Dirac delta function

“Delta function” redirects here. For other uses, see Delta

function (disambiguation).
In mathematics, the Dirac delta function, or δ func-

1.2

1.0

0.8

0.6

0.4

0.2

0.0

-0.2
-2 -1 0 1 2

Schematic representation of the Dirac delta function by a line

surmounted by an arrow. The height of the arrow is usually used
to specify the value of any multiplicative constant, which will give
the area under the function. The other convention is to write the
area next to the arrowhead.

tion, is a generalized function, or distribution, on the real

number line that is zero everywhere except at zero, with
an integral of one over the entire real line.[1][2][3] The
delta function is sometimes thought of as a hypothetical
function whose graph is an infinitely high, infinitely thin
spike at the origin, with total area one under the spike,
and physically represents the density of an idealized point The Dirac delta function as the limit (in the sense of distributions)
mass or point charge.[4] It was introduced by theoretical of the sequence of zero-centered normal distributions δa (x) =
2 2
physicist Paul Dirac. 1
√ e−x /a as a → 0 .
a π
From a purely mathematical viewpoint, the Dirac delta
is not strictly a function, because any extended-real func-
tion that is equal to zero everywhere but a single point of a sequence of functions having a tall spike at the ori-
must have total integral zero.[5][6] The delta function only gin. The approximating functions of the sequence are
makes sense as a mathematical object when it appears thus “approximate” or “nascent” delta functions.
inside an integral. From this perspective the Dirac delta In the context of signal processing the delta function
can usually be manipulated as though it were a function. is often referred to as the unit impulse symbol (or
The formal rules obeyed by this “function” are part of the function).[8] Its discrete analog is the Kronecker delta
operational calculus, a standard tool kit of physics and function, which is usually defined on a discrete domain
engineering. The operational calculus, and in particular and takes values 0 and 1.
the delta function, was viewed with suspicion by math-
ematicians of the early 20th century, until a satisfactory
rigorous theory was introduced by Laurent Schwartz in
the 1950s.[7] Formally, the delta function must be defined 1 Overview
as the distribution that corresponds to a probability mea-
sure supported at the origin. In many applications, the The graph of the delta function is usually thought of as
Dirac delta is regarded as a kind of limit (a weak limit) following the whole x-axis and the positive y-axis. De-

1
2 2 HISTORY

spite its name, the delta function is not truly a function,

at least not a usual one with range in real numbers. For ∫ ∞ (∫ ∞ )
example, the objects f(x) = δ(x) and g(x) = 0 are equal f (x) = 1 e ipx
e −ipα
f (α) dα dp
everywhere except at x = 0 yet have integrals that are dif- 2π −∞ −∞
∫ ∞ (∫ ∞ ) ∫ ∞
ferent. According to Lebesgue integration theory, if f 1 ipx −ipα
and g are functions such that f = g almost everywhere, = e e dp f (α) dα = δ(x − α)f (α) d
2π −∞ −∞ −∞
then f is integrable if and only if g is integrable and the
integrals of f and g are identical. Rigorous treatment of where the δ-function is expressed as:
the Dirac delta requires measure theory or the theory of
distributions.
∫ ∞
The Dirac delta is used to model a tall narrow spike func- δ(x − α) = 1 eip(x−α) dp .
tion (an impulse), and other similar abstractions such as 2π −∞
a point charge, point mass or electron point. For exam-
ple, to calculate the dynamics of a baseball being hit by a A rigorous interpretation of the exponential form and
bat, one can approximate the force of the bat hitting the the various limitations upon the function f necessary
baseball by a delta function. In doing so, one not only for its application extended over several centuries. The
simplifies the equations, but one also is able to calculate problems with a classical interpretation are explained as
[15]
the motion of the baseball by only considering the total follows:
impulse of the bat against the ball rather than requiring
knowledge of the details of how the bat transferred en- The greatest drawback of the classical Fourier
ergy to the ball. transformation is a rather narrow class of func-
In applied mathematics, the delta function is often ma- tions (originals) for which it can be effectively
nipulated as a kind of limit (a weak limit) of a sequence computed. Namely, it is necessary that these
of functions, each member of which has a tall spike at the functions decrease sufficiently rapidly to zero
origin: for example, a sequence of Gaussian distributions (in the neighborhood of infinity) in order to en-
centered at the origin with variance tending to zero. sure the existence of the Fourier integral. For
example, the Fourier transform of such simple
functions as polynomials does not exist in the
classical sense. The extension of the classical
2 History Fourier transformation to distributions consid-
erably enlarged the class of functions that could
Joseph Fourier presented what is now called the Fourier be transformed and this removed many obsta-
integral theorem in his treatise Théorie analytique de la cles.
chaleur in the form:[9]

Further developments included generalization of the

∫ ∞ ∫ ∞ Fourier integral, “beginning with Plancherel’s pathbreak-
1
f (x) = dαf (α) dp cos(px − pα) , ing L2 -theory (1910), continuing with Wiener’s and
2π −∞ −∞
Bochner’s works (around 1930) and culminating with the
which is tantamount to the introduction of the δ-function amalgamation into L. Schwartz’s theory of distributions
in the form:[10] (1945) ...”,[16] and leading to the formal development of
the Dirac delta function.
∫ ∞
An infinitesimal formula for an infinitely tall, unit im-
1 pulse delta function (infinitesimal version of Cauchy dis-
δ(x − α) = dp cos(px − pα) .
2π −∞ tribution) explicitly appears in an 1827 text of Augustin
Louis Cauchy.[17] Siméon Denis Poisson considered the
Later, Augustin Cauchy expressed the theorem using issue in connection with the study of wave propagation
exponentials:[11][12] as did Gustav Kirchhoff somewhat later. Kirchhoff and
Hermann von Helmholtz also introduced the unit impulse
∫ ∞ (∫ ∞ ) as a limit of Gaussians, which also corresponded to Lord
1 ipx −ipα Kelvin's notion of a point heat source. At the end of the
f (x) = e e f (α) dα dp.
2π −∞ −∞ 19th century, Oliver Heaviside used formal Fourier series
to manipulate the unit impulse.[18] The Dirac delta func-
Cauchy pointed out that in some circumstances the order tion as such was introduced as a “convenient notation” by
of integration in this result was significant.[13][14] Paul Dirac in his influential 1930 book The Principles of
As justified using the theory of distributions, the Cauchy Quantum Mechanics.[19] He called it the “delta function”
equation can be rearranged to resemble Fourier’s original since he used it as a continuous analogue of the discrete
formulation and expose the δ-function as: Kronecker delta.
3.2 As a distribution 3

3 Definitions This means that H(x) is the integral of the cumulative

indicator function 1₍₋∞, x] with respect to the measure δ;
The Dirac delta can be loosely thought of as a function on to wit,
the real line which is zero everywhere except at the origin,
where it is infinite, ∫
H(x) = 1(−∞,x] (t) δ{dt} = δ(−∞, x].
{ R
+∞, x = 0
δ(x) = Thus in particular the integral of the delta function against
0, x ̸= 0 a continuous function can be properly understood as a
and which is also constrained to satisfy the identity Stieltjes integral:[24]

∫∞
−∞
δ(x) dx = 1. [20] ∫ ∞ ∫ ∞
f (x)δ{dx} = f (x) dH(x).
This is merely a heuristic characterization. The Dirac −∞ −∞
delta is not a function in the traditional sense as no func-
All higher moments of δ are zero. In particular,
tion defined on the real numbers has these properties.[19]
characteristic function and moment generating function
The Dirac delta function can be rigorously defined either
are both equal to one.
as a distribution or as a measure.

3.1 As a measure 3.2 As a distribution

One way to rigorously capture the notion of the Dirac
delta function is to define a measure, which accepts as an
argument a subset A of the real line R, and returns δ(A) =
1 if 0 ∈ A, and δ(A) = 0 otherwise.[21] If the delta function
is conceptualized as modeling an idealized point mass at
0, then δ(A) represents the mass contained in the set A.
One may then define the integral against δ as the integral
of a function against this mass distribution. Formally, the
Lebesgue integral provides the necessary analytic device.
The Lebesgue integral with respect to the measure δ sat-
isfies
∫ ∞
f (x) δ{dx} = f (0)
−∞
In the theory of distributions a generalized function is
thought of not as a function itself, but only in relation
to how it affects other functions when it is “integrated”
against them. In keeping with this philosophy, to de-
fine the delta function properly, it is enough to say what
the “integral” of the delta function against a sufficiently
“good” test function is. If the delta function is already
understood as a measure, then the Lebesgue integral of a
test function against that measure supplies the necessary
integral.
A typical space of test functions consists of all smooth
functions on R with compact support. As a distribution,
the Dirac delta is a linear functional on the space of test
functions and is defined by[25]

for all continuous compactly supported functions f. The

measure δ is not absolutely continuous with respect to
the Lebesgue measure — in fact, it is a singular measure. for every test function φ.
Consequently, the delta measure has no Radon–Nikodym
derivative — no true function for which the property For δ to be properly a distribution, it must be continu-
ous in a suitable topology on the space of test functions.
In general, for a linear functional S on the space of test
∫ ∞
functions to define a distribution, it is necessary and suf-
f (x)δ(x) dx = f (0) ficient that, for every positive integer N there is an integer
−∞
MN and a constant CN such that for every test function
holds.[22] As a result, the latter notation is a conve- φ, one has the inequality[26]
nient abuse of notation, and not a standard (Riemann or
Lebesgue) integral.
As a probability measure on R, the delta measure is char- ∑
MN

acterized by its cumulative distribution function, which is |S[φ]| ≤ C N sup |φ(k) (x)|.
k=0 x∈[−N,N ]
the unit step function[23]
With the δ distribution, one has such an inequality (with
{ CN = 1) with MN = 0 for all N. Thus δ is a distribution of
1 ifx ≥ 0 order zero. It is, furthermore, a distribution with compact
H(x) =
0 ifx < 0. support (the support being {0}).
4 4 PROPERTIES

The delta distribution can also be defined in a number product of distributions can only be defined under quite
of equivalent ways. For instance, it is the distributional narrow circumstances.[28]
derivative of the Heaviside step function. This means The notion of a Dirac measure makes sense on any
that, for every test function φ, one has set.[21] Thus if X is a set, x0 ∈ X is a marked point, and
Σ is any sigma algebra of subsets of X, then the measure
∫ ∞ defined on sets A ∈ Σ by
δ[φ] = − φ′ (x)H(x) dx.
−∞ {
1 ifx0 ∈ A
Intuitively, if integration by parts were permitted, then δx0 (A) =
0 ifx0 ∈ /A
the latter integral should simplify to
is the delta measure or unit mass concentrated at x0 .
∫ ∞ ∫ ∞ Another common generalization of the delta function is
φ(x)H ′ (x) dx = φ(x)δ(x) dx, to a differentiable manifold where most of its proper-
−∞ −∞ ties as a distribution can also be exploited because of the
and indeed, a form of integration by parts is permitted for differentiable structure. The delta function on a manifold
the Stieltjes integral, and in that case one does have M centered at the point x0 ∈ M is defined as the following
distribution:

∫ ∞ ∫ ∞
− φ′ (x)H(x) dx = φ(x) dH(x).
−∞ −∞ for all compactly supported smooth real-valued functions
φ on M.[29] A common special case of this construction
In the context of measure theory, the Dirac measure gives
is when M is an open set in the Euclidean space Rn .
rise to a distribution by integration. Conversely, equation
(1) defines a Daniell integral on the space of all com- On a locally compact Hausdorff space X, the Dirac delta
pactly supported continuous functions φ which, by the measure concentrated at a point x is the Radon measure
Riesz representation theorem, can be represented as the associated with the Daniell integral (3) on compactly sup-
Lebesgue integral of φ with respect to some Radon mea- ported continuous functions φ. At this level of generality,
sure. calculus as such is no longer possible, however a variety
of techniques from abstract analysis are available. For
Generally, when the term "Dirac delta function" is used,
instance, the mapping x0 7→ δx0 is a continuous embed-
it is in the sense of distributions rather than measures, the
ding of X into the space of finite Radon measures on X,
Dirac measure being among several terms for the corre-
equipped with its vague topology. Moreover, the convex
sponding notion in measure theory. Some sources may
hull of the image of X under this embedding is dense in
also use the term Dirac delta distribution.
the space of probability measures on X.[30]

3.3 Generalizations
4 Properties
The delta function can be defined in n-dimensional
Euclidean space Rn as the measure such that 4.1 Scaling and symmetry

∫ The delta function satisfies the following scaling property

f (x)δ{dx} = f (0) for a non-zero scalar α:[31]
Rn
∫ ∞ ∫ ∞
for every compactly supported continuous function f. As du 1
δ(αx) dx = δ(u) =
a measure, the n-dimensional delta function is the product −∞ −∞ |α| |α|
measure of the 1-dimensional delta functions in each vari-
and so
able separately. Thus, formally, with x = (x1 , x2 , ..., xn),
[8]
one has

In particular, the delta function is an even distribution, in

the sense that
The delta function can also be defined in the sense of
distributions exactly as above in the one-dimensional
δ(−x) = δ(x)
case.[27] However, despite widespread use in engineering
contexts, (2) should be manipulated with care, since the which is homogeneous of degree −1.
4.5 Properties in n dimensions 5

4.2 Algebraic properties the g′ = 0 point. This distribution satisfies δ(g(x)) = 0 if g

is nowhere zero, and otherwise if g has a real root at x0 ,
The distributional product of δ with x is equal to zero: then

xδ(x) = 0. δ(x − x0 )
δ(g(x)) = .
|g ′ (x0 )|
Conversely, if xf(x) = xg(x), where f and g are distribu-
tions, then It is natural therefore to define the composition δ(g(x)) for
continuously differentiable functions g by

f (x) = g(x) + cδ(x) ∑ δ(x − xi )

δ(g(x)) =
for some constant c. [32]
i
|g ′ (xi )|

where the sum extends over all roots of g(x), which are
4.3 Translation assumed to be simple.[34] Thus, for example

The integral of the time-delayed Dirac delta is given by:

( ) 1 [ ]
δ x2 − α2 = δ (x + α) + δ (x − α) .
∫ 2|α|
∞
f (t)δ(t − T ) dt = f (T ). In the integral form the generalized scaling property may
−∞
be written as
This is sometimes referred to as the sifting property[33] or
the sampling property. The delta function is said to “sift ∫
∞ ∑ f (xi )
out” the value at t = T. f (x) δ(g(x)) dx = .
−∞ |g ′ (xi )|
It follows that the effect of convolving a function f(t) with i
the time-delayed Dirac delta is to time-delay f(t) by the
same amount:
4.5 Properties in n dimensions
The delta distribution in an n-dimensional space satisfies
the following scaling property instead:
This holds under the precise condition that f be a
tempered distribution (see the discussion of the Fourier
transform below). As a special case, for instance, we have δ(αx) = |α|−n δ(x)
the identity (understood in the distribution sense)
so that δ is a homogeneous distribution of degree −n. Un-
∫ ∞ der any reflection or rotation ρ, the delta function is in-
δ(ξ − x)δ(x − η) dx = δ(ξ − η). variant:
−∞

4.4 Composition with a function δ(ρx) = δ(x).

As in the one-variable case, it is possible to define the

More generally, the delta distribution may be composed
composition of δ with a bi-Lipschitz function[35] g: Rn
with a smooth function g(x) in such a way that the familiar
→ Rn uniquely so that the identity
change of variables formula holds, that

∫ ∫
∫ ∫
( ) ( ) ′
δ(g(x)) f (g(x)) | det g (x)| dx = δ(u)f (u) du
δ g(x) f g(x) |g ′ (x)| dx = δ(u)f (u) du Rn g(Rn )
R g(R)

provided that g is a continuously differentiable function
with g′ nowhere zero.[34] That is, there is a unique way
to assign meaning to the distribution δ ◦ g so that this
identity holds for all compactly supported test functions
f. Therefore, the domain must be broken up to exclude
for all compactly supported functions f.
Using the coarea formula from geometric measure the-
ory, one can also define the composition of the delta func-
tion with a submersion from one Euclidean space to an-
other one of different dimension; the result is a type of
6 5 FOURIER TRANSFORM

current. In the special case of a continuously differen-

tiable function g: Rn → R such that the gradient of g is
nowhere zero, the following identity holds[36] S ∗ δ = S.

That is to say that δ is an identity element for the con-

∫ ∫
f (x) volution on tempered distributions, and in fact the space
f (x) δ(g(x)) dx = dσ(x)
Rn g −1 (0) |∇g| of compactly supported distributions under convolution
is an associative algebra with identity the delta function.
where the integral on the right is over g−1 (0), the (n − This property is fundamental in signal processing, as con-
1)-dimensional surface defined by g(x) = 0 with respect volution with a tempered distribution is a linear time-
to the Minkowski content measure. This is known as a invariant system, and applying the linear time-invariant
simple layer integral. system measures its impulse response. The impulse re-
More generally, if S is a smooth hypersurface of Rn , then sponse can be computed to any desired degree of accu-
we can associate to S the distribution that integrates any racy by choosing a suitable approximation for δ, and once
compactly supported smooth function g over S: it is known, it characterizes the system completely. See
LTI system theory:Impulse response and convolution.
∫ The inverse Fourier transform of the tempered distribu-
δS [g] = g(s) dσ(s) tion f(ξ) = 1 is the delta function. Formally, this is ex-
S pressed
where σ is the hypersurface measure associated to S. This
generalization is associated with the potential theory of ∫ ∞
simple layer potentials on S. If D is a domain in Rn with 1 · e2πixξ dξ = δ(x)
smooth boundary S, then δS is equal to the normal deriva- −∞
tive of the indicator function of D in the distribution
and more rigorously, it follows since
sense:

∫ ∫
∂1D (x) ⟨1, f ∨ ⟩ = f (0) = ⟨δ, f ⟩
− g(x) dx = g(s) dσ(s),
R n ∂n S
for all Schwartz functions f.
where n is the outward normal.[37][38] For a proof, see e.g.
the article on the surface delta function. In these terms, the delta function provides a suggestive
statement of the orthogonality property of the Fourier
kernel on R. Formally, one has
5 Fourier transform
∫ ∞ ∫ ∞
[ i2πξ2 t ]∗
The delta function is a tempered distribution, and there- e i2πξ1 t
e e−i2π(ξ2 −ξ1 )t dt = δ(ξ2 −ξ1 ).
dt =
−∞ −∞
fore it has a well-defined Fourier transform. Formally,
one finds[39] This is, of course, shorthand for the assertion that the
Fourier transform of the tempered distribution
∫ ∞
δ̂(ξ) = e−2πixξ δ(x) dx = 1.
−∞ f (t) = ei2πξ1 t
Properly speaking, the Fourier transform of a distribu-
tion is defined by imposing self-adjointness of the Fourier is
transform under the duality pairing ⟨·, ·⟩ of tempered dis-
tributions with Schwartz functions. Thus δ̂ is defined as
the unique tempered distribution satisfying fˆ(ξ2 ) = δ(ξ1 − ξ2 )

which again follows by imposing self-adjointness of the

⟨δ̂, φ⟩ = ⟨δ, φ̂⟩ Fourier transform.
By analytic continuation of the Fourier transform, the
for all Schwartz functions φ. And indeed it follows from
Laplace transform of the delta function is found to be[40]
this that δ̂ = 1.
As a result of this identity, the convolution of the delta
∫ ∞
function with any other tempered distribution S is simply
S: δ(t − a)e−st dt = e−sa .
0
6.1 Higher dimensions 7

6 Distributional derivatives 6.1 Higher dimensions

The distributional derivative of the Dirac delta distribu- More generally, onn an open set U in the n-dimensional
tion is the distribution δ′ defined on compactly supported Euclidean space R , the Dirac[45]
delta distribution centered
smooth test functions φ by [41] at a point a ∈ U is defined by

δ ′ [φ] = −δ[φ′ ] = −φ′ (0). δa [φ] = φ(a)

The first equality here is a kind of integration by parts, for all φ ∈ S(U), the space of all smooth compactly sup-
for if δ were a true function then ported functions on U. If α = (α1 , ..., αn) is any multi-
index and ∂α denotes the associated mixed partial deriva-
∫ ∞ ∫ ∞ tive operator, then the αth derivative ∂α δa of δa is given
by[45]
δ ′ (x)φ(x) dx = − δ(x)φ′ (x) dx.
−∞ −∞

The k-th derivative of δ is defined similarly as the distri- 
⟨∂ α δa , φ⟩ = (−1)|α| ⟨δa , ∂ α φ⟩ = (−1)|α| ∂ α φ(x) for all φ ∈ S(U ).
bution given on test functions by x=a
That is, the αth derivative of δa is the distribution whose
value on any test function φ is the αth derivative of φ at
δ (k) [φ] = (−1)k φ(k) (0). a (with the appropriate positive or negative sign).
In particular, δ is an infinitely differentiable distribution. The first partial derivatives of the delta function are
The first derivative of the delta function is the distribu- thought of as double layers along the coordinate planes.
tional limit of the difference quotients:[42] More generally, the normal derivative of a simple layer
supported on a surface is a double layer supported on that
surface, and represents a laminar magnetic monopole.
δ(x + h) − δ(x) Higher derivatives of the delta function are known in
δ ′ (x) = lim .
h→0 h physics as multipoles.
More properly, one has Higher derivatives enter into mathematics naturally as the
building blocks for the complete structure of distributions
1 with point support. If S is any distribution on U supported
δ ′ = lim (τh δ − δ) on the set {a} consisting of a single point, then there is
h→0 h
an integer m and coefficients cα such that[46]
where τh is the translation operator, defined on functions
by τhφ(x) = φ(x + h), and on a distribution S by
∑
S= c α ∂ α δa .
|α|≤m
(τh S)[φ] = S[τ−h φ].
In the theory of electromagnetism, the first derivative of
the delta function represents a point magnetic dipole situ- 7 Representations of the delta
ated at the origin. Accordingly, it is referred to as a dipole
or the doublet function.[43]
function
The derivative of the delta function satisfies a number of The delta function can be viewed as the limit of a se-
basic properties, including: quence of functions
d d
δ(−x) = δ(x)
dx dx δ(x) = lim+ ηε (x),
[44] ε→0
′ ′
δ (−x) = −δ (x) where ηε(x) is sometimes called a nascent delta func-
′ tion. This limit is meant in a weak sense: either that
xδ (x) = −δ(x).

Furthermore, the convolution of δ′ with a compactly sup-

ported smooth function f is for all continuous functions f having compact support, or
that this limit holds for all smooth functions f with com-
pact support. The difference between these two slightly
δ′ ∗ f = δ ∗ f ′ = f ′ ,
different modes of weak convergence is often subtle: the
which follows from the properties of the distributional former is convergence in the vague topology of measures,
derivative of a convolution. and the latter is convergence in the sense of distributions.
8 7 REPRESENTATIONS OF THE DELTA FUNCTION

7.1 Approximations to the identity 7.2 Probabilistic considerations

Typically a nascent delta function ηε can be constructed In the context of probability theory, it is natural to impose
in the following manner. Let η be an absolutely integrable the additional condition that the initial η1 in an approxi-
function on R of total integral 1, and define mation to the identity should be positive, as such a func-
tion then represents a probability distribution. Convolu-
tion with a probability distribution is sometimes favorable
(x)
−1 because it does not result in overshoot or undershoot, as
ηε (x) = ε η .
ε the output is a convex combination of the input values,
and thus falls between the maximum and minimum of
In n dimensions, one uses instead the scaling the input function. Taking η1 to be any probability dis-
tribution at all, and letting ηε(x) = η1 (x/ε)/ε as above will
(x) give rise to an approximation to the identity. In general
ηε (x) = ε−n η . this converges more rapidly to a delta function if, in ad-
ε
dition, η has mean 0 and has small higher moments. For
Then a simple change of variables shows that ηε also has instance, if η1 is the uniform distribution on [−1/2, 1/2],
[50]
integral 1.[47] One shows easily that (5) holds for all con- also known as the rectangular function, then:
tinuous compactly supported functions f, and so ηε con-
verges weakly to δ in the sense of measures. {
1 (x) 1
, − 2ε < x < 2ε
The ηε constructed in this way are known as an ap- ηε (x) = rect = ε
proximation to the identity.[48] This terminology is be- ε ε 0, otherwise.
cause the space L1 (R) of absolutely integrable functions
Another example is with the Wigner semicircle distribu-
is closed under the operation of convolution of functions:
tion
f ∗ g ∈ L1 (R) whenever f and g are in L1 (R). However,
there is no identity in L1 (R) for the convolution product:
no element h such that f ∗ h = f for all f. Nevertheless, { √
2
ε2 − x2 , −ε < x < ε
the sequence ηε does approximate such an identity in the ηε (x) = πε2
sense that 0, otherwise

This is continuous and compactly supported, but not a

mollifier because it is not smooth.
f ∗ ηε → f as ε → 0.

This limit holds in the sense of mean convergence (con- 7.3 Semigroups
vergence in L1 ). Further conditions on the ηε, for in-
stance that it be a mollifier associated to a compactly sup- Nascent delta functions often arise as convolution
ported function,[49] are needed to ensure pointwise con- semigroups. This amounts to the further constraint that
vergence almost everywhere. the convolution of ηε with ηδ must satisfy
If the initial η = η1 is itself smooth and compactly sup-
ported then the sequence is called a mollifier. The stan-
dard mollifier is obtained by choosing η to be a suitably ηε ∗ ηδ = ηε+δ
normalized bump function, for instance for all ε, δ > 0. Convolution semigroups in L1 that form
a nascent delta function are always an approximation to
{ the identity in the above sense, however the semigroup
− 1−|x|
1
e 2
if |x| < 1 condition is quite a strong restriction.
η(x) =
0 if |x| ≥ 1. In practice, semigroups approximating the delta function
arise as fundamental solutions or Green’s functions to
In some situations such as numerical analysis, a piecewise physically motivated elliptic or parabolic partial differ-
linear approximation to the identity is desirable. This can ential equations. In the context of applied mathematics,
be obtained by taking η1 to be a hat function. With this semigroups arise as the output of a linear time-invariant
choice of η1 , one has system. Abstractly, if A is a linear operator acting on
functions of x, then a convolution semigroup arises by
( x ) solving the initial value problem
 
ηε (x) = ε−1 max 1 −   , 0
ε
{∂
which are all continuous and compactly supported, al- ∂t η(t, x) = Aη(t, x), t>0
though not smooth and so not a mollifier. lim+
η(t, x) = δ(x)
t→0
7.4 Oscillatory integrals 9

in which the limit is as usual understood in the weak where the operator is rigorously defined as the Fourier
sense. Setting ηε(x) = η(ε, x) gives the associated nascent multiplier
delta function.
Some examples of physically important convolution [( ) 12 ]
2
semigroups arising from such a fundamental solution in- F ∂
− 2 f (ξ) = |2πξ|Ff (ξ).
clude the following. ∂x

The heat kernel 7.4 Oscillatory integrals

The heat kernel, defined by In areas of physics such as wave propagation and wave
mechanics, the equations involved are hyperbolic and so
may have more singular solutions. As a result, the nascent
1 x2
ηε (x) = √ e− 2ε delta functions that arise as fundamental solutions of the
2πε associated Cauchy problems are generally oscillatory in-
represents the temperature in an infinite wire at time t > tegrals. An example, which comes from a solution of the
[52]
0, if a unit of heat energy is stored at the origin of the Euler–Tricomi equation of transonic gas dynamics, is
wire at time t = 0. This semigroup evolves according to the rescaled Airy function
the one-dimensional heat equation:
( )
ε− 3 Ai xε− 3 .
1 1

∂u 1 ∂2u
= . Although using the Fourier transform, it is easy to see
∂t 2 ∂x2 that this generates a semigroup in some sense, it is not
In probability theory, ηε(x) is a normal distribution of absolutely integrable and so cannot define a semigroup
variance ε and mean 0. It represents the probability den- in the above strong sense. Many nascent delta functions
sity at time t = ε of the position of a particle starting at constructed as oscillatory integrals only converge in the
the origin following a standard Brownian motion. In this sense of distributions (an example is the Dirichlet kernel
context, the semigroup condition is then an expression of below), rather than in the sense of measures.
the Markov property of Brownian motion.
Another example is the Cauchy problem for the wave
In higher-dimensional Euclidean space Rn , the heat ker- equation in R1+1 :[53]
nel is

∂2u
c−2 − ∆u = 0
1 − x·x ∂t2
ηε = e 2ε ,
(2πε)n/2 ∂u
u = 0, =δ fort = 0.
∂t
and has the same physical interpretation, mutatis mutan-
dis. It also represents a nascent delta function in the sense The solution u represents the displacement from equilib-
that ηε → δ in the distribution sense as ε → 0. rium of an infinite elastic string, with an initial distur-
bance at the origin.
The Poisson kernel Other approximations to the identity of this kind include
the sinc function (used widely in electronics and telecom-
The Poisson kernel munications)

∫ ∞ (x) ∫ 1
1 ε 1 1 ε

ηε (x) = = e 2πiξx−|εξ|
dξ ηε (x) = sin = cos(kx) dk
π ε + x2
2
−∞
πx ε 2π − 1ε

is the fundamental solution of the Laplace equation in the and the Bessel function
upper half-plane.[51] It represents the electrostatic poten-
tial in a semi-infinite plate whose potential along the edge ( )
1 x+1
is held at fixed at the delta function. The Poisson kernel ηε (x) = J 1ε .
ε ε
is also closely related to the Cauchy distribution. This
semigroup evolves according to the equation
7.5 Plane wave decomposition
( )1
∂u ∂2 2 One approach to the study of a linear partial differential
=− − 2 u(t, x) equation
∂t ∂x
10 7 REPRESENTATIONS OF THE DELTA FUNCTION

This is essentially a form of the inversion formula for the

Radon transform, because it recovers the value of φ(x)
L[u] = f, from its integrals over hyperplanes. For instance, if n is
where L is a differential operator on Rn , is to seek first a odd and k = 1, then the integral on the right hand side is
fundamental solution, which is a solution of the equation
n+1
∫ ∫ n+1
∫ ∫ ∞
cn ∆x 2
φ(y)|(y−x)·ξ| dωξ dy = cn ∆x2
dωξ |p|R
L[u] = δ. S n−1 S n−1 −∞

When L is particularly simple, this problem can often be where Rφ(ξ, p) is the Radon transform of φ:
resolved using the Fourier transform directly (as in the
case of the Poisson kernel and heat kernel already men- ∫
tioned). For more complicated operators, it is sometimes Rφ(ξ, p) = f (x) dn−1 x.
easier first to consider an equation of the form x·ξ=p

An alternative equivalent expression of the plane wave

decomposition, from Gel'fand & Shilov (1966–1968, I,
L[u] = h
§3.10), is
where h is a plane wave function, meaning that it has the
form ∫
(n − 1)!
δ(x) = (x · ξ)−n dωξ
(2πi)n S n−1
h = h(x · ξ)
for n even, and
for some vector ξ. Such an equation can be resolved
(if the coefficients of L are analytic functions) by the ∫
Cauchy–Kovalevskaya theorem or (if the coefficients of L δ(x) = 1
δ (n−1) (x · ξ) dωξ
are constant) by quadrature. So, if the delta function can 2(2πi)n−1 S n−1
be decomposed into plane waves, then one can in princi-
for n odd.
ple solve linear partial differential equations.
Such a decomposition of the delta function into plane
waves was part of a general technique first introduced 7.6 Fourier kernels
essentially by Johann Radon, and then developed in this
form by Fritz John (1955).[54] Choose k so that n + k is See also: Convergence of Fourier series
an even integer, and for a real number s, put
In the study of Fourier series, a major question consists
 of determining whether and in what sense the Fourier se-
[ ] 

|s|k
nodd
−sk log(−is)  4k!(2πi)n−1 ries associated with a periodic function converges to the
g(s) = Re = function. The nth partial sum of the Fourier series of a
k!(2πi)n 

− |s|k log |s| neven. function f of period 2π is defined by convolution (on the
k!(2πi)n
interval [−π,π]) with the Dirichlet kernel:
Then δ is obtained by applying a power of the Laplacian
to the integral with respect to the unit sphere measure dω
( )
of g(x · ξ) for ξ in the unit sphere S n−1 : ∑
N
sin (N + 12 )x
inx
DN (x) = e = .
sin(x/2)
n=−N
n+k
∫
δ(x) = ∆x 2
g(x · ξ) dωξ . Thus,
S n−1

The Laplacian here is interpreted as a weak derivative,

so that this equation is taken to mean that, for any test ∑
N

function φ, sN (f )(x) = D N ∗ f (x) = an einx

n=−N

∫ n+k
∫ where
φ(x) = φ(y) dy ∆x 2
g((x − y) · ξ) dωξ .
Rn S n−1
∫ π
The result follows from the formula for the Newtonian po- 1
tential (the fundamental solution of Poisson’s equation). an = 2π f (y)e−iny dy.
−π
7.7 Hilbert space theory 11

A fundamental result of elementary Fourier series states

that the Dirichlet kernel tends to the a multiple of the delta ∫ ∞
function as N → ∞. This is interpreted in the distribution ∥f ∥2H 1 = |fˆ(ξ)|2 (1 + |ξ|2 ) dξ < ∞
sense, that −∞

is automatically continuous, and satisfies in particular

∫
sN (f )(0) = DN (x)f (x) dx → 2πf (0)
R
δ[f ] = |f (0)| < C∥f ∥H 1 .
for every compactly supported smooth function f. Thus,
formally one has Thus δ is a bounded linear functional on the Sobolev
space H 1 . Equivalently δ is an element of the continuous
dual space H −1 of H 1 . More generally, in n dimensions,
∑∞
1 one has δ ∈ H −s (Rn ) provided s > n / 2.
δ(x) = einx
2π n=−∞

on the interval [−π,π]. 7.7.1 Spaces of holomorphic functions

In spite of this, the result does not hold for all compactly In complex analysis, the delta function enters via Cauchy’s
supported continuous functions: that is DN does not con- integral formula which asserts that if D is a domain in the
verge weakly in the sense of measures. The lack of con- complex plane with smooth boundary, then
vergence of the Fourier series has led to the introduction
of a variety of summability methods in order to produce
convergence. The method of Cesàro summation leads to I
1 f (ζ) dζ
the Fejér kernel[55] f (z) = , z∈D
2πi ∂D ζ −z

( )2 for all holomorphic functions f in D that are continuous

∑
N −1 on the closure of D. As a result, the delta function δz is
1 1 sin N2x
FN (x) = Dn (x) = . represented on this class of holomorphic functions by the
N N sin x2
n=0 Cauchy integral:
The Fejér kernels tend to the delta function in a stronger
sense that[56] I
1 f (ζ) dζ
δz [f ] = f (z) = .
2πi ζ −z
∫ ∂D

FN (x)f (x) dx → 2πf (0) More generally, let H 2 (∂D) be the Hardy space consist-
R
ing of the closure in L2 (∂D) of all holomorphic functions
for every compactly supported continuous function f. The in D continuous up to the boundary of D. Then functions
implication is that the Fourier series of any continuous in H 2 (∂D) uniquely extend to holomorphic functions in
function is Cesàro summable to the value of the function D, and the Cauchy integral formula continues to hold. In
at every point. particular for z ∈ D, the delta function δz is a continu-
ous linear functional on H 2 (∂D). This is a special case of
the situation in several complex variables in which, for
7.7 Hilbert space theory smooth domains D, the Szegő kernel plays the role of the
Cauchy integral.
The Dirac delta distribution is a densely defined
unbounded linear functional on the Hilbert space L2 of
square integrable functions. Indeed, smooth compactly 7.7.2 Resolutions of the identity
support functions are dense in L2 , and the action of the
delta distribution on such functions is well-defined. In Given a complete orthonormal basis set of functions {φn}
many applications, it is possible to identify subspaces of in a separable Hilbert space, for example, the normalized
L2 and to give a stronger topology on which the delta eigenvectors of a compact self-adjoint operator, any vec-
function defines a bounded linear functional. tor f can be expressed as:

Sobolev spaces
∞
∑
f= α n φn .
The Sobolev embedding theorem for Sobolev spaces on n=1
the real line R implies that any square-integrable function
f such that The coefficients {α } are found as:
12 8 DIRAC COMB

7.8 Infinitesimal delta functions

αn = ⟨φn , f ⟩, Cauchy used an infinitesimal α to write down a unit im-
which may be represented by the notation: pulse, infinitely
∫ tall and narrow Dirac-type delta function
δα satisfying F (x)δα (x) = F (0) in a number of arti-
cles in 1827.[61] Cauchy defined an infinitesimal in Cours
d'Analyse (1827) in terms of a sequence tending to zero.
αn = φ†n f,
Namely, such a null sequence becomes an infinitesimal in
a form of the bra–ket notation of Dirac. Adopting this Cauchy’s and Lazare Carnot's terminology.
[57]

notation, the expansion of f takes the dyadic form:[58] Non-standard analysis allows one to rigorously treat in-
finitesimals. The article by Yamashita (2007) contains a
bibliography on modern Dirac delta functions in the con-
∑∞
( † ) text of an infinitesimal-enriched continuum provided by
f= φn φn f .
the hyperreals. Here the Dirac delta can be given by an
n=1
actual function, having
∫ the property that for every real
Letting I denote the identity operator on the Hilbert function F one has F (x)δα (x) = F (0) as anticipated
space, the expression by Fourier and Cauchy.

∞
∑
I= φn φ†n , 8 Dirac comb
n=1

is called a resolution of the identity. When the Hilbert Main article: Dirac comb
space is the space L2 (D) of square-integrable functions A so-called uniform “pulse train” of Dirac delta mea-
on a domain D, the quantity:

φn φ†n ,

is an integral operator, and the expression for f can be

rewritten as:

∞ ∫
∑
f (x) = (φn (x)φ∗n (ξ)) f (ξ) dξ.
n=1 D

−4T −3T −2T −T 0 T 2T 3T 4T

The right-hand side converges to f in the L2 sense. It need
not hold in a pointwise sense, even when f is a continuous A Dirac comb is an infinite series of Dirac delta functions spaced
function. Nevertheless, it is common to abuse notation at intervals of T
and write
sures, which is known as a Dirac comb, or as the Shah
∫ distribution, creates a sampling function, often used in
f (x) = δ(x − ξ)f (ξ) dξ, digital signal processing (DSP) and discrete time signal
analysis. The Dirac comb is given as the infinite sum,
resulting in the representation of the delta function:[59] whose limit is understood in the distribution sense,

∞
∑ ∞
∑
δ(x − ξ) = φn (x)φ∗n (ξ). ∆(x) = δ(x − n),
n=1 n=−∞

With a suitable rigged Hilbert space (Φ, L2 (D), Φ*)

where Φ ⊂ L2 (D) contains all compactly supported
smooth functions, this summation may converge in Φ*,
depending on the properties of the basis φn. In most cases
of practical interest, the orthonormal basis comes from an
integral or differential operator, in which case the series
converges in the distribution sense.[60]
which is a sequence of point masses at each of the inte-
gers.
Up to an overall normalizing constant, the Dirac comb is
equal to its own Fourier transform. This is significant be-
cause if f is any Schwartz function, then the periodization
of f is given by the convolution
 13

∞
∫ ∞
∑
(f ∗ ∆)(x) = f (x − n). f (x)δ(x − x0 ) dx = f (x0 ).
−∞
n=−∞
This exhibits the Kronecker delta function as a discrete
In particular, analog of the Dirac delta function.[64]

b = fˆ∆
(f ∗ ∆)∧ = fˆ∆ 11 Applications
[62]
is precisely the Poisson summation formula.
11.1 Probability theory

9 Sokhotski–Plemelj theorem In probability theory and statistics, the Dirac delta func-
tion is often used to represent a discrete distribution, or
a partially discrete, partially continuous distribution, us-
The Sokhotski–Plemelj theorem, important in quantum
ing a probability density function (which is normally used
mechanics, relates the delta function to the distribution
to represent fully continuous distributions). For example,
p.v.1/x, the Cauchy principal value of the function 1/x,
the probability density function f(x) of a discrete distri-
defined by
bution consisting of points x = {x1 , ..., xn}, with corre-
sponding probabilities p1 , ..., pn, can be written as
⟨ ⟩ ∫
1 φ(x)
p. v. , φ = lim dx.
x ε→0+ |x|>ε x ∑ n
f (x) = pi δ(x − xi ).
Sokhotsky’s formula states that[63] i=1

As another example, consider a distribution which 6/10

1 1 of the time returns a standard normal distribution, and
lim = p. v. ∓ iπδ(x), 4/10 of the time returns exactly the value 3.5 (i.e. a
ε→0+ x ± iε x
partly continuous, partly discrete mixture distribution).
Here the limit is understood in the distribution sense, that The density function of this distribution can be written as
for all compactly supported smooth functions f,

1 x2
∫ ∞ ∫ f (x) = 0.6 √ e− 2 + 0.4 δ(x − 3.5).
f (x) f (x) 2π
lim dx = ∓iπf (0)+ lim+ dx.
ε→0+ −∞ x ± iε ε→0 |x|>ε x The delta function is also used in a completely different
way to represent the local time of a diffusion process (like
Brownian motion). The local time of a stochastic process
10 Relationship to the Kronecker B(t) is given by
delta
∫ t
The Kronecker delta δij is the quantity defined by ℓ(x, t) = δ(x − B(s)) ds
0

{ and represents the amount of time that the process spends

1 i=j at the point x in the range of the process. More precisely,
δij = in one dimension this integral can be written
0 i ̸= j

for all integers i, j. This function then satisfies the follow- ∫ t

ing analog of the sifting property: if (ai )i∈Z is any doubly ℓ(x, t) = lim 1 1[x−ε,x+ε] (B(s)) ds
infinite sequence, then ε→0+ 2ε 0

where 1[x₋ε, x₊ε] is the indicator function of the interval

∞
∑ [x−ε, x+ε].
ai δik = ak .
i=−∞
11.2 Quantum mechanics
Similarly, for any real or complex valued continuous
function f on R, the Dirac delta satisfies the sifting prop- We give an example of how the delta function is expedient
erty in quantum mechanics. The wave function of a particle
14 11 APPLICATIONS

gives the probability amplitude of finding a particle within

a given region of space. Wave functions are assumed to
be elements of the Hilbert space L2 of square-integrable P φy = yφy .
functions, and the total probability of finding a particle
within a given interval is the integral of the magnitude of That is, φy are the eigenvectors of P. If the eigenvectors
the wave function squared over the interval. A set {φn} are normalized so that
of wave functions is orthonormal if they are normalized
by
⟨φy , φy′ ⟩ = δ(y − y ′ )

⟨φn |φm ⟩ = δnm in the distribution sense, then for any test function ψ,

where δ here refers to the Kronecker delta. A set of

∫
orthonormal wave functions is complete in the space of
ψ(x) = c(y)φy (x) dy
square-integrable functions if any wave function ψ can Ω
be expressed as a combination of the φn:
where
∑
ψ= cn φn ,
c(y) = ⟨ψ, φy ⟩.
with cn = ⟨φn |ψ⟩ . Complete orthonormal systems of
wave functions appear naturally as the eigenfunctions of That is, as in the discrete case, there is a resolution of the
the Hamiltonian (of a bound system) in quantum mechan- identity
ics that measures the energy levels, which are called the
eigenvalues. The set of eigenvalues, in this case, is known
∫
as the spectrum of the Hamiltonian. In bra–ket notation,
I= |φy ⟩ ⟨φy | dy
as above, this equality implies the resolution of the iden- Ω
tity:
where the operator-valued integral is again understood in
the weak sense. If the spectrum of P has both continu-
∑
I= |φn ⟩⟨φn |. ous and discrete parts, then the resolution of the identity
involves a summation over the discrete spectrum and an
Here the eigenvalues are assumed to be discrete, but the integral over the continuous spectrum.
set of eigenvalues of an observable may be continuous The delta function also has many more specialized appli-
rather than discrete. An example is the position ob- cations in quantum mechanics, such as the delta potential
servable, Qψ(x) = xψ(x). The spectrum of the position models for a single and double potential well.
(in one dimension) is the entire real line, and is called
a continuous spectrum. However, unlike the Hamilto-
nian, the position operator lacks proper eigenfunctions. 11.3 Structural mechanics
The conventional way to overcome this shortcoming is
to widen the class of available functions by allowing dis- The delta function can be used in structural mechanics
tributions as well: that is, to replace the Hilbert space to describe transient loads or point loads acting on struc-
of quantum mechanics by an appropriate rigged Hilbert tures. The governing equation of a simple mass–spring
space.[65] In this context, the position operator has a com- system excited by a sudden force impulse I at time t = 0
plete set of eigen-distributions, labeled by the points y of can be written
the real line, given by

d2 ξ
φy (x) = δ(x − y). m + kξ = Iδ(t),
dt2
The eigenfunctions of position are denoted by φy = |y⟩ where m is the mass, ξ the deflection and k the spring
in Dirac notation, and are known as position eigenstates. constant.
Similar considerations apply to the eigenstates of the As another example, the equation governing the static
momentum operator, or indeed any other self-adjoint deflection of a slender beam is, according to Euler–
unbounded operator P on the Hilbert space, provided the Bernoulli theory,
spectrum of P is continuous and there are no degenerate
eigenvalues. In that case, there is a set Ω of real num-
bers (the spectrum), and a collection φy of distributions d4 w
indexed by the elements of Ω, such that EI = q(x),
dx4
 15

where EI is the bending stiffness of the beam, w the [6] Duistermaat & Kolk 2010, pp. 37–38
deflection, x the spatial coordinate and q(x) the load dis-
[7] Schwartz 1950, p. 3
tribution. If a beam is loaded by a point force F at x = x0 ,
the load distribution is written [8] Bracewell 1986, Chapter 5

[9] JB Fourier (1822). The Analytical Theory of Heat (En-

glish translation by Alexander Freeman, 1878 ed.). The
q(x) = F δ(x − x0 ). University Press. p. 408., cf p 449 and pp 546–551. The
original French text can be found here.
As integration of the delta function results in the
Heaviside step function, it follows that the static deflec- [10] Hikosaburo Komatsu (2002). “Fourier’s hyperfunc-
tion of a slender beam subject to multiple point loads is tions and Heaviside’s pseudodifferential operators”. In
described by a set of piecewise polynomials. Takahiro Kawai; Keiko Fujita. Microlocal Analysis and
Complex Fourier Analysis. World Scientific. p. 200.
Also a point moment acting on a beam can be described ISBN 981-238-161-9.
by delta functions. Consider two opposing point forces F
at a distance d apart. They then produce a moment M = [11] Tyn Myint-U.; Lokenath Debnath (2007). Linear Partial
Differential Equations for Scientists And Engineers (4th
Fd acting on the beam. Now, let the distance d approach
ed.). Springer. p. 4. ISBN 0-8176-4393-1.
the limit zero, while M is kept constant. The load distri-
bution, assuming a clockwise moment acting at x = 0, is [12] Lokenath Debnath; Dambaru Bhatta (2007). Integral
written Transforms And Their Applications (2nd ed.). CRC Press.
p. 2. ISBN 1-58488-575-0.
( ) [13] Ivor Grattan-Guinness (2009). Convolutions in French
q(x) = lim F δ(x) − F δ(x − d) Mathematics, 1800–1840: From the Calculus and Me-
d→0
( ) chanics to Mathematical Analysis and Mathematical
M M Physics, Volume 2. Birkhäuser. p. 653. ISBN 3-7643-
= lim δ(x) − δ(x − d)
d→0 d d 2238-1.
δ(x) − δ(x − d)
= M lim [14] See, for example, Des intégrales doubles qui se présentent
d→0 d sous une forme indéterminèe
′
= M δ (x).
[15] Dragiša Mitrović; Darko Žubrinić (1998). Fundamentals
Point moments can thus be represented by the derivative of Applied Functional Analysis: Distributions, Sobolev
of the delta function. Integration of the beam equation Spaces. CRC Press. p. 62. ISBN 0-582-24694-6.
again results in piecewise polynomial deflection. [16] Manfred Kracht; Erwin Kreyszig (1989). “On singular in-
tegral operators and generalizations”. In Themistocles M.
Rassias. Topics in Mathematical Analysis: A Volume Ded-
12 See also icated to the Memory of A.L. Cauchy. World Scientific. p.
553. ISBN 9971-5-0666-1.
• Atom (measure theory) [17] Laugwitz 1989, p. 230

• Delta potential [18] A more complete historical account can be found in van
der Pol & Bremmer 1987, §V.4.
• Dirac measure
[19] Dirac 1958, §15
• Fundamental solution
[20] Gel'fand & Shilov 1968, Volume I, §1.1, p. 1
• Green’s function
[21] Rudin 1966, §1.20
• Laplacian of the indicator [22] Hewitt & Stromberg 1963, §19.61

[23] Driggers 2003, p. 2321. See also Bracewell 1986, Chap-

ter 5 for a different interpretation. Other conventions for
13 Notes the assigning the value of the Heaviside function at zero
exist, and some of these are not consistent with what fol-
[1] Dirac 1958, §15 The δ function, p. 58 lows.
[2] Gel'fand & Shilov 1968, Volume I, §§1.1, 1.3 [24] Hewitt & Stromberg 1965, §9.19
[3] Schwartz 1950, p. 3 [25] Strichartz 1994, §2.2

[4] Arfken & Weber 2000, p. 84 [26] Hörmander 1983, Theorem 2.1.5

[5] Vladimirov 1971, §5.1 [27] Hörmander 1983, §3.1

16
[28] Strichartz 1994, §2.3; Hörmander 1983, §8.2
[29] Dieudonné 1972, §17.3.3
[30] Federer 1969, §2.5.19
[31] Strichartz 1994, Problem 2.6.2
[32] Vladimirov 1971, Chapter 2, Example 3(d)
[33] Weisstein, Eric W. “Sifting Property”. MathWorld.
[34] Gel'fand & Shilov 1966–1968, Vol. 1, §II.2.5
[35] Further refinement is possible, namely to submersions, al-
though these require a more involved change of variables
formula.
[36] Hörmander 1983, §6.1
[37] Lange 2012, pp.29–30
[38] Gelfand Shilov, p. 212
[39] In some conventions for the Fourier transform.
[40] Bracewell 1986
[41] Gel'fand & Shilov 1966, p. 26
[42] Gel'fand & Shilov 1966, §2.1
[43] Weisstein, Eric W. “Doublet Function”. MathWorld.
[44] The property follows by applying a test function and inte-
gration by parts.
[45] Hörmander 1983, p. 56
[46] Hörmander 1983, p. 56; Rudin 1991, Theorem 6.25
[47] Stein Weiss, Theorem 1.18
[48] Rudin 1991, §II.6.31
[49] More generally, one only needs η = η1 to have an inte-
grable radially symmetric decreasing rearrangement.
[50] Saichev & Woyczyński 1997, §1.1 The “delta function” as
viewed by a physicist and an engineer, p. 3
[51] Stein & Weiss 1971, §I.1
[52] Vallée & Soares 2004, §7.2
[53] Hörmander 1983, §7.8
[54] See also Courant & Hilbert 1962, §14.
[55] Lang 1997, p. 312
[56] In the terminology of Lang (1997), the Fejér kernel is a
Dirac sequence, whereas the Dirichlet kernel is not.
[57] The development of this section in bra–ket notation is
found in (Levin 2002, Coordinate-space wave functions
and completeness, pp.=109ff)
[58] Davis & Thomson 2000, Perfect operators, p.344
[59] Davis & Thomson 2000, Equation 8.9.11, p. 344
[60] de la Madrid, Bohm & Gadella 2002
14 REFERENCES
[61] See Laugwitz (1989).
[62] Córdoba 1988; Hörmander 1983, §7.2
[63] Vladimirov 1971, §5.7
[64] Hartmann 1997, pp. 154–155
[65] Isham 1995, §6.2
14 References
• Aratyn, Henrik; Rasinariu, Constantin (2006), A
short course in mathematical methods with Maple,
World Scientific, ISBN 981-256-461-6.
• Arfken, G. B.; Weber, H. J. (2000), Mathe-
matical Methods for Physicists (5th ed.), Boston,
Massachusetts: Academic Press, ISBN 978-0-12-
059825-0.
• Bracewell, R. (1986), The Fourier Transform and Its
Applications (2nd ed.), McGraw-Hill.
• Córdoba, A., “La formule sommatoire de Poisson”,
C.R. Acad. Sci. Paris, Series I, 306: 373–376.
• Courant, Richard; Hilbert, David (1962), Meth-
ods of Mathematical Physics, Volume II, Wiley-
Interscience.
• Davis, Howard Ted; Thomson, Kendall T (2000),
Linear algebra and linear operators in engineering
with applications in Mathematica, Academic Press,
ISBN 0-12-206349-X
• Dieudonné, Jean (1976), Treatise on analysis. Vol.
II, New York: Academic Press [Harcourt Brace
Jovanovich Publishers], ISBN 978-0-12-215502-4,
MR 0530406.
• Dieudonné, Jean (1972), Treatise on analysis. Vol.
III, Boston, Massachusetts: Academic Press, MR
0350769
• Dirac, Paul (1958), The Principles of Quantum Me-
chanics (4th ed.), Oxford at the Clarendon Press,
ISBN 978-0-19-852011-5.
• Driggers, Ronald G. (2003), Encyclopedia of Optical
Engineering, CRC Press, ISBN 978-0-8247-0940-
2.
• Duistermaat, Hans; Kolk (2010), Distributions: The-
ory and applications, Springer.
• Federer, Herbert (1969), Geometric measure the-
ory, Die Grundlehren der mathematischen Wis-
senschaften, 153, New York: Springer-Verlag, pp.
xiv+676, ISBN 978-3-540-60656-7, MR 0257325.
• Gel'fand, I. M.; Shilov, G. E. (1966–1968), Gener-
alized functions, 1–5, Academic Press.

• Hartman, William M. (1997), Signals, sound, and
sensation, Springer, ISBN 978-1-56396-283-7.
• Hewitt, E; Stromberg, K (1963), Real and abstract
analysis, Springer-Verlag.
• Hörmander, L. (1983), The analysis of linear par-
tial differential operators I, Grundl. Math. Wis-
senschaft., 256, Springer, ISBN 3-540-12104-8,
MR 0717035.
• Isham, C. J. (1995), Lectures on quantum theory:
mathematical and structural foundations, Imperial
College Press, ISBN 978-81-7764-190-5.
• John, Fritz (1955), Plane waves and spherical means
applied to partial differential equations, Interscience
Publishers, New York-London, MR 0075429.
• Lang, Serge (1997), Undergraduate analysis,
Undergraduate Texts in Mathematics (2nd ed.),
Berlin, New York: Springer-Verlag, ISBN 978-0-
387-94841-6, MR 1476913.
• Lange, Rutger-Jan (2012), “Potential the-
ory, path integrals and the Laplacian of the
indicator”, Journal of High Energy Physics,
Springer, 2012 (11): 29–30, arXiv:1302.0864 ,
Bibcode:2012JHEP...11..032L,
doi:10.1007/JHEP11(2012)032.
• Laugwitz, D. (1989), “Definite values of infinite
sums: aspects of the foundations of infinitesimal
analysis around 1820”, Arch. Hist. Exact Sci., 39
(3): 195–245, doi:10.1007/BF00329867.
• Levin, Frank S. (2002), “Coordinate-space wave
functions and completeness”, An introduction to
quantum theory, Cambridge University Press, pp.
109ff, ISBN 0-521-59841-9
• Li, Y. T.; Wong, R. (2008), “Integral and se-
ries representations of the Dirac delta function”,
Commun. Pure Appl. Anal., 7 (2): 229–247,
doi:10.3934/cpaa.2008.7.229, MR 2373214.
• de la Madrid, R.; Bohm, A.; Gadella,
M. (2002), “Rigged Hilbert Space Treat-
ment of Continuous Spectrum”, Fortschr.
Phys., 50 (2): 185–216, arXiv:quant-
15 External links
ph/0109154 , Bibcode:2002ForPh..50..185D,
doi:10.1002/1521-3978(200203)50:2<185::AID-
PROP185>3.0.CO;2-S.
• McMahon, D. (2005-11-22), “An Introduction to
State Space”, Quantum Mechanics Demystified, A
Self-Teaching Guide, Demystified Series, New York:
McGraw-Hill, p. 108, doi:10.1036/0071455469,
ISBN 0-07-145546-9, retrieved 2008-03-17.
• van der Pol, Balth.; Bremmer, H. (1987), Op-
erational calculus (3rd ed.), New York: Chelsea
17
Publishing Co., ISBN 978-0-8284-0327-6, MR
904873.
• Rudin, W. (1991), Functional Analysis (2nd ed.),
McGraw-Hill, ISBN 0-07-054236-8.
• Soares, Manuel; Vallée, Olivier (2004), Airy func-
tions and applications to physics, London: Imperial
College Press.
• Saichev, A I; Woyczyński, Wojbor Andrzej (1997),
“Chapter1: Basic definitions and operations”,
Distributions in the Physical and Engineering Sci-
ences: Distributional and fractal calculus, inte-
gral transforms, and wavelets, Birkhäuser, ISBN 0-
8176-3924-1
• Schwartz, L. (1950), Théorie des distributions, 1,
Hermann.
• Schwartz, L. (1951), Théorie des distributions, 2,
Hermann.
• Stein, Elias; Weiss, Guido (1971), Introduction to
Fourier Analysis on Euclidean Spaces, Princeton
University Press, ISBN 0-691-08078-X.
• Strichartz, R. (1994), A Guide to Distribution Theory
and Fourier Transforms, CRC Press, ISBN 0-8493-
8273-4.
• Vladimirov, V. S. (1971), Equations of mathemati-
cal physics, Marcel Dekker, ISBN 0-8247-1713-9.
• Weisstein, Eric W. “Delta Function”. MathWorld.
• Yamashita, H. (2006), “Pointwise analysis
of scalar fields: A nonstandard approach”,
Journal of Mathematical Physics, 47 (9):
092301, Bibcode:2006JMP....47i2301Y,
doi:10.1063/1.2339017
• Yamashita, H. (2007), “Comment on “Point-
wise analysis of scalar fields: A nonstan-
dard approach” [J. Math. Phys. 47, 092301
(2006)]", Journal of Mathematical Physics, 48
(8): 084101, Bibcode:2007JMP....48h4101Y,
doi:10.1063/1.2771422
• Hazewinkel, Michiel, ed. (2001), “Delta-function”,
Encyclopedia of Mathematics, Springer, ISBN 978-
1-55608-010-4
• KhanAcademy.org video lesson
• The Dirac Delta function, a tutorial on the Dirac
delta function.
• Video Lectures – Lecture 23, a lecture by Arthur
Mattuck.
18 15 EXTERNAL LINKS

• Dirac Delta Function on PlanetMath

• The Dirac delta measure is a hyperfunction

• We show the existence of a unique solution and ana-
lyze a finite element approximation when the source
term is a Dirac delta measure

• Non-Lebesgue measures on R. Lebesgue-Stieltjes

measure, Dirac delta measure.
 19

16 Text and image sources, contributors, and licenses

16.1 Text
• Dirac delta function Source: https://en.wikipedia.org/wiki/Dirac_delta_function?oldid=734900835 Contributors: AxelBoldt, Zundark,
The Anome, Tarquin, RAE, Ap, XJaM, Camembert, Patrick, Chas zzz brown, Michael Hardy, David Martland, Kidburla, TakuyaMurata,
Looxix~enwiki, Stevenj, Cyan, AugPi, Charles Matthews, Jitse Niesen, Fibonacci, Omegatron, Donarreiskoffer, Robbot, Benwing, Sver-
drup, Henrygb, Lzur, Tea2min, Tosha, Centrx, Giftlite, Smjg, Mikez, BenFrantzDale, Lethe, MathKnight, Jason Quinn, Mboverload, Delta
G, MarkSweep, Fred Stober, Qef, Moxfyre, Jkl, Pyrop, Gianluigi, Mal~enwiki, Paul August, Bender235, Laurascudder, TMA~enwiki,
C S, Rbj, Mindriot, Knucmo2, Msa11usec, PAR, Wtmitchell, Emvee~enwiki, Oleg Alexandrov, Waldir, Btyner, XaosBits, CronoDAS,
Yurik, Jshadias, FlaBot, Tardis, Kri, Haonhien, Chobot, Krishnavedala, Shervinafshar, YurikBot, Wavelength, Spike Wilbury, Hwasung-
mars, Joel7687, Caiyu, Crasshopper, Klutzy, Bnitin, Light current, Zvika, JJL, SmackBot, Rex the first, Slashme, InverseHypercube, Mel-
choir, Zytsef, Skizzik, Oli Filth, Papa November, Nbarth, Colonies Chris, Bob K, Alexxauw, Darkknight911, Daniel.Cardenas, Wvbailey,
SteffenGodskesen, MagnaMopus, NewTestLeper79, Ivanip, JorisvS, Tyrrell McAllister, Mr Stephen, W0lfie, Rattatosk, CBM, WLior,
Joelholdsworth, Yaris678, Kupirijo, Dr.enh, Karl-H, Thijs!bot, Headbomb, Cj67, Eleuther, Bejitunksu, Salgueiro~enwiki, Thenub314,
Mark.przepiora, Sullivan.t.j, David Eppstein, Stevvers, Mårten Berglund, Shapeev, KorgBoy, SimpsonDG, TheSeven, McSly, Copper-
twig, Plasticup, StefanosNikolaou, Policron, Spendabuk, Treisijs, Steel1943, Cuzkatzimhut, VolkovBot, JohnBlackburne, LokiClock,
FatUglyJo, Lingwitt, TXiKiBoT, M gol, YohanN7, SieBot, Stca74, ToePeu.bot, Gerakibot, Czhangrice, Curtdbz, Classicalecon, Rump-
ing, Genie05, Mild Bill Hiccup, Niceguyedc, Baxtrom, Justin545, Meigel, Philtime~enwiki, Brews ohare, Thingg, Qwfp, TimothyRias,
YouRang?, Aaron north, SilvonenBot, Zinger0, Sinawiki, Addbot, Holmansf, Fyrael, Rmartin16, Ronhjones, LaaknorBot, Bob K31416,
Numbo3-bot, Arbitrarily0, Dpaddy, Legobot, Luckas-bot, Yobot, JWroblewski, Ashpilkin, Rubinbot, Jim1138, AdjustShift, Пика Пика,
Christopher.Gordon3, Citation bot, ArthurBot, Xqbot, Bdmy, TinucherianBot II, Abrhm17, Adavis444, Chjoaygame, Sławomir Biały,
Rkr1991, Citation bot 1, Tkuvho, Tal physdancer, Adlerbot, Ashok567, Heptadecagon, , Jujutacular, Ceplusplus, Inferior Olive, Le
Docteur, Jfmantis, EmausBot, Chaohuang, Dcirovic, Slawekb, Solomonfromfinland, Quondum, TonyMath, Zfeinst, Zueignung, EdoBot,
28bot, Galdobr, KlappCK, Joel B. Lewis, Davidcarfi, Helpful Pixie Bot, Bibcode Bot, Sayanpatra, Paolo Lipparini, Solomon7968, Weare-
deltax, IkamusumeFan, MahdiBot, BurkeMtn, ChrisGualtieri, NoFunction, UnameAlreadyTaken, Jbeyerl, Rutgerjanlange, Ihatedirac2k13,
Wyf92, Susan.grayeff, Reader634, 0user007, Pedro H Morgan Pereira, Kjytay and Anonymous: 224

16.2 Images
• File:Dirac_comb.svg Source: https://upload.wikimedia.org/wikipedia/commons/4/49/Dirac_comb.svg License: CC0 Contributors: Own
work Original artist: Krishnavedala
• File:Dirac_distribution_PDF.svg Source: https://upload.wikimedia.org/wikipedia/commons/4/48/Dirac_distribution_PDF.svg License:
CC BY-SA 3.0 Contributors:
• Manually edited version of Image:Dirac_distribution_CDF.svg Original artist:
• Original SVG by Omegatron
• File:Dirac_function_approximation.gif Source: https://upload.wikimedia.org/wikipedia/commons/b/b4/Dirac_function_
approximation.gif License: Public domain Contributors: self-made with MATLAB Original artist: Oleg Alexandrov
• File:Free-to-read_lock_75.svg Source: https://upload.wikimedia.org/wikipedia/commons/8/80/Free-to-read_lock_75.svg License: CC0
Contributors: Adapted from 9px|Open_Access_logo_PLoS_white_green.svg Original artist: This version:Trappist_the_monk (talk)
(Uploads)
• File:German_integral.gif Source: https://upload.wikimedia.org/wikipedia/commons/b/b5/German_integral.gif License: Public domain
Contributors: Selbst erstellt (Font: Adobe Symbol) Original artist: Aleksaner

16.3 Content license

• Creative Commons Attribution-Share Alike 3.0