There are at least two reasons why mathematicians become interested in representation theory : quantum physics, and numerology! In the early twentieth century, the study of representations of Lie groups was instigated by quantum mechanics and number theory. Here I will present two examples in connection with these two disciplines.
§ The quantum harmonic oscilator. This is a simple but illustrative example of the connection between physics and representation theory of Lie groups. It also provides a snapshot view of the mathematical foundations of quantum mechanics.
Before jumping into the quantum world, let me begin by explaining the classical harmonic oscillator. The classical harmonic oscillator is simply a mass-spring system which is not subject to friction, gravity, or any other external force.

An eighteenth-century physicist would naturally have
been interested in finding a formula for
the position
of the mass in terms of time. We can represent the position by
a one-variable real function
.
In classical mechanics, one can use the fact that the
Hamiltonian (i.e., total energy) of the system is constant
to see that
is the solution to
a simple differential equation:
Here
is the amount of mass,
is the spring constant,
is the momentum,
and
is the total energy of the system (which is a constant).
What is the quantization of the classical harmonic oscillator?
In quantum mechanics we are not supposed to know where a particle
is located exactly! Rather, at any given time we are able to determine the
probability that it lies at a given location. tyle="position: relative;bottom : -1px;"
It follows that the function
should be replaced by a one-parameter family of wave functions.
More precisely,
at any given time
the "position" of the particle is a complex-valued
function
such that
is a probability distribution on the real numbers, i.e.,
Similar to the classical system, the quantum system is governed by an equation involving a Hamiltonian. The difference is that the new Hamiltonian is not a function anymore. In fact it is a linear transformation on the vector space spanned by wave functions. Note that the latter vector space is infinite-dimensional.
Suppose
is the Hamiltonian operator, where
denotes the vector space spanned by wave functions. (Note that in fact
, the space of square-integrable functions on the real line.)
The key difference between classical and quantum mechanics is that
the differential equation obtained by energy conservation is
replaced by the Schrödinger equation:
The Hamiltonian operator
is given in terms of two other
linear transfomations on
: the position operator
,
and the momentum operator
:
An important fact about
and
is that they do not commute. Indeed
we have
is the Planck constant.
This is exactly where representation theory of Lie groups comes into the picture, as in the language of representation theory
the latter equation essentially means that the operators
and
define a representation of
the Heisenberg group on the vector space
.
This representation of the Heisenberg group
is related to the celebrated
Heisenberg uncertainty principle, which states that
we cannot know certain pairs of physical properties (e.g., position and velocity of a particle) up to arbitrary
precision at the same time!
§ Modular functions and automorphic forms. In 1916 Srinivasa Ramanujan, an amateur indian mathematician who turned into
one of the most famous and influential mathematicians of all times, discovered several mysterious properties of a special power series. The power series Ramanujan considered was :
denotes the coefficient of
, then for every two positive integers
and
which are relatively prime we have
.
and
!)
Ramanujan could not prove these mysterious properties, so he stated them as (yet to become famous) conjectures. At least one of these conjectures had to wait more than sixty years to be settled by another big shot named Pierre Deligne.
It turns out that a slightly different way of looking at
the above power series is quite illuminating : If we set
where
,
then the power series can be written as
to the upper half plane
satisfies another mysterious property : if
are integers such that
, then
which are usually called modular forms.
Modular forms are profoundly related to representation theory of Lie groups. Here I can only provide a rough outline of this relationship :
one can form a
matrix from
as follows :
matrices with integer entries and determinant equal to one
is in fact a group (under the usual matrix multiplication),
and is usually denoted by
.
Every element of
acts on
as follows :
on 
is a subgroup of a bigger group : the group
, which consists of
matrices
with real entries whose
determinant is equal to one. The action of
, which is a
Lie group,
on
by Möbius transformations yields a representation
of
on the space of complex-valued functions on
.
Many properties of modular forms (including Ramanujan's conjectures)
are understood by studying (refinements of) this representation.
Ramanujan's observations have flourished to turn into the theory of automorphic forms and the Langlands program, which is currently a very active area of research.