Lv Jindong

3C
The Lagrangian Method
and
The Hamiltonian Method
Physics Portfolio work 5
Introduction
I was reading up about classical mechanics in David. J Morin’s book, when I came across
this chapter about Lagrangian mechanics. I wanted to share my experience of learning about this
new chapter of physics as I found it to be extremely enlightening and it taught me to think about
classical mechanics in new ways and to solve problems in faster and more efficient methods.
However, most importantly the research and study of Lagrangian mechanics brought me across
this new branch of physics known as the Hamiltonian method, which is similar to the Lagrangian
method.

I wanted to share my experience learning about the Hamiltonian Method too, and its
applications in not only classical mechanics, but also in other physics areas too, most importantly
quantum physics and special relativity. In fact, quantum mechanics is derived from the
Hamiltonian method! I also wanted to do a comparison between thee 2 methods, their weakness
and strengths, their differences and similarities to understand these 2 methods better.

Last but not least, I always think of physics as not just facts and knowledge, and I like to
appreciate the math part in physics too. That is why I always seek to learn how to apply the
knowledge that I have learnt in practice. What use is there memorizing facts when you can learn
how to use them? Understanding is never complete without the knowledge to apply what you
have learnt. This is why I have selected a myriad of problems from the book itself, other sources
and some being my own questions to practice on both the Hamiltonian Method and the
Lagrangian Method.

However, unfortunately the biggest problem in this learning process is the lack of time.
Although the 1 month break seems long, in reality if you throw in all the homework, projects,
Olympiads practices and question, time passes in a blink of an eye. I tried to learn as much about
these topics as possible, and I focused especially on mechanics, which is why mechanics has
much greater content in this portfolio then other topics like quantum physics and special
relativity. Those 2 topics are quite content heavy and with more time, perhaps I can try to learn
more about these areas.
Partial Differentiation
What is a partial differentiation? Usually what I commonly see is normal dx/dt
differentiation, but I have not come across the term partial differential before. It is easier to
visualize normal differentiation as simply the slope of a curve, but what is partial
differentiation?

I learnt that with a function of many variables, partial differentiation is required if we

want to take only the derivative on one variable, say x. Then we use the terminology “partial
derivative with respect to x” with the notation δx/δt. To evaluate this derivative, we assume
all other variable in the function to be constant i.e., not a function of x. The, we treat it as a
normal derivative with some constants.

For example, if we have a function f(x,y)= x+y+x2+y2, then the partial derivative of
this function with respect to x is 2x+1. If we plot the value of f(x,y) in cartesian coordinates,
then you can visualize the partial derivative with respect to x as the slop of the curve formed
by the intersection of the function’s surface with the vertical plane parallel to the x axis and
passing through the point in question, and similarly for y.

Differential equations
A differential equation is a linear equation in one variable but involving derivatives of
that. We usually denote the nth derivative of x as the variable x with n dots above x.
However, since we cannot type that out in word, I will try to denote it with a power instead.
For example, dx/dt=x1. Usually, an nth order homogenous differential equations has n
solutions of x in terms of t. I won’t go in detail in solving differential equations in this
portfolio work but just this brief introduction.

Very briefly, the main idea for solving differential equations is to plug some form of x
in terms of t into the equation and try to solve it. In the case of a second order homogeneous
differential equation, a usual substitution is Aeat, and then solving for a. A is determined by
the initial conditions.
The Euler-Lagrange equations
The Euler-Lagrange equations are a set of second order differential equations discovered
by mathematicians Leonhard Euler and Joseph Lagrange in the 1750s. In classical mechanics, it
is equivalent to Newton’s Laws of motion. However, it has the advantage that it takes the form in
any system of generalized coordinates. The procedure is as such:

The Lagrangian is defined as L= T-V, a combination of the kinetic energy, K and the potential
energy, V.

For example, in the case of a spring, the Lagrangian L= ½ mv2+ ½ kx2

Then, write

The second order differential equation obtained is what we call an equation of motion. Why we
say this is remarkably similar to Newtons Law of motions is that they are equivalent in the 3D
planes. However, the Lagrangian method is often a quicker and more elegant way to solve
problems in complex situations with lots of forces involved.

In a 3 dimensional set-up written in cartesian coordinates, the potential energy is a function of x,

y and z so it takes the form V(x,y,z). Then, the Lagrangian is

Applying the Euler-Lagrange equations to the 3 coordinates, we obtain the statement

The right-hand side of this equation is the definition of force. This means that the Euler-
Lagrange equations equate to F=ma!

To fully understand the Euler-Lagrange equations, I researched a little about differential

equations and partial differentiation as shown earlier.
Hamiltonian’s principle of stationary action
Consider the quantity

S is known as the action. It is a quantity with the dimensions of (energy)×(time). S is

dependent on the Lagrangian. Given any function x(t), x as a function of time, we can find S. If
the function yields a stationary value (i.e., maximum, minimum or saddle point of S), then the
Euler-Lagrange theorem can be applied to x. This has been proven by Hamilton via calculus.

The principle of Stationary action states that the path of a particle is one that yields a
stationary value of the action. Hamilton developed the entire branch of Hamiltonian Mechanics
based on this principle. This principle is equivalent to F=ma, as if you have a stationary value of
S, then the Euler Lagrange equations apply, and the Euler-Lagrange equations are equivalent to
Newton’s Laws of motions.

Noether’s theorem
Now we will look at one of the most beautiful theorems in physics. It deals with 2
fundamental concepts, symmetry and conserved quantities. The theorem was introduced and
proved by famous Mathematician Emmy Noether.

Noether’s theorem states that for each symmetry of the lagrangian, there is a conserved
quantity. By symmetry, what she means is that if the coordinates are changed by some small
quantities, then the Lagrangian has no first-order change in these quantities. By “conserved
quantity,” we mean a quantity that does not change with time.
Noether proved that the quantity remains constant and does not
change with time.

P is given the generic name of conserved momentum, but it does not necessarily have the
units of linear momentum. K(q) is a function of q that the Lagrangian is invariant under the
change of coordinates q=q+xK(q). Noether’s theorem is extremely important as it gives
insightful ideas into conservation laws and allows investigators to determine the invariants in a
physical system. I applied Noether’s theorem later as an extension to our Atwood machine
problems.
Problems

Atwoods:
Lagrangian in Optics
I learnt this by online research too on other applications of the Lagrangian. To
understand how we can use Lagrangian in Optics, firstly we have to understand Fermat’s
principle. Fermat’s principle states that the optical length of the path followed by light
between two fixed points, A and B, is a stationary point. The stationary point may be a
maximum, a minimum, constant or an inflection point. What this means is that the path
taken by the light ray to travel from A to B is one that results in the minimum time
elapsed. Most of the time, the optical path length is considered. The OPL is defined as the
refractive index of the medium multiplied by the integral of the path taken by the light
ray, and OPL= Fermat’s principle is just like Hamiltonian’s principle earlier,

which is why we can adapt the Lagrangian method for Optics.

Letting the path of the light ray be characterized in a 3D Euclidean space with
coordinates x1,x2 and x3, the optical Lagrangian is

Using the Euler-Lagrange equations, we can solve for x1,x2, and x3. You can also use the
Hamiltonian method, but it involves using the Legendre transforms to change the
function into a function based on momentum. The Euler-Lagrange equation can be used
to derive Snell’s law of refraction and the law of specular reflection. It can also give
insights into rays and wavefronts, as well as phase diagrams.

Unfortunately I am not able to comprehend the calculations involved in obtaining

these laws and equations, as I have to regrettably admit that Optics in my worst subject in
Physics(I just don’t like physics geometry), and thus I was unable to put in some
problems and solutions.
The Hamiltonian Method
Energy is defined as the quantity

This expression is equivalent to our normal mechanical energy of the system, as

the first term on the RHS of the equation can be calculated to be simply twice the kinetic
energy of the system. As, the Lagrangian is equal to T-V, the RHS is equivalent to 2T-(T-
V)=T+V. The Hamiltonian basically represents the total energy of a system. However, it
differs a little from our earlier representation of energy.

The Hamiltonian is a function of position and momentum of the system. It looks a

little like this:

What this represents is an alternate form of writing the energy of the system. As
for the Lagrangian, it is transformed from a function of position and velocity into a
function of position and momentum by a mathematical concept known as a Legendre
transform, which I was naturally interested in, being an avid mathematician myself. From
the Hamiltonian, we can derive Hamiltonian’s equations through the Legendre transforms.
Hamiltonian’s equations are:

Thus, Hamilton’s equations give us 2 first order differential equations in position

and momentum. Comparing this to the Euler-Lagrange equations which gives us a 2nd
order differential equation, we can see that they both have different methods of solving
and obtaining the equations of motion.
Legendre transform
Let us consider a function F(x). the derivative of F(x) is dF(x)/dx= F’(x), let us
represent it as s(x). Note that we can invert the function s(x) to solve for x as a function
of s, yielding x(s). For example, if F(x)=3x2, then s(x)=3x, then x(s)=s/3.

The purpose of the Legendre transform is to construct another function G(s) as a

function of s such that dG(s)/ds = x(s). G(s) is then called the Legendre transform of F(x).
Now, the most obvious way to find G(s) is to simply integrate this equation. However,
there is another interesting method that does not involve the task of integrating. If such a
G exists, then we add the 2 equations, dF=s dx and dG=x ds to obtain d(F+G) =sdx+xds.
The right-hand side of the equation looks very familiar, doesn’t it? It reminds you of the
product rule of differentiation! Rewriting the RHS, we have d(F+G)=d(sx). This implies
that F+G=sx. (we take the additive constant by 0 by convention). The Legendre
transform of F is therefore G(s)=sx(s)-F(x(s)).

How do we apply this to obtain the Hamiltonian? We can obtain it by forming the
Legendre transform of the Lagrangian with respect to q̇ Replacing G with the
Hamiltonian H, F with the Lagrangian L, x with velocity q̇ and s with momentum p, we
obtain the Hamiltonian earlier:

I learnt the Legendre transform as part of the background of learning about

Hamiltonian mechanics. It really helped me understand how Hamiltonian is based on the
Lagrangian and yet it is a function of position and momentum while the Lagrangian is a
function of position and velocity.
Problems:

Atwood 2: Atwood 3:

Atwood 4:
Hamiltonian Method in quantum Mechanics
I wanted to further my understanding of the Hamiltonian by looking up its
applications in other areas of physics, and my research have led me to quantum physics. I
do not consider myself familiar with quantum mechanics, but I have learnt enough of it
beforehand to understand what is going on.

The original founder of the Hamiltonian mechanics, Sir William Rowan

Hamiltonian, did not actually work on quantum mechanics. The Hamiltonian exists in
quantum mechanics because physicists find the Hamiltonian method useful to express
things in quantum mechanics and named the operator in honor of Sir William Hamilton

The Hamiltonian method to represent the Hamiltonian as an expression for the

total energy of the system is widely used in quantum mechanics. This is because in
quantum mechanics, a particle can be in multiple positions at the same time, following a
probabilistic curve, and position becomes less useful. Thus, physicists use the
Hamiltonian to express the energy in generalized coordinates and their conjugate
momenta. The Hamiltonian in quantum mechanics is an operator, denoted by the function

, where the h with a slash is h/π, where h is Planck’s constant.

This operation produces specific constants, or eigenvalues for the energy of the system.
Erwin Schrodinger used this for the wavefunction to derive the famous Schrodinger’s
equation:

The math behind Schrodinger’s equations was intense. Schrodinger’s equation

gives the evolution of a wave function over time and is used to describe waves. This is so
important in quantum physics because of the wave-particle duality.
Reflecting on the Lagrangian and the Hamiltonian

The Euler-Lagrange equations and the Hamiltonian equations of motion are,

speaking in terms of classical mechanics, equivalent, as they are both developed and
based on Hamiltonian’s principle of stationary action, and it has been shown earlier that
they are equivalent to Newton’s laws of motions. Why do we use these methods instead
of Newton’s laws? From the questions that I have done, I noticed one difference between
those and the normal kind of questions that you can easily solve using F=ma: there were
myriads of forces everywhere.

When situations become a little wacky and the diagram has too many forces, or
the forces related in the question are constantly changing in direction and magnitude,
such as the 2 falling stick problem, consider the forces may not give you anything useful
to solve the question. In these cases, considering the usual energy conversion is also not
very helpful as you cannot really calculate the energy conversion easily. It simplifies the
question a lot to just consider the total energy of the system and using the basic definition
and derivations of energy. For example, in the sliding block problem, the Euler-Lagrange
equations produced only 2 equations that can be solved to produce the acceleration.
However, if you go about it in the F=ma method, you will find yourself looking for more
equations as your current equations cannot be solved for all the unknown variables.
Eventually I found them, but it took me much more time than just doing the Euler-
Lagrange equations, and in total I had to solve 5 equations consisting of energy,
momentum, force and some geometrical aspects of the question.

So what is the difference between the Lagrangian and the Hamiltonian? In terms
of classical mechanics, they are equivalent methods to solving a problem. I found that the
disadvantage of using the Euler-Lagrange equations is that it only produces 2 variables in
2 2nd order differential equations, namely the acceleration to be solved. The Hamiltonian
method generates more equations than the Euler-Lagrange equations, and they are all 1st
order differential equations. However, the Hamiltonian method takes much longer than
the Lagrangian method, as you can tell from my solving of identical problems. The
Hamiltonian took way longer than the Lagrangian to fully solve, as the Hamiltonian
requires you to compute the Lagrangian first, then using a Legendre transform to change
it into a function of momentum.

However, in other areas of physics apart from classical mechanics, the

Hamiltonian method is, as I currently understand, more adaptable and useful than the
Lagrangian method. This is because you can technically use the Hamiltonian to do
everything the Lagrangian method does, and for some cases, like quantum mechanics
where everything is now based on probability, the Hamiltonian method is more helpful as
it calculates based on the total energy of the system. This is also why the entire quantum
mechanics system is based on Hamiltonian’s methods.

Personal self-reflection

This has been a very enlightening learning process. It was both arduous and
enjoyable solving and writing out all the solutions to all the problems I gave myself (I
think I gave myself too much to be honest). It was extremely satisfying watching every
piece of the puzzle coming together as one and assembling all the questions. This
portfolio on the Lagrangian and the Hamiltonian method taught me 5 very important
ideas about physics and mechanics.

First of all, this taught me that there are always more to learn about each topic.
Before I learnt about the Lagrangian and the Hamiltonian, I thought what I learnt before
was already everything. Apparently, dynamics, torque, rotational motion and oscillation
aren’t the only mechanics topics. After I came into contact with Lagrangian and the
Hamiltonian, I realized that there are more than what I have learnt, and that I should
continue to seek out new theorems, new theories and new methods to solve equations.
My next step for mechanics is after this is to continue to learn the applications of calculus
in mechanics, and how to use infinitesimal analysis to solve a question.
 Secondly, I realized that merely learning the Euler-Lagrange equations and
Hamiltonian’s equation is never enough if you want to have a complete understanding of
the methods. You have to learn the theory behind these methods and some of their
continuations and applications in real life. If I had not learnt about Hamiltonian’s
principle of stationary action, or the Legendre transform, my knowledge of the
Hamiltonian and the Lagrangian would be incomplete. Similarly, this applies for other
theorems in other topics too. I must try to figure out where I lack understanding and
supplement my learning.

Thirdly, I learnt that memorizing equations are useless if you don’t understand real
life equations. I had known this originally, but this portfolio just reinforced and
strengthened my belief. What use is there memorizing an equation when you come across
a new problem, and you do not know how to apply the equations? That is why I strived to
do as many problems as I can.

Fourthly, practice makes perfect. Originally when I was doing questions on the
Lagrangian and the Hamiltonian, I kept stumbling, and making calculation mistakes,
being unable to derive the answer. However, as I progressed, my speed became faster and
I learnt how to solve questions more efficiently. This showed to me that the more
questions you do, the higher your understanding of the topic.

Lastly, I learnt that there are different methods to solve the same question. I should
try to learn as many methods to solving a problem as I can, as each method has their own
advantages and disadvantages.