University of Mary Washington

Eagle Scholar

Student Research Submissions

Spring 4-29-2022

Einstein's Equations, Lagrangians for General Relativity, and ADM

Formalism
Timothy Corbett

Follow this and additional works at: https://scholar.umw.edu/student_research

Part of the Elementary Particles and Fields and String Theory Commons

Recommended Citation
Corbett, Timothy, "Einstein's Equations, Lagrangians for General Relativity, and ADM Formalism" (2022).
Student Research Submissions. 454.
https://scholar.umw.edu/student_research/454

This Honors Project is brought to you for free and open access by Eagle Scholar. It has been accepted for inclusion
in Student Research Submissions by an authorized administrator of Eagle Scholar. For more information, please
contact archives@umw.edu.
Einstein’s Equations, Lagrangians for General
Relativity, and ADM Formalism

Timothy K. Corbett

submitted in partial fulfillment of the requirements for Honors in

Mathematics at the University of Mary Washington

Fredericksburg, Virginia

April 2022
This thesis by Timothy K. Corbett is accepted in its present form as satisfying the thesis
requirement for Honors in Mathematics.

Date Approved

4/28/2022

Dr. Yuan-Jen Chiang, Professor of Mathematics,

thesis advisor

Dr. Julius Esunge, Professor and Chair of Mathematics,

committee member

Dr. Suzanne Sumner, Professor of Mathematics,

committee member
Contents
1 Preliminaries 1
1.1 Basic Concepts of Manifolds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Riemann Curvature Tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Differential Forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Einstein’s Equation 4
2.1 The Stress-Energy Tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2 The Einstein Tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.3 The Schwarzschild Metric . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

3 Cosmology 8
3.1 Robertson-Walker Cosmological Model . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.2 Cosmological Evolution Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

4 Lagrangians for General Relativity 11

5 ADM Formalism 14
5.1 Extrinsic Curvature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
5.2 The Gauss-Codazzi Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
5.3 Canonical Quantization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
 Abstract
Einstein’s equations describe the relation of spacetime curvature and present matter. We will
consider the case of a Lorentzian manifold diffeomorphic to R × Σ, where time t ∈ R and
the three-dimensional manifold Σ represents space. In a homogeneous and isotropic universe,
the three-dimensional manifold Σ of the Lorentzian manifold has Riemann curvature tensor
(3)
R = KI, where K is a constant and I is the 3 × 3 identity matrix. Space is flat when K = 0,
spherical when K is positive, and hyperbolic when K is negative. In this paper, we will show
models agreeing with each. We will derive the Schwarzschild metric, find Einstein’s equations
from the Lagrangian, and analyze them using Arnowitt-Deser-Misner formalism.

1 Preliminaries
The subjects covered in this paper are important features in general relativity. While in special
relativity, we can approximate the geometry of spacetime to be R4 , this assumption is deemed
inappropriate in general relativity, as, when making this assumption, there are errors in predicting
the gravitation between two particles. We use a Lorentzian metric, which forces that spacetime
be curved in the presence of a gravitational field. “The world lines of freely falling bodies in a
gravitational field are simply the geodesics of the spacetime metric.”[10] Because of this, we cannot
meaningfully describe gravitation as a force field, but, rather, we find meaning in the relative
gravitational forces between free falling bodies. Now, what is a ”geodesic”? To answer this question,
we must cover some basic differential geometry for the sake of the reader.

1.1 Basic Concepts of Manifolds

Imagine we have a surface M , a two-dimensional manifold. The surface can be any shape of
membrane - a sphere, a torus, a cylinder, a paraboloid, a hyperboloid, a flag flapping in the wind,
etc. For a given point p on our surface, there is an open neighborhood Up of p. The surface, as a
whole, is composed of a collection of these open neighborhoods {Up }p∈M . A coordinate patch is a
C k one-to-one function x: Up −→ R3 for some k ≥ 1, for which the coordinates of Up are u1 , u2 and

∂x ∂x
1
× 2 = x1 × x2 ̸= 0. (1.1)
∂u ∂u
The unit normal at a p =x(a, b) is equal to
x1 × x2
n(a, b) = . (1.2)
|x1 × x2 |

This normal is perpendicular to the tangent plane Tp M at p. The tangent plane sitting at p =x(a, b)
is the plane perpendicular to x1 (a, b)×x2 (a, b). Therefore, if x(a, b) = (x1 , y 1 , z 1 ) and x1 ×x2 =
(x2 , y 2 , z 2 ), then the tangent plane would be

x2 (x − x1 ) + y 2 (y − y 1 ) + z 2 (z − z 1 ) = 0. (1.3)

We denote the tangent space at p, the set of all tangent vectors of a surface passing through p,
as Tp M . The metric is defined as the map,

g : Tp M × Tp M −→ R, (1.4)

1
for which its components, the first fundamental form of a surface, are

gij (u1 , u2 ) = ⟨xi (u1 , u2 ) · xj (u1 , u2 )⟩, i = 1, 2, j = 1, 2. (1.5)

In the Lorentzian manifold M = R × Σ, the metric takes the following form:

 
−1 0 0 0
0 
gµν = 
0 3
 , i, j = 1, 2, 3. (1.6)
gij 
0

where (3) gij is the metric on the three-dimensional manifold Σ. When we use an inverse of a
matrix, typically we will either raise or lower the indices. In the case of the metric, we define
the reverse in our notation as g µν . The second fundamental form is defined for tangent vectors
X= ΣX i xi ,Y= ΣY i yi
II(X, Y) = Σi,j Lij X i Y j , (1.7)
where coefficients Lij = ⟨xij , n⟩. The Christoffel symbols, the metric connections, are intrinsic
functions on Up defined as

1 ∂gνσ ∂gµσ ∂gνσ

Γρµν = Σσ g ρσ { µ + µ
− }. (1.8)
2 ∂x ∂x ∂xµ
Let x: Up −→ R3 be a local surface. Then we have Gauss’s formulas:

xij = Lij n + Σk Γkij xk . (1.9)

A geodesic on a surface is a unit speed curve with geodesic curvature equal to zero everywhere.
The geodesic curvature of a unit speed curve is the component of the second derivative of the curve
in the direction of S = n × T for tangent vector T . It is represented by Kg = [n, T, T ′ ]. For any
unit speed curve, γ(s) = x(γ 1 (s), γ 2 (s)), the normal curvature is

κn = Σi,j Lij (γ i )′ (γ j )′ (1.10)

and the geodesic curvature is

Kg S = Σk [(γ k )′′ + Σi,j Γkij (γ i )′ (γ j )′ ]xk . (1.11)

1.2 Riemann Curvature Tensor

For a point on a vector space V , we can define a vector bundle above the vector space. For this
vector bundle, we can define a connection, denoted by D. In particular, if this V is a tangent space,
the Levi-Cevita connection, a unique, metric-preserving and torsion free connection on Tp M for
metric g, is given the symbol ∇. If ∇gµν = 0, ∇ is a metric connection for gµν on M . Below is a
sample of the Lie algebra of this connection with vector fields ∂µ :

∇α ∂β − ∇β ∂α = [∂β , ∂α ]. (1.12)

We also witness some important algebra between ∇ and something we call the Riemann
curvature tensor, the curvature of ∇:

R(u, υ)w = (∇u ∇υ − ∇υ ∇u − ∇[u,υ] )w, (1.13)

2
 R(∂β , ∂γ )∂δ = (∇β ∇γ − ∇γ ∇β )∂δ . (1.14)
We can define the Riemann tensor itself in terms of Christoffel symbols,
α
Rβγδ = ∂β Γαγδ − ∂γ Γαβδ + Γσγδ Γαβσ − Γσβδ Γαγσ . (1.15)
γ
When we contract the indices of the Riemann tensor, we can get the Ricci tensor, Rαβ = Rαγβ ,
where we can describe this as
∂ λ ∂
Rµν = Σλ λ
Γµν − µ (Σλ Γλλν ) + Σα,λ (Γαµν Γλαλ − Γαλν Γλαµ ). (1.16)
∂x ∂x
We then can find the scalar curvature, R = Rαα , which is important when finding the action later.
We can take a moment to mention a few symmetries of the Riemann tensor:
λ
Rαβγδ = gαλ Rβγδ , (1.17)
λ λ
Rβγδ = −Rγβδ , (1.18)
Rαβγδ = −Rδβγα , (1.19)
λ
R[βγδ] = 0. (1.20)

1.3 Differential Forms

One other small detail we must go over is the differential form. A 1-form is an exterior derivative
of a function f . We can define derivative this as

df = ∂µ f dxµ . (1.21)

Since
dxµ (∂ν ) = ∂ν xµ = δνµ , (1.22)
we can define a 1-form ω on Rn when
ωµ = ω(∂µ ), (1.23)
as
ω = ωµ dxµ . (1.24)
We must mention the star operator, as it is used briefly later. When we take the cross product
of two vectors, in actuality, we are taking the wedge product and applying the star operator. The
wedge product will look similar to the cross product:

ω ∧ µ = −µ ∧ ω. (1.25)

The following is a list of star operators in the dx, dy, dz basis:

⋆ :dx ∧ dy 7→ dz (1.26)
⋆ :dy ∧ dz 7→ dx (1.27)
⋆ :dz ∧ dx 7→ dy. (1.28)

We can rewrite Maxwell’s equations for electromagnetism in these terms as

⋆d ⋆ F = J, (1.29)
d ⋆ J = 0, (1.30)

3
when J is the current density and
F = B + E ∧ dt, (1.31)
for magnetic and electric fields B and E. These equations are important because much of the
mathematics from electromagnetism can be used to justify certain relationships in general relativity,
as we will see in the section Einstein’s Equation.
Furthermore, we may find the Bianchi identity for a field F and connection D useful,

dD F = 0, (1.32)

when dD is the exterior covariant derivative.

2 Einstein’s Equation
2.1 The Stress-Energy Tensor
Einstein’s equation describes how spacetime is curved by the presence of matter (anything which
possesses energy or momentum). This equation is a (0,2) tensor of chosen local coordinates t = x0 ,
x1 , x2 , x3 . It is a symmetric tensor, so entries of the stress-energy tensor Tµν = Tνµ . We use the
stress-energy tensor to describe the law of conservation of energy-momentum on curved spacetime
as
g µλ ∇λ Tµν = ∇µ Tµν = 0. (2.1)
We know from electromagnetism that d ⋆ J = 0 for the current 1-form J, and we can show that
⋆d ⋆ J = −∇µ Jµ , which can also be written ∇µ Jµ = 0. We are working in Minkowski spacetime,
meaning the Levi-Civita connection is the flat connection, ∂ µ Jµ = 0. The following is our continuity
equation
∂ρ
+ ∇ · ⃗j = 0. (2.2)
∂t
This equation says any divergence in J gives a corresponding change in charge density ρ.
Similarly, T µν is divergence free on Minkowski spacetime, and with ∇µ Tµν = 0, if ν = 0, we
get the law of conservation of energy:

∂T 00 ∂T i0
+ =0 (2.3)
∂t ∂xi
and the law of conservation of momentum when j=1,2,3:

∂T 0j ∂T ij
+ = 0. (2.4)
∂t ∂xi

2.2 The Einstein Tensor

We want equations for gravity which are consistent with the law of conservation of energy, so we
choose for the stress energy tensor to be set equal to some divergence-free tensor depending only
on the curvature of spacetime, Cµν = Tµν . The Riemann tensor is such. We can find the simplest
divergence-free symmetric result using the Bianchi identity. To get there, we start with the Jacobi
identity for any vector fields α, β, γ on spacetime:

[∇α , [∇β , ∇γ ]] + [∇β , [∇γ , ∇α ]] + [∇γ , [∇α , ∇β ]] = 0. (2.5)

4
 The following Lie algebra is true: [∇µ , ∇ν ] = R(∂µ , ∂ν ). This relationship allows the following
substitution:
[∇α , R(∂β , ∂γ )] + [∇β , R(∂γ , ∂α )] + [∇γ , R(∂α , ∂β )] = 0. (2.6)
α
We change this equation in terms of curvature as the Riemann tensor. We describe Rβγδ with
α α
Rβγδ = −Rγβδ as R to get a nice form of the Bianchi identity, d∇ being the exterior covariant
derivative from the Levi-Civita connection:

d∇ R = 0. (2.7)

Another way we can describe this equation is in local coordinates. First, the exterior derivative
of a p-form is d(ωI dxI ) = (∂µ ωI )dxµ ∧ dxI = (∇µ ωI )dxµ ∧ dxI . This substitution changes the
Bianchi identity to
λ
∇[α Rβγ]δ = 0. (2.8)
α
By contracting indices, we have ∇[α Rβγ]δ = 0. Using the explicit antisymmetrization, contraction
and raising of indices, we can perform the following,
α α α
0 = ∇α Rβγδ + ∇β Rγαδ + ∇γ Rαβδ (2.9)
α α α
= ∇α Rβγδ + ∇β Rγαδ − ∇γ Rβαδ (2.10)
= ∇α Rαβγδ + ∇β Rγδ − ∇γ Rβδ (2.11)
= ∇α Rδγβα + ∇β Rγδ − ∇γ Rβδ (2.12)
= ∇α Rγα + ∇β Rγβ − ∇γ R (2.13)
= ∇α Rγα + ∇α Rγα − ∇γ R (2.14)
= 2∇α Rγα − ∇γ R (2.15)
1
= ∇α Rγα − ∇γ R (2.16)
2
1
= ∇α (Rγα − gγα R). (2.17)
2
The last is because ∇α gγα = 0. We then have Gµν = Rγα − 12 gγα R, the Einstein tensor. Now,
we can say
∇µ Gµν = 0. (2.18)
We find that the Einstein tensor is the result we were seeking. This result allows the Einstein
tensor to be a multiple of the stress-energy tensor.

Gµν = 8πκTµν . (2.19)

Now, since ∇µ gµν = 0, we could add a term proportional to the metric such that

Gµν + Λgµν = 8πκTµν , (2.20)

Λ being the cosmological constant.

2.3 The Schwarzschild Metric

With the Schwarzschild Solution, Einstein’s equation will describe the exterior gravitational field
(spacetime geometry of the surrounding empty space) of a static, spherically symmetric body. In

5
this case, spacetime coordinates will be t, r, θ, and ϕ. The metric of a static, spherically symmetric
spacetime is of this form:

ds2 = −f (r)dt2 + h(r)dr2 + r2 (dθ2 + sin2 θdϕ2 ). (2.21)

An orthonormal basis for this metric is shown here,

(e0 )µ = f 1/2 (dt)µ (2.22)

(e1 )µ = h1/2 (dr)µ (2.23)
(e2 )µ = r(dθ)µ (2.24)
(e3 )µ = rsinθ(dϕ)µ , (2.25)

then we use the derivative ∂µ where f ′ = df /dr,

1
∂[µ (e0 )ν] = f −1/2 f ′ (dr)[µ (dt)ν] (2.26)
2
∂[µ (e1 )ν] = 0 (2.27)
∂[µ (e2 )ν] = (dr)[µ (dθ)ν] (2.28)
∂[µ (e3 )ν] = sinθ(dr)[µ (dϕ)ν] rcosθ(dθ)[µ (dϕ)ν] . (2.29)

Now, using
∂[a (eσ )b] = Σµ,ν η µν (eµ )[a ωb]σν (2.30)
we solve for the connection 1-forms ωµαβ = −ωµβα :

1 −1/2 ′
f f (dr)[µ (dr)ν] = h1/2 (dr)[µ ων]01 + r(dθ)[µ ων]02 + rsinθ(dϕ)[µ ων]03 (2.31)
2
0 = f 1/2 (dt)[µ ων]01 + r(dθ)[µ ων]12 + rsinθ(dϕ)[µ ων]13 (2.32)
(dr)[µ (dθ)ν] = −f 1/2 (dt)[µ ων]20 + h1/2 (dr)[µ ων]21 + rsinθ(dϕ)[µ ων]23 (2.33)
1/2 1/2
sinθ(dr)[µ (dϕ)ν] + rcosθ(dθ)[µ (dϕ)ν] = −f (dt)[µ ων]30 + h (dr)[µ ων]31 r(dθ)[µ ων]32 . (2.34)

We offer the guess ων02 = ων03 = 0, so

1 f′
ων01 = (dt)ν + α1 (dr)ν . (2.35)
2 (f h)1/2

in which α1 is an undetermined function.

We plug these terms into equation (2.32) from just above to find that α1 = 0. We can also
guess from (2.32) that

ων12 = α2 (dθ)ν + α3 (dϕ)ν , (2.36)

α3
ων13 = α4 (dϕ)ν + (dθ)ν . (2.37)
sinθ
We plug everything into (2.33) to get

α2 = −h−1/2 , (2.38)
h1/2
ων23 = − α3 (dr)ν + α5 (dϕ)ν . (2.39)
rsinθ

6
 Next we substitute into (2.34) and have

α3 = 0, (2.40)
−1/2
α4 = −h sinθ, (2.41)
α5 = −cosθ. (2.42)

There have been no bumps in the road, so we have all of our values:

f′
ων01 = (dt)ν , (2.43)
2(f h)1/2
ων12 = −h−1/2 (dθ)ν , (2.44)
ων13 = −h−1/2 sinθ(dθ)ν , (2.45)
ων23 = −cosθ(dϕ)ν . (2.46)

Now, we can find all of the Riemann tensors:

d
Rµν01 = −Rµν 10 = [(f h)−1/2 f ′ ](dr)[µ (dt)ν] , (2.47)
dr
Rµν02 = f −1/2 h−1 f ′ (dθ)[µ (dt)ν] , (2.48)
−1/2 −1 ′
Rµν03 = f h f sinθ(dϕ)[µ (dt)ν] , (2.49)
−3/2 ′
Rµν12 = h h (dr)[µ (dθ)ν] , (2.50)
Rµν13 = sinθh−3/2 h′ (dr)[µ (dϕ)ν] , (2.51)
−1
Rµν23 = 2(1 − h )sinθ(dθ)[µ (dϕ)ν] . (2.52)

We find the Ricci tensors from the Riemann tensors and set them equal to zero for the vacuum
Einstein equation when Rαβ = Rµν (eα )µ (eβ )ν :

1 d
1
0 = R00 = R010 2
+ R020 + R0303
= (f h)−1/2 [(f h)−1/2 f ′ ] + (rf h)−1 f ′ , (2.53)
2 dr
1 −1/2 d −1/2 ′
0 = R11 = − (f h) [(f h) f ] + (rh2 )−1 h, (2.54)
2 dr
1 1
0 = R22 = − (rf h)−1 f ′ + (rh2 )−1 h′ + r−2 (1 − h−1 ) = R33 . (2.55)
2 2
The off-diagonals of Rαβ vanish as well.
We see from the sum

f ′ h′
R00 + R11 = + = 0, (2.56)
f h
√
that f = Kh−1 ⇒ t −→ Kt, we set constant K = 1. From these parameters,

1−f
R22 = −f ′ + = 0, (2.57)
r
or

d
(rf ) = 1. (2.58)
dr

7
 Now, f = 1 + Cr when C is constant.
We can now write out our general solution,

C 2 C
ds2 = −(1 + )dt + (1 + )−1 dr2 + r2 (dθ2 + sin2 θdϕ2 ). (2.59)
r r
When r −→ ∞, this solution reaches flatness asymptotically, so we can say that the behavior
of a body agrees with Newtonian behavior when r is large, or C = − 2GM
c2
. Therefore, this result is
our Schwarzschild metric,

2GM 2 2GM −1 2
ds2 = −(1 − 2
)dt + (1 − ) dr + r2 (dθ2 + sin2 θdϕ2 ). (2.60)
rc rc2
Here, r is our critical radius in which, if the entirety of the mass of the spherical object were
to be scrunched, then not even light could escape its gravity.[6] The equation is essentially setting
light speed to be the escape velocity from the gravitational field of a mass. The surface at this
radius is called the horizon and appears to be a black, circular void to the observer.

3 Cosmology
3.1 Robertson-Walker Cosmological Model
It has been a long-time assumption that our universe is homogeneous and isotropic. If we were
situated anywhere else in the universe, the laws of physics that we observe here would apply there,
and, regardless of where we look, observations which yield results should not change. “There are no
preferred directions in space.”[10] Through the ages of technology broadening our horizons, we have
continued to see this assumption holds. We can define homogeneity and isotropy mathematically.
Let us say there is a one-parameter family of spacelike hypersurfaces we call Σt . We will take
a look at this figure from General Relativity by R. Wald, (Fig. C1). Spacetime is spatially
homogeneous if for all t, and for any p,q ∈ Σt , there is an isometry of the spacetime metric
(some ϕ s.t. (ϕ ∗ g)ab = gab ) which takes p into q.[10] Next, we look to another figure from General
Relativity, (Fig. C2). A spacetime is isotropic at each point if there exist timelike curves with
tangents ua orthogonal to Σt , such that any point p connecting tangent vectors to Σt , sa1 , sa2 ∈ Vp ,
there is an isometry of gab for which p and ua are fixed, but sa1 rotates into sa2 . We see, therefore,
that we cannot construct a preferred tangent vector orthogonal to ua .
Because of the restrictions of homogeneity and isotropy, the Riemann tensor can of Σt be
described as a multiple of the identity operator,

(3) γδ γ δ
Rαβ = κI = δ[α δβ] . (3.1)

Fig. C1

8
 Fig. C2

We then lower the indices,

Rαβγδ = κhγ[α hβ]δ . (3.2)

Since κ is constant, which is implied by homogeneity, a space where this property is satisfied
is called a space of constant curvature. Any two spaces of constant curvature of equal dimension,
metric, and κ are locally isometric.
The 3-sphere spacial geometry provides all positive κ, negative κ comes from three-dimensional
hyperboloids, and κ = 0 for flat space. Given that the isotropic and homogeneous surfaces are
orthogonal to each other, we can define the four-dimensional metric for spacetime as

gµν = −uµ uν + hµν (t), (3.3)

for which hµν is the metric of a spatial geometry from the selection below for a proper time τ
carried by an isotropic observer:

ds2 = −dτ 2 + a2 (τ )(dψ 2 + sin2 ψ(dθ2 + sin2 θdϕ2 )), (3.4)

ds2 = −dτ 2 + a2 (τ )(dx2 + dy 2 + dz 2 ), (3.5)
2 2 2 2 2 2 2 2
ds = −dτ + a (τ )(dψ + sinh ψ(dθ + sin θdϕ )), (3.6)

for which (3.4) is the spherical metric, (3.5) is the flat space metric, and (3.6) is the hyperbolic
space metric.
These metrics are called the Robertson-Walker cosmological model. We want to use the
spacetime metric to find ordinary differential equations to predict the general evolution for homogeneous,
isotropic cosmology.

3.2 Cosmological Evolution Equations

First, we define the stress-energy tensor of matter in the present universe to be the average mass
density of matter times these tangent vectors for points on observation lines, perpendicular to
foliating hypersurfaces of spacetime, Tµν = ρuµ uν . We must also include a term involving thermal
radiation pressure P in our expression, so we multiply it by the metric of the hypersurfaces:

Tµν = ρuµ uν + P (gµν + uµ uν ). (3.7)

9
 Due to our two restrictions of homogeneity and isotropy, the only components we must compute
of the Einstein tensor are Gτ τ and Gss , so these terms are our only components for Einstein’s
equations:

Gµν uµ uν = Gτ τ = 8πTτ τ = 8πρ, (3.8)

µ ν
Gµν s s = Gss = 8πTss = 8πρ. (3.9)

First, we compute these expressions for flat spatial geometry, ds2 = −dτ 2 +a2 (τ )(dx2 +dy 2 +dz 2 ).
The Christoffel components are, ȧ being with respect to τ ,

Γτxx = Γτyy = Γτzz = aȧ, (3.10)

ȧ
Γxxτ = Γxτx = Γyyτ = Γyτ y = Γzzτ = Γzτ z = . (3.11)
a
From the definition of the Ricci tensor, we have

3ä
Rτ τ = − , (3.12)
a
ä ȧ2
Rss = a−2 Rxx = + 2 2. (3.13)
a a
ȧ2
Therefore, 12 R = 12 (−Rτ τ + 3Rss ) = 3( aä + a2
).
We then compute

ä ä ȧ2 ȧ2
Gτ τ = −3 + 3( + 2 ) = 3 2 = 8πρ, (3.14)
a a a a
ä ȧ 2 ä ȧ2 ä ȧ2
Gss = + 2 2 − 3( + 2 ) = −2 − 2 = 8πP. (3.15)
a a a a a a
2
We rewrite Gss to be 3 äa2 = −4π(ρ + 3P ).
Next, we compute Gτ τ and Gss for spherical and hyperbolic space, k being either ±1.

Γτθθ = r2 aȧ (3.16)

ȧ
Γrτ r = Γrrτ = Γθτ θ = Γθθτ = Γϕτϕ = Γϕϕτ = (3.17)
a

Γrθθ = −r(1 − kr2 ) (3.18)

1
Γθrθ = Γθθr = Γϕrϕ = Γθϕr = (3.19)
r
cosθ
Γϕθϕ = Γϕϕθ = (3.20)
sinθ

3ȧ′
Rτ τ = − (3.21)
a

10
 2k + 2ȧ2 + ȧä
Rrr = (3.22)
1 − kr2

Rθθ = −1 + 2kr2 − cot2 θ + csc2 θ + 2r2 ȧ2 + r2 aä (3.23)

Rϕϕ = r2 sin2 θ(2k + 2ȧ2 + aä) (3.24)

1 3k 3ȧ2
Gτ τ = Rτ τ + R = 2 + 2 = 8πρ (3.25)
2 a a
Rµν X µ X ν Rϕ
Rss = µ ν
= (3.26)
gµν X X gϕϕ

k ȧ2 2ä
Gss = − 2
− − = 8πP (3.27)
a a a
The general evolution differential equations are found to be

ȧ2 k
3 = 8πρ − 3 2 , (3.28)
a2 a
ä
3 = −4π(ρ + 3P ). (3.29)
a
We see that the universe must always be either expanding or contracting.
We will not work out the exact solutions for our three geometries in this paper, but, for the
curious reader,Rthe below
R table displays exact solutions taken from General Relativity by R. Wald,
for which η = dη = dt a:

Dust and Radiation Filled Robertson-Walker Cosmologies

Spatial Geometry ”Dust”; P = 0 Radiation;
q P = 13 ρ q
4 8πρa4 8πρa4 2 1/2
3-sphere, k = +1 a = 4πρa
3 (1 - cos η) a = 3 [1 - (1 - τ / 3 ) ]
4
τ = 4πρa
3 (η - sin η) .
4
Flat, k = 0 a = (6πρa3 )1/3 τ 2/3 a ( 32πρa
=q 3 )1/4 τ 1/2q
4πρa4 4 8πρa4 2
Hyperboloid, k = -1 a= 3 (cosh η - 1) a = 8πρa 3 [(1 + τ / 3 ) - 1 ]1/2
4πρa4
τ = 3 (sinh η - η)

4 Lagrangians for General Relativity

We want to derive Einstein’s equation from an action principle for gravity. The action principle
is a path along which an action is minimized, the action being the integral of the Lagrangian, the
kinetic energy minus the potential energy of a particle as a function of time.
When spacetime is an oriented manifold M on which there is a metric g, the Lagrangian for
general relativity is Rvol, R being the Ricci scalar curvature of g and vol g’s volume form. The
Einstein-Hilbert action is then Z
S(g) = Rvol. (4.1)
M

11
 p √
This integral can be rewritten in local coordinates as S(g) = M R | det g|dn x = M R − det gdn x.
R R

We are working in the Lorentzian case, so we calculate the variation of the action as
d
δS(g) = S(g + sδg)|s=0 , (4.2)
ds
with Lorentzian matrix g, (0,2)-tensor δg which vanishes outside a compact set when M is such,
and s is a real number that will form a path of metrics through g when in this expression as it
varies. This method is how we find the variation of any quantity depending on g, so we start with
our original definition Z Z
δS = δ(Rvol) = (δR)vol + Rδvol, (4.3)
M M
and we translate it into local coordinates:
Z p Z p p
δS = δ(R | det g|dn x) = (δR) | det g| + Rδ | det g|dn x. (4.4)
M M
We utilize the fact that, for any two matrices A and B,

d d
det (A + sB)|s=0 = det (A) det (1 + sA−1 B)|s=0 (4.5)
ds ds
= det (A)tr(A−1 B),

to say
d
δ det g = det (g + sδg)|s=0 (4.6)
ds
= det gg −1 δg
= (det g)g αβ δgαβ .

Since trace g αβ gαβ is equal to its number of dimensions, δ(g αβ gαβ ) = 0, so g αβ δgαβ = −gαβ δg αβ .
Therefore we can perform the following,
p 1p
δ(det g) = −(det g)gαβ δg αβ ⇒ δ − det g = − − det ggαβ δg αβ . (4.7)
2
This expression is just δvol= − 12 gαβ (δg αβ )vol.
We begin calculation of the Ricci scalar, δR,

δR = δRαα = δ(g αβ Rαβ ) = (δg αβ )Rαβ + g αβ δRαβ . (4.8)

γ
We need to find δRαβ = δRαγβ . This calculation means first finding the variation of the
Christoffel symbols:
1
Γαβγ = g αη (∂β gγη + ∂γ gβη − ∂η gβγ ), (4.9)
2
the variation being
1
δΓαβγ = g αη (∇β δgγη + ∇γ δgβη − ∇η δgβγ ). (4.10)
2
We can now look for the variation of the Riemann tensor,

12
 α
Rβγη = ∂β Γαγη − ∂γ Γαβη + Γσγη Γαβσ − Γσβη Γαγσ , (4.11)
as
α
δRβγη = ∇β δΓαγη − ∇γ δΓαβη . (4.12)
or
γ
δRαβ = δRαγβ = ∇α δΓγγβ − ∇γ δΓγαβ . (4.13)
We next include the variation of the Christoffel symbols:
1
δRαβ = (g γη ∇α ∇β δgγη + g γη ∇γ ∇η δgαβ − g γη ∇γ (∇β δgαη + ∇α δgβη )). (4.14)
2
We plug this term back into our δR expression:

δR = δ(g αβ Rαβ ) (4.15)

= (δg αβ )Rαβ + g αβ δRαβ
1
= Rαβ δg αβ + g αβ ( (g γη ∇α ∇β δgγη + g γη ∇γ ∇η δgαβ − g γη ∇γ (∇β δgαη + ∇α δgβη )))
2
= Rαβ δg + ∇ ∇γ (g αβ δgαβ ) − ∇α ∇β δgαβ .
αβ γ

If we so choose, we can make this expression look nicer by defining a 1-form ω to be

ωα = gαγη δgγη − ∇β δgαβ , (4.16)

so we have
δR = Rαβ δg αβ + ∇α ωα . (4.17)
Finally, now that we have our full variation expression,

Z
δS = (δR)vol + Rδvol (4.18)
ZM
1
= (Rαβ δg αβ + ∇α ωα − Rgαβ δg αβ )vol,
M 2
we can start deriving Einstein’s equation. If we look at the part of this equation without the
ω-term, we have Z
1
(Rαβ − Rgαβ )(δg αβ )vol. (4.19)
M 2
Remember Einstein’s equation,
1
Rαβ − Rgαβ = 0. (4.20)
2
When this equation is true, our integral is zero for all variations δg αβ which vanish outside a
compact set M . Now for the remaining term, it is true that for any 1-form,

∇α ωα = − ⋆ d ⋆ ω, (4.21)

so we are left with the following through Stokes’ theorem, ⋆2 = ±1,

Z Z Z
− vol ∧ ⋆d ⋆ ω = ± d ⋆ ω ∧ ⋆vol = ± d ⋆ ω = 0. (4.22)
M M M

13
 We can now know the linearized vacuum Einstein equation. We solve using a perturbation
g + ϵh such that the new metric is a solution up to first order in ϵ. The Ricci tensor for g vanishes,
since g is a solution, so the Ricci tensor of g + ϵh is
ϵ γη
(g ∇α ∇β hγη + g γη ∇γ ∇η hαβ − g γη ∇γ (∇β hαη + ∇α hβη )), (4.23)
2
along with higher order terms. If h is to vanish, it must satisfy the linearized Einstein equation,
which we find to be

g γη ∇α ∇β hγη + g γη ∇γ ∇η hαβ − g γη ∇γ (∇β hαη + ∇α hβη ) = 0. (4.24)

The linearized Einstein equation is used to study gravitational waves. Gravitational waves are
interesting in that they are ripples across spacetime caused by massive objects moving at extreme
accelerations.[4] This example includes neutron stars and black holes. The detection of gravitational
waves offers a separate detection method to electromagnetic waves, as gravitational waves are largely
unimpeded by matter. We can track ripples farther away than with light detection, which can be
bent by massive celestial bodies, or even blocked by space materials.

5 ADM Formalism
Let us say we have a Lorentzian manifold M and a diffeomorphism defined

ϕ : M −→ R × S, (5.1)
for which t ∈ R is our time and S is space. We define a time coordinate on spacetime M as the
pullback of t on R × S, τ = ϕ ∗ t. We will say that Σ ⊂ M is a slice of spacetime M if time τ is
constant on Σ. We call the metric on Σ the 3-metric, or 3 g. The way Σ sits in M , or the extrinsic
curvature, represents the time derivative of 3 g, which is K in our context, and our Cauchy data for
the metric will be (3 g,K).
The Einstein tensor contains ten independent components, so, in actuality, rather than the
Einstein equation, we have the Einstein equations. This section is about the formulation of these ten
equations, the Arnowitt-Deser-Misner (ADM) formulation. The first four equations we will consider
are constraint equations for the Cauchy data. The six other equations describe the evolution of the
3-metric over time.

5.1 Extrinsic Curvature

When Σ is a spacelike slice of spacetime M , with the 3-metric, a field of timelike normal unit
vectors n exists such that

g(n, n) = −1 (5.2)
and ∀ υ ∈ Tp Σ, g(n, υ) = 0, for which Tp Σ is the tangent space on Σ. If we switch the sign of n,
we can designate directions for the future and past. We can expand any vector υ ∈ Tp M into two
components

υ = −g(υ, n)n + (υ + g(υ, n)n). (5.3)

We can check that the normal component of n is just n,

14
 −g(n, n)n = −(−1)n = n, (5.4)
and that the tangential component of υ is orthogonal to n,

g(υ + g(υ, n)n, n) = g(υ, n) + g(υ, n)g(n, n) = g(υ, n) − g(υ, n) = 0. (5.5)

Then, for any vector fields u, υ on Σ,

∇u υ = −g(∇u υ, n)n + (∇u υ + g(∇u υ, n)n) (5.6)

= K(u, υ)n + (∇u υ + g(∇u υ, n)n),

where K(u, υ) is the extrinsic curvature. When parallel translated using the Levi-Civita connection
∇ on M , a tangent vector of Σ will fail to a certain degree to be tangent. This failure is the extrinsic
curvature.
The second term is the Levi-Civita connection associated to 3 g:

3
∇u υ = ∇u υ + g(∇u υ, n)n. (5.7)
To prove this fact, we first show this expression is a connection for some u, υ, w ∈ Vect(Σ),
f ∈ C ∞ (Σ), and g(n, w) = 0,

3
∇u (f w) = ∇υ (f w) + g(n, ∇υ (f w))n (5.8)
= υ(f )w + f ∇υ w + g(n, υ(f )w)n + g(n, f ∇υ w)n
= υ(f )w + f ∇υ w + f g(n, ∇υ w)n
= υ(f )w + f (∇υ w + g(n, ∇υ w)n))
= υ(f )w + f 3 ∇u w.

Next, we prove metric preservation with g(n, w) = g(υ, n) = 0:

ug(υ, w) = g(∇u , w) + g(υ, ∇u w) (5.9)

3 3
= g(K(u, υ)n + ∇u , w) + g(υ, K(u, w)n + ∇u w)
= g(3 ∇u υ, w) + g(υ,3 ∇u w).

Lastly, we prove this result is torsion-free, using K(u, υ) = K(υ, u):

3
∇u υ −3 ∇υ u = ∇u υ − K(u, υ)n − ∇υ u + K(υ, u)n (5.10)
= ∇u υ − ∇ υ u
= [u, υ].

A tensor depends C ∞ (Σ)-linearly on its associated vector fields, so, given that K(u, υ) is a
symmetric tensor,

K(u, υ) = Kij ui υ j (5.11)

such that
Kij = K(∂i , ∂j ). (5.12)

15
 We can show that K(u, υ) is C ∞ (Σ)-linear in u,

K(f u, υ) = −g(∇f u υ, n) = −g(f ∇u υ, n) = −f g(∇u υ, n) = f K(u, υ), (5.13)

and in υ,

K(u, f υ) = −g(∇u f υ, n) (5.14)

= −g(u(f )υ + f ∇u υ, n)
= −f g(∇u υ, n)
= −f K(u, υ).

For symmetry, using that ∇ is torsion free,

Kij –Kji = K(∂i , ∂j )–K(∂j , ∂i ) (5.15)

= −g(∇i ∂j , n) + g(∇j ∂i , n)
= −g(∇i ∂j − ∇j ∂i , n)
= −g([∂i , ∂j ], n)
= 0.

We can also define K(u, υ) = g(∇u n, υ), which agrees with the argument from the metric
preservation portion of the proof, substituting n for υ and υ for w,

0 = ug(n, υ) = g(∇u υ, n) + g(n, ∇u υ). (5.16)

We see that this definition allows for extrinsic curvature to be seen from the perspective that it is
how much parallel translating in the direction of u is n forced to rotate in the direction of υ.

5.2 The Gauss-Codazzi Equations

We will now show that four of Einstein’s equations in 4-dimensional spacetime, the Gauss-Codazzi
equations describe constraints on 3 g and K. We revisit our diffeomorphism:

ϕ : M −→ R × S. (5.17)
Let us define a particular vector field ∂t on M as the pushforward by ϕ−1 of ∂t on R × S. We
will focus on the slice Σ for which τ = 0. We split ∂τ into normal and tangential components

⃗,
∂τ = −g(∂τ , n)n + (∂τ + g(∂τ , n)n) = N n + N (5.18)
⃗ . We then have an expression for the unit
the lapse function being N and the shift vector being N
normal,
1 ⃗ ).
n= (∂τ − N (5.19)
N
Now, we choose a point p on Σ with local coordinates x0 = τ , x1 , x2 , x3 and vector fields
∂0 = ∂τ , ∂1 , ∂2 , ∂3 tangent to Σ. The Christoffel symbols of 3 ∇ will be 3 Γijk and the Riemann
tensor of 3 g will be 3 Rijk
m . We want Rα in terms of K and 3-Riemann. We start with
ijk ij

16
 R(∂i , ∂j )∂k = ∇i ∇j ∂k − ∇j ∇i ∂k . (5.20)
For any vector fields u, υ on Σ,

∇u υ = K(u, υ)n +3 ∇u υ ⇒ ∇i ∂j = Kij n +3 Γm

ij ∂m , (5.21)
and, since K(u, υ) = g(∇u n, υ), ∇i n = Cim ∂m .
We can now find the first term,

∇i ∇j ∂k = ∇i (Kjk n +3 Γm
jk ∂k ) (5.22)
= Kjk,i n + Kjk ∇i n + Γm
3 3 m
jk,i ∂m + Γjk ∇i ∂m
= Kjk,i n + Kjk Kim ∂m +3 Γm 3 m 3 ℓ
jk,i ∂m + Γjk (Kim n + Γim ∂ℓ )
= (Kjk,i n +3 Γm m 3 m 3 ℓ 3 m
jk Kim )n + Kjk Ki ∂m + ( Γjk,i + Γjk Γiℓ )∂m .

The second term just has switched i, j indices, so completing the subtraction,

R(∂i , ∂j )∂k =(Kjk,i –Kik,j +3 Γm 3 m

jk Kim − Γik Kjm )n (5.23)
+ (Kjk Kim –Kik Kjm )∂m
+ (3 Γm 3 m 3 ℓ 3 m
jk,i − Γik,j + Γjk Γiℓ −3 Γℓik 3 Γm
jℓ )∂m .

We can condense the first line to

(3 ∇i Kjk −3 ∇j Kik )n (5.24)

and the third line to

3 m
Rijk ∂m (5.25)
to retrieve our Gauss-Codazzi equations:

R(∂i , ∂j )∂k = (3 ∇i Kj k −3 ∇j Kik )n + (3 Rijk

m
+ Kjk Kim –Kik Kjm )∂m . (5.26)
The formulae are easier to deal with if we assume the lapse is 1 and the shift is 0. If we apply
dx0 to both sides, we find the Gauss equation,

0
Rijk =3 ∇i Kjk −3 ∇j Kik , (5.27)
or if we apply dxm , we find the Codazzi equation,

m
Rijk =3 Rijk
m
+ Kjk Kim –Kik Kjm . (5.28)
This result says that intrinsic curvature 3 Rijk
m = Rm when extrinsic curvature K = 0.
ijk
0
To find the four Gα Einstein equations, we break down the Riemann tensor symmetries:

α α α α
Rµαν = −Rαµν = −Rµν α = Rµ να = Rµν . (5.29)
If we raise the µ index, Rνµ = Rνα
µα
, therefore,

17
 1 αβ
Gµν = Rνα
µα
− δνµ Rαβ . (5.30)
2
If µ = ν = 0, G00 = −(R12
12 + R23 + R31 ). Then, with the Codazzi equation,
23 31

−G0 0 =(3) R12 12 +(3) R23 23 +(3) R31 31 +(K 2 1 K 1 2 –K 2 2 K 1 1 )+(K 3 2 K 2 3 –K 3 3 K 2 2 )+(K 1 3 K 3 1 –K 1 1 K 3 3 ).
(5.31)
1 i 2 ij
Now, the K terms are equal to − 2 ((Ki ) –Kij K ), so

(3) 13
R12 12 +(3) R23 23 +(3) R31
31
= R. (5.32)
2
We use these terms to find that
1
G0 0 = − (3 R + (K i i )2 –Kij K ij ), (5.33)
2
and, if K is a matrix,
1
G0 0 = − (3 R + (trK)2 –tr(K 2 )) = 8πκT00 , (5.34)
2
which is a constraint relating the K of a Σ slice to its scalar curvature.
If we instead set µ = 0 and ν = 1, we ignore the second term to get

G0 1 = R0α 1α = R0 α1 α = R0 01 0 + R0 11 1 + R0 21 2 + R0 31 3 . (5.35)
The first two terms disappear, so G01 = R21
0 2 + R0 3 . Then, with the Gauss equation,
31

G01 = (3 ∇2 K12 −3 ∇1 K22 ) + (3 ∇3 K13 −3 ∇1 K33 ) (5.36)

= (3 ∇1 K11 −3 ∇1 Kjj )–(3 ∇1 K11 −3 ∇j K1j )
=3 ∇j K1j −3 ∇1 Kjj .

This expression is general for all G0i constraints:

G0i =3 ∇j Kij −3 ∇i Kij = 8πκTi0 , (5.37)

constraints on K for any Σ.
If we were to drop the original assumption that the lapse is 1 and the shift is 0, we would find
the general constraints
1
Gµν nµ nν = − (3 R + (trC)2 –tr(C 2 )) (5.38)
2
and
Gµi nµ =3 ∇j Cij −3 ∇i Cjj . (5.39)
We have successfully found our four restraint equations.

18
5.3 Canonical Quantization
We can quantize gravity into particles called gravitons when gravity is weak and spacetime is
approximately flat.[9] In this section, we look at how the quantum mechanics of a particle in n-
dimensional space, Rn applies to quantum gravity. Let the particle be traveling a path q(t) in this
configuration space satisfying the Euler-Lagrange equations

∂L d ∂L
i
= , (5.40)
∂q dt ∂ q̇ i
for which the Lagrangian
1
L(q, q̇) = mq̇ 2 –V (q), (5.41)
2
where q̇ is the velocity, m is mass, and potential energy is V . Between the two equations, we have
the equation of motion,
mq̈ = −∇V (q). (5.42)
We then have the momentum conjugate to the position,
∂L
pi = . (5.43)
∂ q̇ i
When a particle is in a potential, momentum p = mq̇, and, in general, the Hamiltonian is
described by

H(p, q) = p · q̇–L(q, q̇).

For a particle in a potential,

H(p, q) = p2 /2m–V (q), (5.44)
but this relationship does not hold in general relativity due to constraints.
To convert the Euler-Lagrange equations into evolutionary equations for the state over time,
we first compute
∂H ∂H
dH = dpi + i dq i (5.45)
∂pi ∂q
then, using the definition of momentum,
∂L i ∂L i
dH = ṗi dq̇ i + q̇ i dpi − dq − i dq̇ (5.46)
∂q i ∂ q̇
d ∂L
= ṗi dq̇ i + q̇ i dpi − dq i –pi dq̇ i
dt ∂ q̇ i
= q̇ i dpi − ṗi dq i .

When we combine these expressions at dH, we find Hamilton’s equations:

∂H ∂H
q̇ i = , ṗi = . (5.47)
∂pi ∂pi
We can write a neater definition if we use the Poisson bracket as an algebra on phase space,

∂f ∂g ∂f ∂g
{f, g} = i
– (5.48)
∂pi ∂q ∂q i ∂pi

19
and the Leibniz Law,
{f, gh} = {f, g}h + g{f, h}, (5.49)
and get

q̇ i = {H, q i }, (5.50)
ṗi = {H, pi }. (5.51)

If we consider the time derivative of observable f ,

d ∂f ∂f
f (p, q) = i q̇ i + ṗi (5.52)
dt ∂q ∂pi
∂H ∂f ∂H ∂f
= i
− i
∂pi ∂q ∂q ∂pi
= {H, f },

we see that the Poisson bracket with the Hamiltonian gives the rate of change of the observable.
The Hamiltonian generates time evolution.[1]
If we wish to quantize this particle, we need to have observables which are self-adjoint operators
on L2 (Rn ). We exchange f for fˆ such that if {f, g} = k, [fˆ, ĝ] = −ik̂ℏ, then let ℏ = 1 in our units.
If we manage to assign operators to observables in the ideal manner, we can set

fˆt = eitĤ fˆe−itĤ (5.53)

to find the time evolution of observables to be
d ˆ
ft = i[Ĥ, fˆt ]. (5.54)
dt
In the case of a free particle in potential V = 0, position and momentum have the Poisson
brackets pj , q k = δjk and pj , pk = q j , q k = 0. Now, our operators will be

(q̂ j ψ)(x) = xj ψ(x), (5.55)

(p̂j ψ)(x) = −i∂j ψ(x), (5.56)

and our canonical commutation relations

[p̂j , q̂ k ] = −iδjk , (5.57)

[p̂j , p̂k ] = [q̂ j , q̂ k ] = 0, (5.58)

p̂ 2
while converting our original Hamiltonian to Ĥ = 2m .
To apply this concept to general relativity, we consider spacetime M to be diffeomorphic to
R × S with our fixed spacelike slice Σ. We will only work in the vacuum Einstein equation so as
not to make things more complicated. Instead of q, we use 3 gij =3 qij , and, instead of Rn , we use
Met(Σ) as the configuration space for gravity. If we go back to our Kij derivations, in terms of
operators, we find that

20
 1
Kij = N −1 (q̂ij −3 ∇i Nj −3 ∇j Ni ), (5.59)
2
and p
L=R − det gd4 x, (5.60)
which we will alter to have
p
L = R − det g = q 1/2 N R = q 1/2 N (3 R + tr(K 2 )–(trK)2 ). (5.61)

From the above forms of the Lagrangian and extrinsic curvature,

∂L
pij = = q 1/2 (K ij –tr(K)q ij ). (5.62)
∂ q̇ij

When we integrate the Hamiltonian density over the slice, we find the Hamiltonian:
Z Z
ij 3
H= H(p , qij )d x = pij q̇ ij − Ld3 x (5.63)
Σ Σ
when in a compact space. Since we are in the compact case, we can ignore total divergences and
get the form

H = q 1/2 (N C + N i Ci ), (5.64)
for which C = −3 R + q −1 (tr(p2 ) − 12 tr(p)2 ) = −2Gµν nµ nν and Ci = −23 ∇j (q −1/2 pij ) = −2Gµi nµ .
Because of these C constraints, the vacuum equation implies that the Hamiltonian density
disappears. This result is because C = Ci = 0 are the four constraint Einstein equations.
The phase space may be the space of the cotangent bundle T ∗ Met(Σ), but not all position-
momentum coefficient pairs represent allowed states.[1] The constraints provide us with a phase
space of allowed states called the physical phase space,

X = {C = Ci = 0} ⊂ T ∗ Met(Σ), (5.65)
over which H vanishes. Our Hamiltonian equations, however, still provide nontrivial dynamics. We
can perform variation calculations on the Poisson brackets of pij and q ij :

{pij (x), qkℓ (y)} = (δki δℓj + δℓi δkj )δ (3) (x − y) (5.66)
ij kℓ
{p (x), p (y)} = 0 (5.67)
{qij (x), qkℓ (y)} = 0. (5.68)

We then use Einstein’s equations Gij = 0 of the second time derivative of 3 qij to get our six
time evolutionary equations when i, j = 1,2,3:

q̇ ij = {H, q ij }, (5.69)
ṗij = {H, pij }, (5.70)

the expansions of which are the non-trivial

1
q̇ij = 2q −1/2 N (pij − pkk qij ) + 23 ∇[i Nj] , (5.71)
2

21
 1 3 ij ij 1 1
ṗij = − N q −1/2 (3 Rij −R q ) + N q −1/2 q ij (pab pab − (paa )2 ) (5.72)
2 2 2
−1/2 ia j 1 a ij 1/2 i j ij 3 a3
–2N q (p pa − pa p ) + q (∇ ∇ N –q ∇ ∇a N )
2
1/2 −1/2 a ij
+ q ∇a (q N p )–2pa[i3 ∇a N j]

When we attempt to quantize gravity, the first problem we face is defining a Hilbert space. Since
Met(Σ) is infinite-dimensional, we cannot find an obvious square-integrable function to use, but we
pretend that L2 (Met(Σ) works regardless. Our substitute operator for the 3g , for g ∈ Met(Σ), and
x a point on Σ,

(qˆij (x)ψ)(q) = gij (x)ψ(q), (5.73)

with our momentum operator as

∂
(p̂ij (x)ψ)(q) = −i ψ(q). (5.74)
∂qij (x)
These forms give us our canonical commutation relations,

[p̂ij (x), q̂kℓ (y)] = −i(δki δℓj + δℓi δkj )δ (3) (x, y) (5.75)
[p̂ij (x), p̂kℓ (y)] = 0 (5.76)
[q̂ij (x), q̂kℓ (y)] = 0. (5.77)

When trying to quantize the Hamiltonian, we run into an operator ordering problem, since,
if we replace qij , pij with qˆij , pˆij , we are confronted with the fact that these expressions do not
commute. The constraints, therefore, are difficult to achieve. Ideally, we would find constraints
satisfying

[Ĉ(N ⃗ ′ )] = −iĈ([N
⃗ ), Ĉ(N ⃗,N⃗ ′ ]) (5.78)
[Ĉ(N⃗ , Ĉ(N ′ )] = −iĈ(N⃗ N ′) (5.79)
[Ĉ(N ), Ĉ(N ′ )] = −iĈ((N ∂ i N ′ –N ′ ∂ i N )∂i ), (5.80)

so we will act forward as though we could. We would then find that our quantum theory Hamiltonian
is
Z
Ĥ = (N Ĉ + N i Ĉi )q 1/2 d3 x. (5.81)
Σ

If we use Dirac algebra, for which ψ ∈ L2 Met(Σ) is a physical state, it must satisfy the quantum
restraints

⃗ )ψ = 0.
Ĉ(N )ψ = Ĉ(N (5.82)
We could also say that, for all N, N⃗ , the Wheeler-DeWitt equation holds with Ĥψ = 0. In
either case, we can force the Hamiltonian to disappear in the physical phase space. Unfortunately,
few solutions have been found for this Wheeler-DeWitt equation. In addition to this fact, if we
were to find physical states, the physical phase space would be

22
 ⃗ Ĉ(N )ψ = Ĉ(N
Hphys = {ψ : ∀N, N ⃗ )ψ = 0}, (5.83)
of which we cannot expect the physically relevant inner product to agree with the L2 Met(Σ)
inner product. Since our Hamiltonian vanishes on Hphys , any operator will commute with it, leading
to the revelation that these operators do not change with time:

d
At = i[Ĥ, At ] = 0. (5.84)
dt
This result is because Hphys describes quantum gravity for all time, rather than with time
dependency. Therefore, quantum gravity faces many roadblocks to drawing a connection between
gravity and the three other fundamental forces. This topic this, however, a growing field. For
example, there has been achieved quantum entanglement of diamonds and - possibly - waterbears.
It is believed that this approach is the future to finding a method of experiencing the quantization
of gravity.

23
References
[1] John Baez, Javier P. Muniain (1994) Gauge Fields, Knots and Gravity, World Scientific.

[2] Alessandro Danieli (2019) ADM formalism: a Hamiltonian approach to General Relativity,
Universitas Studiorum Mediolanensis.

[3] Andrew Hamilton (2019) Schwarzschild Geometry, https://jila.colorado.edu/ ajsh/bh/schwp.html.

[4] LIGO Caltech website, https://www.ligo.caltech.edu/page/gw-sources.

[5] Richard S. Millman, George D. Parker (1977) Elements of Differential Geometry, Prentice-Hall.

[6] Aula Raul (2018) Solutions to the Friedmann Equation, http://fma.if.usp.br.

[7] Joel W. Robbin, Dietmar A. Salamon (2021) Introduction to Differential Geometry, Lecture
notes.

[8] Iva Stavrov (2020) Curvature of Space and time, with an Introduction to Geometric Analysis,
American Mathematical Society Institute for Advanced Study.

[9] Hajime Urakawa (1993) Calculus of Variations and Harmonic Maps, American Mathematical
Society.

[10] Robert M. Wald (1984) General Relativity, The University of Chicago Press.

24