relativity, Annalen
der Physik, 49, 1916. (For an English translation see: H. A. Lorentz, A.
Einstein, H. Minkowski, H. Weyl, The Principle of Relativity.)
The paper has two sections. The first section is a smash course on the
semi-Riemannian geometry tools needed to understand Einsteins theory.
The second section looks at Einsteins derivation of his field equations in vac-
uum and in the presence of matter and/or electromagnetism as he worked
them out in his paper.
This paper commemorates the 100th centennial of Einsteins General The-
ory of Relativity, which he finalized near the end of November 1915 and pub-
lished in 1916.

Date: Jan./Feb. 2016 (last update July 2016) LATEX File: EinsteinRelativityLectureFeb2016.tex.
2000 Mathematics Subject Classification. 35Q76 83C05 35Q75 37N20 83-01 83-XX.
Key words and phrases. Einstein field equations, tensors, relativity, gravity, curvature, spacetime.
1
2 S. Walters
1. Semi-Riemannian
Geometry
In this section we outline briefly some of
the basic and standard concepts known from
semi-Riemannian geometry that are used in
Einsteins formulation of his theory of grav-
ity.
Spacetime Curvilinear coordinates.
Spacetime coordinates are written using
superscripts simply as x where = 1, 2, 3, 4.
This is short for vectors (x1 , x2 , x3 , x4 ) describ-
ing points in spacetime with respect to a cer-
tain coordinate system (cartesian, cylindri-
cal, spherical, etc). The coordinates x1 , x2 , x3
refer to spatial coordinates and x4 to the time
coordinate. Any other coordinate system can
be written as x0 . (An observer and the coor-
dinate system used by him/her will be iden-
tified.)
Einsteins Summation Convention.
Any time an upper index is repeated as a
lower index we are automatically summing
over that index. For example, we write
4 4
Examples.
V U V U , A A .
=1 =1
(In such cases one drops the notation.)
A vector field V is a differentiable func-
tion defined on a certain region of space time
whose values are tangent vectors to space-
time (much as vectors one the sphere that
are tangent to it at each point of a certain
region on it). There is a natural way to de-
scribe V by means of its components V rela-
tive to a given coordinate system x . (Essen-
tially, each coordinate axis x has a vector
along it similar to the usual i, j, k vectors in
3-space R3 , as illustrated in Figure 1.)
The components of such vector fields are
written with superscripts and called contra
variant because their components in differ-
ent coordinate systems V relative to x
and V 0 relative to x0 transform according
to the rule
x0
V 0 =V . (1.1)
x
Covariant vector fields are written using
subscripts V and they transform in a dual
manner via
x
V0 = 0 V . (1.2)
x
Figure 1. At each point p in space time
one has natural vectors e1 , e2 that are tan-
gent to the curved coordinate axes, quite
analogous to the i, j, k vectors in calculus.
(1) If a test particle is moving in space and
its spacetime coordinates are expressed in
terms of its proper time , i.e. x = x (),
then its velocity 4-vector is defined by
dx
V = .
d
Defining a particles velocity in this way is
quite convenient since it gives us a covari-
ant way of describing velocities. This is so
simply because from the chain rule from cal-
culus we see that the velocity 4-vector trans-
forms just like a contravariant vector field:
dx0 x0 dx
= .
d x d
(2) As a second example, consider a scalar
field = (x ) = (x0 ), and look at its gradi-
ent vector field with components V = x .
Given any other coordinate system x0 , the
 How Einstein Got His Field Equations 3

chain rule tells us hows the gradient com- by 4 symmetric matrix with real entries) be-
ponents are related: ing + ++ means that it has one one neg-
x ative eigenvalue and 3 positive eigenvalues.
= 0 . (The only negative eigenvalue is related to
x0 x x
the time coordinate.)
This is exactly how covariant vectors trans- In Relativity we are interested in semi- Rie-
form as in equation (1.2). mannian (or Lorentzian) metric tensors ex-
A tensor (field) is like a vector (field) except emplified by the Minkowski metric
that it can have two or more index compo-
1 0 0 0

nents, e.g. 0 1 0 0
=
0 0 1 0
(1.3)
A B C R .
0 0 0 1
The number of indices is called the rank of
the tensor. (So vectors have rank 1, scalar (used in Special Relativity) and its associ-

fields have rank 0, and C has rank 3.) ated element of proper spacetime interval
The way tensor components transform be-
ds2 = dx dx = d 2
tween different coordinate systems is just
like equation (1.1) for each upper index, and where d 2 is called the proper time (by anal-
like equation (1.2) for each lower (covariant) ogy with element of arc length in Riemann-

index. Thus, for the rank 2 tensor A , its ian geometry).
components in another coordinate system In General Relativity, the spacetime inter-
A0
are related by val extends to any coordinate system using
the metric tensor components g according
0 x0 x to
A = 0 A .
x x ds2 = g dx dx = d 2 .
It is for such elegant transformation prop-
The essential property of proper time is
erties of tensors that Einsteins principle of
that it is an invariant for all coordinate sys-
general covariance (discussed in the next sec-
tems:
tion) requires that the laws of physics be
expressed in covariant form using tensors. d 2 = d 0 2 .
Once such a law holds in one coordinate All observers (coordinate systems) will agree
system the same form of the law holds in on this quantity. (Again, this follows by the
any other coordinate system. chain rule.)
The most important rank 2 tensor is the Since the metric tensor g is an invertible
metric tensor g because it contains the matrix, its inverse is written in superscript
essential ingredients of gravity, distance, form g . Thus using the usual rules of ma-
time, and curvature of spacetime. It has the trix multiplication one has
essential property that it is symmetric
g g =
g = g .

where is the Kronecker delta function (which
One can have either a Riemannian or a semi-
is 1 when = and is 0 when 6= ).
Riemannian metric tensor depending on its
signature being + + ++ or + ++, respec- Both forms of the metric tensor are used
tively. The signature of g (treated as a 4 to raise and lower indices on tensors. For
4 S. Walters

example, Einstein regarded the Christoffel symbols

as the components of the gravitational
A = g A , g R = R .
field. In his paper, he also referred to the
The contraction of a mixed tensor is ob- metric tensor g as describing the grav-
tained by setting an upper index and a lower itational field in relation to the chosen
index equal and then summing. For instance, system of coordinates. So, in a manner

if one contracts the rank 4 tensor R with of speaking, gravity affects how we take the
respect to the indices and , one gets the derivative!2
rank 2 tensor
A test particle moving freely in vacuum
4
under the influence of the geometry of space
S = R R
=1
(i.e., under gravity) obeys the geodesic equa-
tion
where the index now disappeared.1 d 2 x
dx dx

+ =0
d 2 d d
Covariant Derivative. Since spacetime is
(where is the particles proper time). This
generally curved, the notion of the deriva-
equation describes the straight lines
tive from calculus becomes curved also. It
(geodesics) of spacetime. Thus a ray of light
becomes what we call the covariant deriva-
or photons moving freely in (vacuum) space-
tive. The covariant derivative of a covariant
time will follow a path that is the analogue of
vector field A is given by
straight lines (or great circles on a sphere).
A For a slowly moving particle in low grav-
A; = A
x
ity, this equation essentially comes down to
where Newtons second law F = ma. If 0 (zero
2

1 g g g
 gravity) the geodesic equation becomes ddx2 =

= g + 0, i.e. zero acceleration or constant velocity,
2 x x x
so it moves along a straight line (i.e., were
is called the Christoffel symbol. The covari- in flat spacetime).
ant derivative of a contravariant vector field
is quite similar except with a + in place of The Riemann-Christoffel Curvature Ten-
the ; thus sor is the rank 4 tensor

A R = + .
A ; = + A . x x
x
It governs all aspect of the curvature of space-
For example, if the metric tensor g is time. If it is zero everywhere then our space-
constant in some coordinate system, then time is flat. This looks like a fairly com-
s are all 0 and so the covariant derivative plicated expression, but heres a way to re-
reduces to the usual partial derivative. The late it to something that is more familiar.
advantage of the covariant derivative is that From advanced calculus we know that par-
it gives rise to tensors: the covariant deriva- tial derivatives commute on most functions
tive of tensors are tensors also.
2The same thing happens in QFT where the de-
1One can contract between any upper index and rivative is similarly modified by means of the 4-vector
any lower index. We can also contract between two potential of the electromagnetic field in order that the
lower or two upper indices after one raises or lowers Lagrangian remain invariant e.g., see [3], Section
one of them. Its just like taking the trace of a matrix. 10.3.
 How Einstein Got His Field Equations
that appear in practice: i.e., x y f y x f = 0
(usually called Clairauts Theorem). If one
works out a similar difference for the covari-
ant derivative in curved space, one does not
get 0 in general, but instead gets an expres-
sion that involves the components R of
the Riemann curvature tensor. So the cur-
vature tensor can be seen as the degree to
which covariant derivatives depend on the
order that they are performed.3
The Riemann tensor has an important con-
traction called the Ricci4 tensor:
R = R
(of rank 2). Lastly, contracting the Ricci ten-
sor once more we obtain the scalar curva-
ture tensor
R = R
which is a scalar field on spacetime (related
to the Gaussian curvature).
There is a rank 2 tensor that has special
physical significance. It is called the Stress-
Energy-Momentum tensor denoted by T .
It contains or models all the essential physics
of a system pertinent to gravitation and space
time - specifically, it contains information
regarding energy density, energy flow, and
momenta of matter particles and/or radia-
tion. It is an essential component of Ein-
steins field equations. It is normally as-
sumed that one has local energy-momentum
conservation, which translates into saying
that the energy momentum tensor satisfies
the covariant form of the continuity equa-
tion
T ; = 0.
(That is, its covariant divergence vanishes.)
Einstein knew that the Ricci tensor R
has strong similarity to energy-momentum
tensor. And at one point, early in November
3For a detailed calculation see Weinberg [5], Sec-
tion 6.5.
4Pronounced Reeshi.
5
1915, he conjectured5 that the two should
be proportional: R = T . Though this
equation is not quite correct, since Ricci does
not generally satisfy the continuity equation,
its a significant step closer to his ultimate
field equation (which he eventually reached
by the end of November).
We now turn to how Einstein, in his pa-
per, finally arrived at his field equation after
a long and elusive search which ended No-
vember 25, 1915.
2. Einsteins Derivation of
the Field Equations
Principle of Equivalence. This princi-
ple arose from what Einstein described as
the happiest thought of my life. Einstein
wrote6:
The gravitational field has only
a relative existence in a way sim-
ilar to the electric field gener-
ated by magentoelectric induc-
tion. Because for an observer
falling freely from the roof of a
house there exists at least in
his immediate surroundings no
gravitational field.
Einsteins Equivalence Principle comes down
to saying that spacetime is locally Minkowski
(locally inertial). That is, at each point p
of spacetime there is a coordinate system
where the metric tensor g is exactly the
Minkowski metric at p (see equation (1.3))
and where its derivatives vanish at p (i.e.,
gravity vanishes).7 In essence, then, in curved
spacetime each point has a local coordinate
5Pais [4], Chapter 14 (page 253).
6Pais [4], p. 178.
7Here we are referring to the usual partial deriva-
tives g / x not the covariant derivative, since it
is a theorem that the covariant derivatives of the met-
ric tensor vanish everywhere.
6 S. Walters

system where the Special Theory of Relativ- by suitably

p scaling the coordinate system so
ity holds good. that det(g ) = 1; he then gets the result
Einsteins Principle of General Covari-
ance states that all the laws of physics (at R = + = 0. (2.1)
x
least those pertaining to gravitational effects)
are to be expressed in covariant form using Using the Lagrangian/Action formalism, Ein-
tensors. That is, the laws of Physics should stein rewrote this equation in a more phys-
be expressible in terms of tensors since ten- ically meaningful form that has an impor-
sors transform in a covariant manner i.e., tant term in it that represents the energy-
the laws should have the same form regard- momentum tensor t of the gravitational
less of which coordinate system (observer) is field11; so Einstein recomputes equation (2.1)
used.8 to the following
Einsteins Principle of Covariance is also  
a method for finding how physical equations   1
g R = g + t t = 0
look in any coordinate system: just find out x 2
what the physical equation looks like in in- (2.2)

ertial coordinate systems (which exist locally for some constant and t = t , where t is
by the Equivalence Principle), and then make given by
the replacements
1
A t = g g .
g , A ; 2
x
Einsteins next insight: to get the grav-
so usual derivatives get replaced by covari- itational field equations in the presence of
ant derivatives. The resulting equation will matter and radiation, in the previous equa-
be the covariant form of the physical law in tion he replaced the energy momentum ten-
any new (accelerated) coordinate system.9 sor t of the gravitational field by the total
Einsteins Postulate: in vacuum10, the energy momentum tensor t + T of gravity
contracted curvature tensor, the Ricci ten- and of matter12 together! Once this is done,
sor, vanishes: he then retraced his steps back the way he
R = 0. arrived to (2.2) to obtain his field equations,
which now incorporate matter T .
Since this vacuum equation has 4 terms (see Replacing t by t +T in the vacuum equa-
definition above of the Riemann tensor given tion (2.2), Albert Einstein obtains the matter
by 4 terms), Einstein simplified it to 2 terms field equation:
8
This means, for example, that if A = B is a co-   
1

variant (tensor) form for a law of physics in one coor- g + (t + T )
2 (t + T ) =0
x
dinate system, then in another coordinate system it
would read quite similarly: A0 = B0 . For instance,
Maxwells equations can be so expressed in covariant 11It isnt a tensor in general relativity though it is
form - e.g., see [5], Chapter 5, Section 2. a tensor in Special Relativity since it is acts covari-
9See [5], Section 7.6. antly with respect to Lorentz transformations see
10In this context, vacuum means no matter, no [5], Section 7.6, page 167.
electromagnetic fields, no forces that are nongravita- 12Non-gravitational matter and radiation now be-
tional, etc. only spacetime (i.e., gravity). ing represented by the tensor T .
 How Einstein Got His Field Equations
where t + T = t + T is the contraction; or
  
1
 
1
g + t 2 t = T 2 T . per century.
x - He calculates the bending of light as well
Now he eliminates t from view (since it is
not really a tensor in General Relativity) and
he calculates the left side, which in view of
(2.2) is g R , and obtains it in terms of the
Ricci tensor thus:
 
g R = T 12 T . (2.3) experiment which Einstein was already aware
Notice that if there is no matter present, so
that the tensor T is zero, then this equation
reduces to the vacuum field equation (2.2)
g R = 0. Taking the g in (2.3) on the
other side he gets
 
R = g T 21 T = T 12 g T
or
R = T + 12 g T. (2.4) itational waves is [6].)
Now contract this equation with respect to
the indices and to get the scalar fields
1
R = T + 2 4T = T
and express the T in (2.4) in terms of R so
that the general field equation can now be
written in the form that is most familiar in
the literature:
R 12 g R = T .
This is Einsteins gravitational field equa-
tion in the presence of matter. (Matter and
electromagnetism are represented by the ten-
sor T .) He later calculates the constant
= 8G (in units where the speed of light
c is set to 1).
Applications/Predictions. Einstein ends
his paper with these applications:
- Einstein verifies that energy-momentum
tensor for gravity and matter t + T has en- VERSITY OF N OR THERN B.C., P RINCE G EORGE , B.C.
ergy conservation.
- He shows that to first order approxima-
tion he gets Newtons laws.
7
- Einstein calculates the precise preces-
sion of Mercurys orbit = 43 second of arc

as its redshift in a gravitational field. (This
was verified experimentally in 1919, three
years after his paper was published.)
- The slowing down of clocks in a gravita-
tional field. (This was verified by the Eotvos
of.)
There are numerous tests and applications
of his General Theory which Einstein him-
self did not anticipate. He did however pre-
dict the existence of gravitational waves [1]
which have just recently (in 2015-16) been
detected thru the intense gravitational bat-


tles between two black holes. (A good short
historical account of Einsteins work on grav-
R EFERENCES
[1] A. Einstein, Approximative Integration of the
Field Equations of Gravitation, Prussian Acad-
emy of Sciences, June 1916. Sitzungsber
Preuss. Akad. Wiss. Berlin (Math. Phys.) (1916),
688-696. https://inspirehep.net/record/42612
[2] H. A. Lorentz, A. Einstein, H. Minkowski, H.
Weyl, The Principle of Relativity. (Includes Eng-
lish translation of Einsteins general relativity
paper which appeared in Annalen der Physik
49,1916.)
[3] D. Griffiths, Introduction to Particle Physics,
Wiley-VCH Verlag, 2008, 2nd edition.
[4] Abraham Pais, Subtle is the Lord, The Science
and The Life of Albert Einstein, Oxford University
Press, 1982.
[5] Steven Weinberg, Gravitation and Cosmology,
John Wiley and Sons, 1972.
[6] G. Weinstein, Einsteins Discovery of
Gravitational Waves 1916-1918, 2016,
http://arxiv.org/abs/1602.04040
D EPAR TMENT OF M ATHEMATICS AND S TATISTICS , U NI -
V2N 4Z9, C ANADA .
E-mail address: walters@unbc.ca
URL: http://hilbert.unbc.ca/