The Dirac Delta and Green's Functions

George E. Hrabovsky

MAST

Introduction

We all know about particles. Classically they occupy a point, and the mathematics has been built-up around that concept. There is often a need to consider a density (mass, energy, charge,  etc.) for objects being represented as particles. How do we get around the fact that the volume of a particle is zero?

The Idea of a Dirac Delta—A First Approach

How can we develop a mathematical object that has a radius within an arbitrarily close distance of zero? In fact we must force this distance to be indistinguishable from zero—what we call a neighborhood around zero. We must also make sure that this region has some area to it that is more than 0 (positive). Why? There is no point in considering such an object otherwise—the goal is to make something where any density will not become infinite, so its area/volume has to be larger than 0. We are effectively asking to be able to make a concentration of the essence of an object into a very small, but non-zero, region. We can make this a definition of our mathematical object: A positive area concentrated into the neighborhood of zero. The density of this region, its integral, will be 1. This is what we call the Dirac Delta. Some have said of it, “It is higher than the Devil and narrow as Hell.” We will denote this object with the symbol δ(x). Were we to plot this, it would look like this,

retreat2-3_1.gif

Figure DDG-1: The Dirac Delta.

Some careful consideration, which we do not have time for right now, will lead to the following property, for any non-pathological univariate function f(x),

retreat2-3_2.png

DDG-1: Why is this true?

Since we are in the neighborhood of retreat2-3_3.png we can specify this neighborhood by establishing the interval retreat2-3_4.png, for a sufficiently small ε. Since our function is not pathological, it will not blow up at retreat2-3_5.png. By our definition f(x) will not vary anywhere within the neighborhood. Thus retreat2-3_6.png in the neighborhood. We can then write

retreat2-3_7.png

Since retreat2-3_8.png does not change over the interval, it is a constant. so,

retreat2-3_9.png

retreat2-3_10.png

From our original specification,

retreat2-3_11.png

so

retreat2-3_12.png

We can see some examples of this.

retreat2-3_13.png

retreat2-3_14.png

retreat2-3_15.png

retreat2-3_16.png

retreat2-3_17.png

retreat2-3_18.png

retreat2-3_19.png

retreat2-3_20.png

retreat2-3_21.png

retreat2-3_22.png

Many problems lead to a Dirac delta.

Properties of the Delta

The first property is that the Dirac delta doesn’t care about sign.

retreat2-3_23.png

retreat2-3_24.png

Other properties are:

retreat2-3_25.png

retreat2-3_26.png

retreat2-3_27.png

here retreat2-3_28.png are the roots of f(x).

If have an integral of the form,

retreat2-3_29.png

retreat2-3_30.png

we see that this is the equivalent of a difference from the product of the Dirac delta and the initial value of our function, and the product of a new function by the derivative around the initial value of the function. But what the heck is represented in Mathematica by the HeavisideTheta? In normal mathematics we call this the step function.

The Step Function

The step function is defined as follows,

retreat2-3_31.png

Thus the integral above becomes

retreat2-3_32.png

The derivative of an arbitrary step function is the Dirac delta.

retreat2-3_33.png

retreat2-3_34.png

From this we see another way to define the step function, as the integral of a Dirac delta.

retreat2-3_35.png

retreat2-3_36.png

Green’s Functions

Let’s say we have the differential equation

retreat2-3_37.png

We can think of f(x) as the input and retreat2-3_38.png as the output. What happens when the input is a Dirac delta? This converts the differential equation into the form

retreat2-3_39.png

over some interval retreat2-3_40.png. If we know what G is, then we have solved the differential equation.

retreat2-3_41.png

retreat2-3_42.png

If we compare this with (DDG.6), we can see that

retreat2-3_43.png

So, G is what we call a Green’s function, named for the British mathematician George Green (1793 - 1841).

This example is due to Hermann Schultz’s book, “A Theoretical Physics Primer,” 2015, published by Verlag Europa-Lerhmittel. We will examine the falling body problem, where we have one-dimensional motion a(t)=-g. Thinking about it, we need our interval I to be a time interval, (0,T), with T being some later time. The operator is retreat2-3_44.png. We can then write the characteristic equation

retreat2-3_45.png

Integrating this, we have

retreat2-3_46.png

retreat2-3_47.png

retreat2-3_48.png

This gives us an expression for the velocity

retreat2-3_49.png

retreat2-3_50.png

retreat2-3_51.png

retreat2-3_52.png

retreat2-3_53.png

retreat2-3_54.png

retreat2-3_55.png

If, as retreat2-3_56.png we have the condition retreat2-3_57.png then we say that the operator exhibits translational invariance. In this case the left hand term vanishes,

retreat2-3_58.png

we then have a familiar looking solution,

retreat2-3_59.png

where the constant retreat2-3_60.png is the only thing unfamiliar.

Mathematica allows you to work with Green’s functions.

retreat2-3_61.png

retreat2-3_62.png

So let’s try this for the heat operator,

retreat2-3_63.png

retreat2-3_64.png

if we define our spatial region as the entire real line we write

retreat2-3_65.png

retreat2-3_66.png

We can choose b to be arbitrarily close to t,

retreat2-3_67.png

retreat2-3_68.png

We can also choose a value of a close to x,

retreat2-3_69.png

retreat2-3_70.png

We can now plot this,

retreat2-3_71.png

retreat2-3_72.gif

You can see the effect of the Dirac delta, this is why it is called an impulse response.

Using these methods we can get around the issue of infinite densities.

Created with the Wolfram Language

Click here to go back to the quantum mechanics page.

Click here to go back to our home page.