Chapter 20

The Dirac Delta Function

I do not know what I appear to the world; but to myself I seem to have been only like a boy playing on a seashore, and diverting myself now and then by nding a smoother pebble or a prettier shell than ordinary, whilst the great ocean of truth lay all undiscovered before me. - Sir Issac Newton

20.1

Derivative of the Heaviside Function

The Heaviside function H(x) is dened H(x) = 0 1 for x < 0 for x > 0.

The derivative of the Heaviside function is zero for x = 0. At x = 0 the derivative is undened. We will represent the derivative of the Heaviside function by the Dirac delta function, (x). The delta function is zero for x = 0 and innite at the point x = 0. Since the derivative of H(x) is undened, (x) is not a function in the conventional sense of the word. One can derive the properties of the delta function rigorously, but the treatment in this text will be almost entirely heuristic. The Dirac delta function is dened by the properties 0 for x = 0 and (x) = for x = 0

(x) dx = 1.

The second property comes from the fact that (x) represents the derivative of H(x). The Dirac delta function is conceptually pictured in Figure 20.1.

Figure 20.1: The Dirac Delta Function.

637

Let f (x) be a continuous function that vanishes at innity. Consider the integral f (x)(x) dx.

We use integration by parts to evaluate the integral. f (x)(x) dx = f (x)H(x) f (x)H(x) dx = f (x) dx
0

= [f (x)] 0 = f (0)

We assumed that f (x) vanishes at innity in order to use integration by parts to evaluate the integral. However, since the delta function is zero for x = 0, the integrand is nonzero only at x = 0. Thus the behavior of the function at innity should not aect the value of the integral. Thus it is reasonable that f (0) = f (x)(x) dx holds for all continuous functions. By changing variables and noting that (x) is symmetric we can derive a more general formula. f (0) = f ()() d f ( + x)() d f (x) = f ()( x) d f (x) = f ()(x ) d f (x) =

This formula is very important in solving inhomogeneous dierential equations.

20.2

The Delta Function as a Limit

0
1

Consider a function b(x ) dened by b(x ) = for |x| > /2 for |x| < /2.

The graph of b(x 1/10) is shown in Figure 20.2.

10 5 -1 1

Figure 20.2: Graph of b(x 1/10). The Dirac delta function (x) can be thought of as b(x ) in the limit as 0. Note that the delta function so dened satises the properties, 0 for x = 0 (x) dx = 1 (x) = and for x = 0

638

Delayed Limiting Process. When the Dirac delta function appears inside an integral, we can think of the delta function as a delayed limiting process. f (x)(x) dx lim f (x)b(x ) dx.
0

Let f (x) be a continuous function and let F (x) = f (x). We compute the integral of f (x)(x).

f (x)(x) dx = lim

1 /2 f (x) dx 0 /2 1 /2 = lim [F (x)]/2 0 F (/2) F (/2) = lim 0 = F (0) = f (0)

20.3

Higher Dimensions

We can dene a Dirac delta function in n-dimensional Cartesian space, n (x), x Rn . It is dened by the following two properties. n (x) = 0 for x = 0 n (x) dx = 1
n

It is easy to verify, that the n-dimensional Dirac delta function can be written as a product of 1-dimensional Dirac delta functions. n n (x) = (xk )
k=1

20.4

Non-Rectangular Coordinate Systems

We can derive Dirac delta functions in non-rectangular coordinate systems by making a change of variables in the relation, n (x) dx = 1
n

Where the transformation is non-singular, one merely divides the Dirac delta function by the Jacobian of the transformation to the coordinate system. Example 20.4.1 Consider the Dirac delta function in cylindrical coordinates, (r z). The Jacobian is J = r. 2 3 (x x0 ) r dr d dz = 1
0 0

For r0 = 0, the Dirac Delta function is 3 (x x0 ) = 1 (r r0 ) ( 0 ) (z z0 ) r

since it satises the two dening properties. 1 (r r0 ) ( 0 ) (z z0 ) = 0 r 639 for (r z) = (r0 0 z0 )

2 0

1 (r r0 ) ( 0 ) (z z0 ) r dr d dz r (r r0 ) dr =
0

( 0 ) d
0

(z z0 ) dz = 1

1 (r) (z z0 ) 2r since this again satises the two dening properties. 3 (x x0 ) = for (r z) = (0 z0 ) 2 1 1 (r) (z z0 ) r dr d dz = (r) dr d (z z0 ) dz = 1 2r 2 0 0 1 (r) (z z0 ) = 0 2r

For r0 = 0, we have

640

20.5

Exercises

Exercise 20.1 Let f (x) be a function that is continuous except for a jump discontinuity at x = 0. Using a delayed limiting process, show that f (0 ) + f (0+ ) = f (x)(x) dx. 2 Hint, Solution Exercise 20.2 Show that the Dirac delta function is symmetric. (x) = (x) Hint, Solution Exercise 20.3 Show that (cx) = Hint, Solution Exercise 20.4 We will consider the Dirac delta function with a function as on argument, (y(x)). Assume that y(x) has simple zeros at the points {xn }. y(xn ) = 0 y (xn ) = 0 (x) . |c|

Further assume that y(x) has no multiple zeros. (If y(x) has multiple zeros (y(x)) is not well-dened in the same sense that 1/0 is not well-dened.) Prove that (y(x)) = Hint, Solution Exercise 20.5 Justify the identity

(x xn )
n

|y (xn )|

f (x) (n) (x) dx = (1)n f (n) (0)

From this show that

(n) (x) = (1)n (n) (x) Hint, Solution

and

x (n) (x) = n (n1) (x).

Exercise 20.6 Consider x = (x1 . . . xn ) Rn and the curvilinear coordinate system = (1 . . . n ). Show that (x a) = ( ) |J|

where a and are corresponding points in the two coordinate systems and J is the Jacobian of the transformation from x to . x J Hint, Solution

641