January 4, 2013 9:31 WSPC/103-M3AS 1250054

January 4, 2013 9:31 WSPC/103-M3AS 1250054
Mathematical Models and Methods in Applied Sciences

Vol. 23, No. 3 (2013) 493540
c World Scientic Publishing Company
DOI: 10.1142/S0218202512500546
A NONLOCAL VECTOR CALCULUS,
NONLOCAL VOLUME-CONSTRAINED PROBLEMS,
AND NONLOCAL BALANCE LAWS
QIANG DU
Department of Mathematics,
Pennsylvania State University,
University Park, PA 16802, USA
qdu@math.psu.edu
MAX GUNZBURGER
Department of Scientic Computing,

Florida State University,
Tallahassee FL 32305-4120, USA
gunzburg@fsu.edu
R. B. LEHOUCQ
Sandia National Laboratories,
P. O. Box 5800, MS 1320, Albuquerque,
NM 87185-1320, USA
rblehou@sandia.gov
KUN ZHOU
Department of Mathematics,
Pennsylvania State University,
University Park, PA 16802, USA
zhou@math.psu.edu
Received 7 September 2011
Revised 15 March 2012
Accepted 5 April 2012
Published 26 September 2012
Communicated by I. Fonseca
A vector calculus for nonlocal operators is developed, including the denition of nonlocal
divergence, gradient, and curl operators and the derivation of the corresponding adjoint
operators. Nonlocal analogs of several theorems and identities of the vector calculus for
dierential operators are also presented. Relationships between the nonlocal operators
and their dierential counterparts are established, rst in a distributional sense and
then in a weak sense by considering weighted integrals of the nonlocal adjoint operators.
The operators of the nonlocal calculus are used to dene volume-constrained problems
Corresponding author
493
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
494 Q. Du et al.
that are analogous to elliptic boundary-value problems for dierential operators; this is
demonstrated via some examples. Another application discussed is the posing of abstract
nonlocal balance laws and deriving the corresponding nonlocal eld equations; this is
demonstrated for heat conduction and the peridynamics model for continuum mechanics.
Keywords: Nonlocal operators; vector calculus; volume-constrained problems; balance
laws; peridynamics; nonlocal diusion.
AMS Subject Classication: 26B12, 26B15, 26B20, 45A05, 45P05, 45E99, 46F12
1. Introduction
Our principal goal is to develop a vector calculus for nonlocal operators that mimics
the classical vector calculus for dierential operators. We also show how the nonlocal
vector calculus can be used to dene nonlocal balance laws and volume-constrained
problems that mimic boundary-value problems for dierential operators.
Nonlocal analogs of the divergence, gradient, and curl operators are dened
and the corresponding nonlocal adjoint operators are deduced. Nonlocal analogs
of the Gauss theorem and Greens identities of the vector calculus for dierential
operators are also derived. Relationships between the nonlocal operators and their
dierential counterparts are established. In addition, the nonlocal vector calculus
has an important application to balance laws
a
that are nonlocal in the sense that
subregions not in direct contact may have a nonvanishing interaction. This is accom-
plished by dening a nonlocal ux in terms of interactions between regions having
positive measure, possibly not sharing a common boundary. As a result, the nonlo-
cal vector calculus provides an alternative to standard approaches for circumventing
the technicalities associated with the lack of sucient regularity in local balance
laws.
Preliminary attempts at a nonlocal calculus are found in Refs. 11 and 12 that
include applications to image processing and steady-state diusion, respectively. In
particular, in Ref. 26, which is cited in Ref. 11, a discrete nonlocal divergence and
gradient are introduced within the context of machine learning; see also Refs. 4, 14
and 16 where a discrete calculus is also discussed. However, the discussion in those
papers is limited to scalar problems. In contrast, this paper extends the ideas in
Refs. 11 and 12 to vector and tensor elds and beyond the consideration of image
processing and steady-state diusion. For example, the ideas presented here enable
an abstract formulation of the balance laws of momentum and energy in the peri-
dynamic continuum model
b
for solid mechanics that parallels the classical vector
calculus formulation of the balance laws of elasticity. In fact, the nonlocal vector
a
A balance law postulates that the rate of change of an extensive quantity over any domain is
given by the rate at which that quantity is produced in the domain minus the ux out of the
domain.
b
The peridynamic continuum model was introduced in Refs. 21 and 23; Ref. 24 reviews the peri-
dynamic balance laws of momentum and energy and provides many citations for the peridynamic
model and its applications. See Sec. 7.3 for a brief discussion.
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
Nonlocal Vector Calculus, Nonlocal Volume-Constrained Problems 495
calculus presented in this paper is suciently general that we envisage application
to balance laws beyond those of elasticity, e.g. to the laws of uid mechanics and
electromagnetics.
The paper is organized as follows. The remainder of this section is devoted to
establishing notation. In Sec. 2, the notions of local and nonlocal uxes into or
out of a region are compared and contrasted. The next two sections contain our
principal contributions. In Sec. 3, nonlocal divergence, gradient, and curl operators
are introduced as are the corresponding adjoint operators, several vector identities,
and other results about the operators. The nonlocal vector calculus is developed in
Sec. 4; in particular, nonlocal integral theorems and nonlocal Greens identities are
derived. In Sec. 5, connections between the nonlocal operators and distributional
and weak representations of the associated classical dierential operators are made.
Sections 6 and 7 deal with applications of the nonlocal vector calculus. In Sec. 6,
examples are given of nonlocal volume-constrained problems formulated in terms of
the nonlocal operators. Then, in Sec. 7, a brief review of the conventional notion of
a balance law is provided after which abstract nonlocal balance laws are discussed.
The notion of nonlocal uxes discussed in Sec. 2 is used in developing nonlocal
balance laws and the vector calculus developed in Sec. 4 plays a crucial role in
transforming balance laws into eld equations. Also, in Sec. 7, a brief discussion is
given of the application of our nonlocal vector calculus to the peridynamic theory
for continuum mechanics.
Throughout, wherever it is illuminating, we associate the denitions and results
of the nonlocal vector calculus with the analogous denitions and results of the
classical dierential vector calculus.
1.1. Notation
We have need of two types of functions and two types of nonlocal operators. Point
functions refer to functions dened at points whereas two-point functions refer to
functions dened for pairs of points. Point operators map two-point functions to
point functions whereas two-point operators map point functions to two-point func-
tions so that the nomenclature for operators refer to their ranges. Point and two-
point operators are both nonlocal. Point operators involve integrals of two-point
functions whereas two-point operators explicitly involve point functions evaluated
at two dierent points.
We now make more precise the denitions given above. Let m, k, and n denote
positive integers. Points in R
n
are denoted by the vectors x, y, or z and the natural
Cartesian basis is denoted by e
1
, . . . , e
n
. For any domain

R
n
, functions from

into R
mk
or R
m
or R are referred to as point functions or point mappings and are
denoted by Roman letters, upper-case bold for tensors, lower-case bold for vectors,
and plain face for scalars, respectively, e.g. U(x), u(x), and u(x), respectively.
Functions from

into R
mk
or R
m
or R are referred to as two-point functions
or two-point mappings and are denoted by Greek letters, upper-case bold for tensors,
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
496 Q. Du et al.
lower-case bold for vectors, and plain face for scalars, respectively, e.g. (x, y),
(x, y), and (x, y), respectively. Symmetric and antisymmetric scalar two-point
functions (x, y) satisfy (x, y) = (y, x) and (x, y) = (y, x), respectively,
and similarly for vector and tensor two-point functions.
For the sake of notational simplicity, in much of the rest of the paper we adopt
the following notation:
:= (x, y),
:= (y, x), := (x, y),
:= (y, x),
u := u(x), u
:= u(y), u := u(x), u
:= u(y),
and similarly for other functions.
The dot (or inner) product of two vectors u, v R
m
is denoted by u v R;
the dyad (or outer) product is denoted by u w R
mk
whenever w R
k
; given
a second-order tensor (matrix) U R
km
, the tensor-vector (or matrix-vector)
product is denoted by U v and is given by the vector whose components are the
dot products of the corresponding rows of Uwith v.
c
In R
3
, the cross product of two
vectors u and v is denoted by u v R
3
. The Frobenius product of two second-
order tensors A R
mk
and B R
mk
, denoted by A: B, is given by the sum of
the elementwise product of the two tensors, i.e. A: B =
i=1,...,m,j=1,...,k
A
ij
B
ij
.
The trace of B R
mm
, denoted by tr
_
B
_
, is given by the sum of the diagonal
elements of B.
Inner products in L
2
(
) and L
2
(
) are dened in the usual manner. For

example, for vector functions, we have
_
_
(u, v)
e
=
_
e
u vdx for u(x), v(x) L

2
(
),
(, )
e
=
_
e
_
e
dydx for (x, y), (x, y) L

2
(
)
with analogous expressions involving the Frobenius product and the ordinary prod-
uct for tensor and scalar functions, respectively.
Many of the results in the paper require that domains under consideration
are Lebesgue measurable and functions are Lebesgue integrable with their point-
wise properties interpreted in the Lebesgue measure sense, i.e. as holding almost
everywhere.
2. Local and Nonlocal Fluxes and ActionReaction Principles
A key concept in the development of a vector calculus is the notion of a ux that
accounts for the interaction of points in a domain with points outside the domain.
As a result, the notion of a ux is also fundamental to the understanding of balance
laws in mechanics, heat transfer, and many other settings; see Sec. 7. In the classical
c
In matrix notation, the inner, outer, and matrix-vector products are given by x y = x
T
y,
x y = xy
T
, and U v = Uv, respectively.
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
setting of local interactions, that interaction occurs at the boundary of the domain
whereas, in the nonlocal case, the interaction occurs over volumes external to the
domain. In order to compare and contrast the notion of a nonlocal ux with the
classical local ux, we begin by briey reviewing the latter notion.
2.1. Local uxes
Let
1
R
n
and
2
R
n
denote two disjoint open regions having Lipschitz
boundaries. If
1
and
2
have a nonempty common boundary
12
=
1

2
,
then, for a suitably smooth vector-valued function q(x), the expression
_
12
q n
1
dA (2.1)
represents the classical local ux out of
1
into
2
, where n
1
denotes the unit
normal on
12
pointing outward from
1
and dA denotes a surface measure in
R
n
; q n
1
is referred to as the ux density along
12
in the direction of the normal
vector n
1
to the surface at that point. Because the local ux density q n
1
at
a point is dened in terms of any orientable surface passing through that point,
the ux conveys a notion of direction out of and into regions and is a proxy for
the interaction between
1
and
2
, i.e. the ux is an oriented interaction between
two domains. An important observation is that the local ux from
1
into
2
occurs across their common boundary and that if the two disjoint regions have no
common boundary, then the ux from one to the other is zero. The classical ux
(2.1) is then deemed to be local because there is no interaction between
1
and
2
when separated by a nite distance. The classical ux satises the actionreaction
principle
d
_
12
q n
1
dA +
_
21
q n
2
dA = 0, (2.2)
where, of course,
12
=
21
and n
2
= n
1
denotes the unit normal on
12
pointing outward from
2
. In words, the ux
_
12
q n
1
dA from
1
into
2
across
their common boundary
12
is equal and opposite to the ux
_
21
q n
2
dA from
2
into
1
across that same surface. The vector q is often expressed in terms of an
intensive variable through a constitutive relation.
e
2.2. Nonlocal uxes
For any point x R
n
, for integrable (x, y), we identify
_
e
(x, y)dy
R
n
(2.3)
d
An example is in mechanics for which Newtons third law, i.e. the force exerted upon an object
is equal and opposite to the force exerted by the object, is an actionreaction archetype.
e
For example, if q n
1
denotes the heat ux density, then q is related to the temperature via the
Fourier heat law; see Sec. 7.2.2.
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
498 Q. Du et al.
as the nonlocal ux density
f
at x into a measurable volume
. Because of nonlocality,
x does not necessarily have to belong to the closure of

. The nonlocal ux density
(2.3) has units of an intensive quantity per unit volume, in contrast to the local
ux density q n that has units of an intensive quantity per unit area. In (2.3),
(x, y) may be viewed as a ux density per unit volume; no such notion exists in
the local case because the ux density is not dened in terms of an integral as is
the nonlocal ux density (2.3).
The following result is fundamental to the development of the nonlocal calcu-
lus. Here and throughout, we assume that domains such as
1
,
2
, and

are
measurable.
Theorem 2.1. The following are equivalent:
(i) The antisymmetry of a two-point function
(x, y) = (y, x) for almost all x, y R
n
. (2.4a)
(ii) The alternating or no self-interaction principle
_
e
_
e
(x, y)dydx = 0
R
n
. (2.4b)
(iii) The nonlocal actionreaction principle
_
1
_
2
(x, y)dydx +
_
2
_
1
(x, y)dydx = 0
1
,
2
R
n
. (2.4c)
(iv) Additivity for disjoint regions
_
2
_
R
n
\(
1
2
)
(x, y)dydx
=
_
1
_
R
n
\
1
(x, y)dydx +
_
2
_
R
n
\
2
(x, y)dydx

1
,
2
R
n
such that
1

2
= . (2.4d)
Similar results hold for antisymmetric vector and tensor two-point functions.
Proof. A change of variables under integration followed by change in the order of
integration shows that
_
e
_
e
(x, y)dydx =
_
e
_
e
(y, x)dydx
R
n
,
so that
2
_
e
_
e
(x, y)dydx =
_
e
_
e
((x, y) + (y, x))dydx
R
n
.
f
In Sec. 3.2, we show how can be expressed as an inner product of two vectors so that (2.3)
mimics the local ux density q n.
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
Thus, obviously, (i) implies (ii). This relation also shows that (ii) implies (i) because
if (ii) holds, we have
_
e
_
e
((x, y) + (y, x))dydx = 0
R
n
,
which implies that (x, y) + (y, x) = 0 almost everywhere.
Setting T(
1
,
2
) =
_
1
_
2
(x, y)dydx, we easily see that (2.4a) and (2.4b)
are satised. Then, the equivalence of (ii) and (iii) follows from Proposition 7.1 and
the equivalence of (ii) and (iv) follows from Proposition 7.3.
With (2.3) denoting a nonlocal ux density, for any two open regions
1
R
n
and
2
R
n
, we identify
_
1
_
2
(x, y)dydx (2.5)
as a scalar interaction or nonlocal ux from
1
into
2
. As in the local case, the
nonlocal ux is associated with the notion of an orientation so that nonlocal uxes
are again oriented interactions between two domains.
Because we require no self-interactions, i.e. the ux from a region into itself
vanishes so that (2.4b) holds whenever
1
=
2
, by Theorem 2.1 we equivalently
require that : (
1
2
) (
1
2
) R be an antisymmetric function. By (2.3),
_
2
(x, y)dy is the nonlocal ux density at a point x
1
into the region
2
. As
is the case for the local ux density q n
1
, the nonlocal ux density
_
2
(x, y)dy
is related to an intensive variable through a constitutive relation; see Sec. 7.2.3.
Based on the above discussion, we see that (2.4c) is the nonlocal analogue
of (2.2). In words, (2.4c) states that the ux (or interaction) from
1
into
2
is equal and opposite to the ux (or interaction) from
2
into
1
. The ux is
nonlocal because, by (2.5), the interaction may be nonzero even when the closures
of
1
and
2
have an empty intersection. This is in stark contrast to classical
local interactions for which we have seen that the interaction between
1
and
2
vanishes if their closures have empty intersection, i.e. if they have no common
boundary.
3. Nonlocal Operators
The nonlocal vector calculus developed in Sec. 4 involves nonlocal operators that
mimic the classical local dierential divergence, gradient, and curl operators. An
important distinction between local and nonlocal operators is that the adjoint oper-
ators for the former involve the same operators, i.e. the adjoint of is , of is
, and of is , whereas the adjoint of nonlocal operators involve dierently
dened nonlocal gradient, divergence, and curl operators. See Sec. 3.3 for a further
discussion.
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
500 Q. Du et al.
3.1. Motivating the denitions of nonlocal operators
To provide motivation for the denitions given in Denition 3.1 for the basic oper-
ators of the nonlocal vector calculus, we recall that the Gauss theorem
_
e
qdx =
_
q ndA
R
n
(3.1)
shows that the integral of the divergence of a vector q over any region

R
n
is equal to the ux out of that region. Of course, the latter is the integral over
of the ux density q n. We adopt the same word denition for the nonlocal
divergence, i.e. with T denoting the nonlocal divergence operator, we have that the
integral of the nonlocal divergence of a vector over any region

R
n
is equal
to the ux out of that region. In general, a point x

can interact with all other
points y R
n
so that, by (2.3), the nonlocal ux density out of a point x

is
given by
_
R
n
(x, y)dy.
Then, according to our denition of the nonlocal divergence operator, we have from
(2.5) that
_
e
T()(x)dx =
_
e
_
R
n
(x, y)dydx
R
n
. (3.2)
In (3.1), the ux density q n is obviously related to the vector q, i.e. to the
vector operand of the local dierential divergence operator. In the nonlocal case,
to dene the action of the nonlocal divergence operator on the vector , we have
to analogously determine how the operand of the nonlocal divergence operator
T is related to (x, y), where, because of nonlocality, (x, y) is a vector two-point
function. The information we have in hand to determine this relation is:
because we assume no self-interactions,
(2.4a) requires that the scalar function (x, y) be antisymmetric; (3.3a)
because of our denition of the nonlocal divergence operator,
(3.2) relates T to ; (3.3b)
because we emulate the fact that is a linear operator,
(x, y) and therefore also T() are linear in (x, y). (3.3c)
This information is sucient to apply the Schwartz kernel theorem (see, e.g. Refs. 8,
10 and 20) to determine the relation between and that leads to a denition of
the nonlocal divergence operator. Before justifying this claim, we need to introduce
some notation.
Let
U = [C
(R
n
R
n
)]
k
and V = C
),
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
so that U consists of vector two-point functions whereas V consists of scalar point
functions. The dual spaces are given by
U
= [D
(R
n
R
n
)]
k
and V

= D
),
respectively, where D
denotes the space of distributions. The corresponding duality

pairings are given as follows:
, )
U,U
=
_
R
n
_
R
n
(z, y) (z, y)dzdy and v, u)
V,V
=
_
e
v(x)u(x)dx
for U, U
and v V , u V

, respectively. We also use the product space
U V , its dual (U V )
, and, for U V and (U V )
, the corresponding
duality pairing
, )
UV,(UV )
=
_
e
_
R
n
_
R
n
(x, y, z) (x, y, z)dzdydx.
Specialized to our needs, the Schwartz kernel theorem is given as follows.
Theorem 3.1. (Schwartz kernel theorem) Let T : U V

denote a continuous
linear map. Then, there exists a unique kernel (x, y, z) (U V )
such that
v, T)
V,V
= v, )
(UV ),(UV )
v V, U,
or, using the denitions of the duality pairings,
_
e
v(x)(T)(x)dx =
_
e
_
R
n
_
R
n
(x, y, z) (z, y)v(x)dzdydx
v V, U. (3.4)
Corollary 3.1. Assume that (3.3) holds. The nonlocal ux density per unit volume
(x, y) is uniquely expressed in terms of the vector (x, y) by
(x, y) =
1
2
_
R
n
((x, y, z) (z, y) (y, x, z) (z, x))dz x, y R
n
(3.5)
and the nonlocal divergence operator T is uniquely expressed in terms of the vector
(x, y) by
(T)(x) =
1
2
_
R
n
_
R
n
((x, y, z) (z, y) (y, x, z) (z, x))dzdy
x R
n
. (3.6)
Proof. Because

R
n
is arbitrary, we conclude from (3.4) with v = 1 that
(T)(x) =
_
R
n
_
R
n
(x, y, z) (z, y)dzdy x R
n
. (3.7)
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
502 Q. Du et al.
Comparing (3.2) and (3.7), we see that
(x, y) =
_
R
n
(x, y, z) (z, y)dz x, y R
n
. (3.8)
The antisymmetry of (x, y) with respect to x and y requires that
(y, x, z) (z, x) = (x, y, z) (z, y) x, y, z R
n
, (3.9)
so that (3.5) and (3.6) follow from (3.8) and (3.7), respectively.
The general denition for the nonlocal divergence is thus given by (3.6). We have
found that we can simplify that denition and still develop a nonlocal vector calculus
that meets our needs; however, it is possible that in some applications, retaining
the more general denition (3.6) could prove useful. Our simplifying assumption is
to let
(x, y, z) = 2(x, y)(x z).
Substituting into (3.6), we obtain
(T)(x) =
1
2
_
R
n
_
R
n
((x, y, z) (z, y) (y, x, z) (z, x))dzdy
=
_
R
n
_
R
n
((x, y) (z, y)(x z) (y, x) (z, x)(y z))dzdy
=
_
R
n
((x, y) (x, y) (y, x) (y, x))dy.
At this point there are no restrictions on the two-point function (x, y) as far
as its symmetry or antisymmetry with respect to its two arguments is concerned.
However, in Sec. 5, we provide reasons for why (x, y) should be antisymmetric so
that we assume this is true from the outset, leading to the denition
(T)(x) =
_
R
n
((x, y) +(y, x)) (x, y)dy, (3.10)
with (x, y) antisymmetric.
We now have that = ( +
) ; note that the integrability of does not

imply that is necessarily integrable. For example, the integral appearing in (3.10)
could be interpreted as the L
2
(R
n
) duality pairing of with that would allow
for to be nonintegrable or to be a distribution.
3.2. Nonlocal point divergence, gradient, and curl operators
Based on the discussion of Sec. 3.1, the nonlocal point divergence, gradient, and
curl operators mapping two-point functions to point functions are dened in terms
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
of their action on two-point functions as follows. These operators along with their
adjoints are the building blocks of our nonlocal calculus.
g
Denition 3.1. (Nonlocal operators) Given the vector two-point function
(x, y) : R
n
R
n
R
k
and the antisymmetric vector two-point function
h
(x, y) : R
n
R
n
R
k
, the action of the nonlocal point divergence operator T
on is dened as
T()(x) :=
_
R
n
( +
) dy for x R
n
, (3.11a)
where T() : R
n
R. Given the scalar two-point function (x, y) : R
n
R
n
R
and the antisymmetric vector two-point function (x, y) : R
n
R
n
R
k
, the action
of the nonlocal point gradient operator ( on is dened as
(()(x) :=
_
R
n
( +
) dy for x R
n
, (3.11b)
where (() : R
n
R
k
. Given the vector two-point function (x, y) : R
n
R
n
R
3
and the antisymmetric vector two-point function (x, y) : R
n
R
n
R
3
, the action
of the nonlocal point curl operator ( on is dened as
(()(x) :=
_
R
n
( +
) dy for x R
n
, (3.11c)
where (() : R
n
R
3
.
The nonlocal point operators T, (, and ( map vectors to scalars, scalars to vec-
tors, and vectors to vectors, respectively, as is the case for the divergence, gradient,
and curl dierential operators. Relationships between the nonlocal point operators
and dierential operators are made in Sec. 5 where we demonstrate circumstances
under which the nonlocal point operators are identied with the corresponding
dierential operators in the sense of distributions and also as weak representations.
Because the integrands in (3.11a)(3.11c) are antisymmetric, (2.4b) immediately
implies that
_
R
n
T()dx = 0,
_
R
n
(()dx = 0, and
_
R
n
(()dx = 0. (3.12)
These relations may be viewed as free-space nonlocal integral theorems.
i
g
Throughout, we assume that all integrals encountered are well dened; this implies assumptions
about the regularity and, for integrals over R
n
, the decay at innity functions appearing in the
integrands.
h
There is no practical reason for having k = n; however, this is a possibility for the nonlocal
divergence and gradient operators, in contrast to the corresponding local operators for which
k = n is not possible.
i
For example,
R
R
n
D()dx = 0 can be viewed as a nonlocal analog of the free-space classical local
Gauss theorem
R
R
n
udx = 0.
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
504 Q. Du et al.
3.3. Nonlocal adjoint operators
The adjoint operators corresponding to nonlocal point operators are two-point oper-
ators that are dened as follows.
Denition 3.2. Given a point operator Q that maps two-point functions F to
point functions dened over R
n
, the adjoint operator Q
is a two-point operator
that maps point functions G to two-point functions dened over R
n
R
n
that
satises
(G, Q(F))
R
n (Q
(G), F)
R
n
R
n = 0, (3.13)
where, for Q = T, (, or (, we have that F and G denote pairs of vectorscalar,
scalarvector, or vectorvector functions, respectively.
Denition 3.2 can be used to determine the nonlocal adjoint two-point operators
corresponding to the nonlocal point operators introduced in Denition 3.1.
Proposition 3.1. (Nonlocal adjoint operators) Given the scalar point function
u(x) : R
n
R, the adjoint of T is the two-point operator whose action on u is
given by
T
(u)(x, y) = (u
u) for x, y R
n
, (3.14a)
where T
(u) : R
n
R
n
R
k
. Given the vector point function v(x) : R
n
R
k
, the
adjoint of ( is the two-point operator whose action on v is given by
(
(v)(x, y) = (v
v) for x, y R
n
, (3.14b)
where (
(v) : R
n
R
n
R. Given the vector point function w(x) : R
n
R
3
, the
adjoint of ( is the two-point operator whose action on w is given by
(
(w)(x, y) = (w
w) for x, y R
n
, (3.14c)
where (
(w) : R
n
R
n
R
3
.
Proof. Let = u. Then,
( +
) = (u + u
) = u( +
) + (u
u)
, (3.15)
so that, from (3.11a) and the rst equation in (3.12), we have
0 =
_
R
n
T()dx =
_
R
n
_
R
n
(u( +
) + (u
u)
)dydx
=
_
R
n
u
_
R
n
( +
) dydx +
_
R
n
_
R
n
(u
u)
dydx
=
_
R
n
uT()dx +
_
R
n
_
R
n
(u
u) dydx, (3.16)
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
where the last equality follows because, due to the antisymmetry of , we have that
(u
u)(
) is an antisymmetric two-point function so that, by (2.4b),

_
R
n
_
R
n
(u
u)(
) dydx = 0. (3.17)
Associating F with , G with u, and Q with T, we see that (3.16) is exactly of the
form (3.13) with Q
= T
, where T
is given by (3.14a).
In a similar manner, (3.14b) can be derived from (3.11b) and the second equation
in (3.12) and (3.14c) can be derived from (3.11c) and the third equation in (3.12).
The operator (
, the negative of the adjoint of the nonlocal gradient oper-

ator (, can be used to dene a second form for a nonlocal divergence operator.
Similarly, the pairs (, T
and (, (
serve to dene two forms for the nonlocal

gradient and curl operators, respectively. It is not surprising that there are two
forms for each operator because, in the nonlocal case, we have two dierent types
of functions, i.e. one-point and two-point functions. Naturally, one needs two sets
of operators, one operating on two-point functions, i.e. T, (, and (, and the other
operating on one-point functions, i.e. T
, (
, and (
.
j
Furthermore, the composi-
tion of operators requires the two sets of operators. For example, if one wants to
compose a divergence and a gradient operator, the composition T(() (where
denotes a two-point function) involves the application of T to the one-point func-
tion (, whereas in general, one expects T to operate on two-point functions. On
the other hand, the composition T(T
u) (where u denotes a one-point function)

involves the application of T to the two-point function T
u.
3.4. Further observations and results about nonlocal operators
In this subsection, we collect several observations and results that can be deduced
from the denitions and results of Secs. 3.2 and 3.3.
3.4.1. A nonlocal divergence operator for tensor functions and a nonlocal
gradient operator for vector functions
A nonlocal divergence operator for tensor functions is dened by applying (3.11a)
to each row of the tensor.
Denition 3.3. Given the tensor two-point function : R
n
R
n
R
mk
and
the antisymmetric vector two-point function : R
n
R
n
R
k
, the action of the
j
Of course, in the local case, one only deals with one-point functions so that there is no need to
dene two dierent sets of operators.
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
506 Q. Du et al.
nonlocal point divergence operator T for tensors on is dened as
T()(x) :=
_
R
n
(+
) dy for x R
n
, (3.18a)
where T()(x) : R
n
R
m
.
Denition 3.2 implies that, given the vector point function v(x) : R
n
R
m
,
the action of the adjoint of T on v is given by
T
(v)(x, y) = (v
v) for x, y R
n
, (3.18b)
where T
(v) : R
n
R
n
R
mk
, i.e. T
maps a vector point function to a second-

order tensor two-point function. One can also extend Denition (3.11b) of the
nonlocal gradient operator of a two-point scalar function to dene the nonlocal gra-
dient operator ( acting on a two-point vector function (x, y) as the point tensor
function
(()(x) :=
_
R
n
(
+) dy.
The corresponding nonlocal adjoint operator (
acting on a point tensor function

U(x) is then given by the two-point vector function
(
(U)(x, y) = (U
U) .
3.4.2. Nonlocal vector identities
Compositions of the point operators dened in Secs. 3.2 and 3.4.1 with the cor-
responding adjoint two-point operators derived in Secs. 3.3 and 3.4.1 lead to the
following nonlocal vector identities. In this proposition, we set = = and, for
the rst, second, and fourth results, i.e. those involving nonlocal curl operators, we
set m = k = 3; we also set m = k for the third result.
k
k
The four identities in (3.19) are analogous to the vector identities associated with the dierential
divergence, gradient, and curl operator:
(u) = 0, (u) = 0, u = tr(u),
(u) +( u) = (u),
respectively. Because ()
= ,
= , and ()
= (), these identities can be written

in the form
(()
u) = 0, (()
u)) = 0,
u = tr(()
u),
(()
u) (()
u) = (()
u)
that more directly correspond to (3.19a)(3.19d), respectively. In particular, the identities (3.19a),
(3.19b) and (3.19d) suggest that D
, G
, and C
can also be viewed as nonlocal analogs of the

dierential gradient, divergence, and curl operators, respectively, that, when operating on point
functions, result in two-point functions.
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
Proposition 3.2. The nonlocal divergence, gradient, and curl operators and the
corresponding adjoint operators satisfy
T((
(u)) = 0 for u: R
n
R
3
, (3.19a)
((T
(u)) = 0 for u : R
n
R, (3.19b)
(
(u) = tr(T
(u)) for u: R
n
R
k
, (3.19c)
T(T
(u)) (((
(u)) = (((
(u)) for u: R
n
R
3
. (3.19d)
Proof. We prove (3.19d); the proofs of (3.19a)(3.19c) are immediate after direct
substitution of the operators involved.
Let u: R
n
R
3
. Then, by (3.11c), (3.14c), (3.18a) and (3.18b) and recalling
that is an antisymmetric function, we have
T(T
(u)) (((
(u)) =
_
R
n
((u
u) + (u u
) dy
+
_
R
n
((u
u) + (u u
)dy
= 2
_
R
n
((u
u)( ) ((u
u) ))dy
= 2
_
R
n
(((u
u) ))dy,
where, for the last equality, we have used the vector identity a(bc) = b(c a)
c(a b). A simple computation shows that the last expression is equal to (((
(u))
so that (3.19d) is proved.
Another set of results for the nonlocal operators that mimic obvious properties of
the corresponding dierential operators are given in the following proposition whose
proof is a straightforward consequence of the results given in Proposition 3.1.
Proposition 3.3. Let b and b denote a constant scalar and vector, respectively.
Then, the adjoints of the nonlocal divergence, gradient, and curl operators satisfy
T
(b) = 0, (
(b) = 0, and (
(b) = 0.
These results do not hold for the point divergence, gradient, and curl operators,
i.e. T(b), ((b), and ((b) do not necessarily vanish for constants b and constant
vectors b.
3.4.3. Nonlocal curl operators in two and higher dimensions
The nonlocal point and two-point curl operators dened in (3.11c) and (3.14c),
respectively, are dened in three dimensions. These denitions can be generalized to
arbitrary dimensions by replacing the vector cross product with the wedge product
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
508 Q. Du et al.
so that, e.g. instead of (3.14c) we would have (
(w) : R
n
R
n
R
r
given by
(
(w)(x, y) = (w
w),
where (x, y) : R
n
R
n
R
r
with r a positive integer. This suggests a possible
exterior calculus-based formalism. However, such a formalism is beyond the scope
of this paper so that only the special cases of r = 3 and r = 2 are considered and
the vector cross product is retained.
We next dene nonlocal point and two-point curl operators in two-space dimen-
sions, i.e. for R
2
; in fact, we have two types of each kind of curl operator.
l
First,
we assume, without loss of generality, that e
3
= 0, w e
3
= 0, and e
3
= 0 in
(3.11c) and (3.14c). Then, for a vector two-point function (x, y) : R
n
R
n
R
2
,
we can view (() as the nonlocal scalar point function dened by
(()(x) =
_
R
n
((
2
+
2
)
1
(
1
+
1
)
2
)dy for x R
n
and, for a vector point function w: R
n
R
2
, we can view (
(w) as the nonlocal

scalar two-point function dened by
(
(w)(x, y) = (w
2
w
2
)
1
(w
1
w
1
)
2
for x, y R
n
.
Next, assume, again without loss of generality, that = e
3
, w = we
3
, and e
3
=
0 in (3.11c) and (3.14c). Then, for a scalar two-point function (x, y) : R
n
R
n
R,
we can view (() as the nonlocal vector point function dened by
(()(x) =
_
R
n
( +
)(
2
e
1

1
e
2
)dy for x R
n
and, for a scalar point function w(x) : R
n
R, we can view (
(w) as the nonlocal

vector two-point function dened by
(
(w)(x, y) = (w
w)(
2
e
1

1
e
2
) for x, y R
n
.
4. A Nonlocal Vector Calculus
We develop a nonlocal vector calculus that mimics the classical vector calculus for
dierential operators. In the classical calculus, interactions are local which is why,
e.g., in the Gauss theorem, the contribution coming from interactions with points
outside a domain appears in the form of a ux through the boundary . If
interactions are nonlocal, then points outside of interact with points in so that
interaction cannot be accounted for merely by an integral along the boundary .
In fact, one must account for interactions with points in the complement domain
R
n
.
l
This is analogous to the two types of dierential curl operators in two dimensions, one operating
on vectors, the other on scalars, respectively, given by
curl u =
u
2
x
1
u
1
x
2
and curl u =
u
x
2
e
1
u
x
1
e
2
.
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
To treat the most general case, we introduce the notion of an interaction domain.
Denition 4.1. Given an open subset R
n
, the interaction domain correspond-
ing to is given by
I
:= y R
n
such that (x, y) ,= 0 for some x , (4.1)
so that
I
consists of those points outside of that interact with points in .
For simplicity, we assume that the corresponding interactions domains for
and are also given by (4.1). No assumption is made about the geometric relation
between and
I
so that, e.g. the four congurations of Fig. 1 are possible. We
have that the situation
I
= R
n
is allowable as is = R
n
in which case, of
course,
I
= .
From (4.1), we see that points in R
n
(
I
) do not interact with points in ,
i.e.
(x, y) = 0 for x and y R
n
(
I
) (4.2)
and, by antisymmetry,
(x, y) = 0 for y and x R
n
(
I
).
We do not assume that points in
I
interact only with points in
I
, i.e. (4.2)
does not hold for x
I
, so that, in general, points in
I
may interact with points
in R
n
(
I
) as well.
4.1. Nonlocal interaction operators
We want to dene operators that can be used to account for the interaction of
points in
I
with points in
I
. To this end, we obtain, successively using
the antisymmetry of and (2.4b), the linearity of the integral operator, (4.2),
Fig. 1. Four of the possible congurations for (the shaded regions) and
I
(the white regions).
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
510 Q. Du et al.
and (3.11a),
0 =
_
I
_
I
( +
) dydx
=
_
I
( +
) dydx +
_
I
_
I
( +
) dydx
=
_
_
R
n
( +
) dydx +
_
I
_
I
( +
) dydx
=
_
T()dx +
_
I
_
I
( +
) dydx. (4.3)
We could not express the last term in terms of T because points in
I
can interact
with points in R
n
(
I
). By (2.5), we identify the last term in (4.3) as the
ux from
I
into
I
and
_
I
( +
) dy as the ux density at x
I
into
I
. Thus, we could use
_
I
( +
) dy to dene the operator we

are seeking. In (4.5a), we instead choose
_
I
( +
) dy for this purpose

because, using the antisymmetry of , (2.1), and (2.4c),
I
_
I
( +
) dydx =
_
I
_
( +
) dydx
=
_
I
( +
) dydx, (4.4)
i.e. it is convenient to have the integral over
I
of the nonlocal operator we seek
to be the ux out of into
I
.
Similar discussions can be made for the nonlocal gradient and curl operators.
Thus, we dene the nonlocal interaction operators as follows.
Denition 4.2. (Nonlocal interaction operators) Given a domain R
n
, let the
interaction domain
I
be dened by (4.1). Then, corresponding to the point diver-
gence operator T() : R
n
R dened in (3.11a), we dene the point interaction
operator ^() :
I
R through its action on by
^()(x) :=
_
I
( +
) dy for x
I
. (4.5a)
Corresponding to the point gradient operator (() : R
n
R
k
dened in (3.11b),
we dene the point interaction operator o() :
I
R
k
through its action on by
o()(x) :=
_
I
( +
)dy for x
I
. (4.5b)
Corresponding to the point curl operator (() : R
n
R
3
dened in (3.11c), we
dene the point interaction operator T () :
I
R
3
through its action on by
T ()(x) :=
_
I
( +
)dy for x
I
. (4.5c)
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
Note that, because of (4.1), or equivalently (4.2), the operator ^ (and likewise
o and T ) can be equivalently dened as ^()(x) :=
_
R
n
( +
) dy for x
I
.
Supercially, this would seem to imply a denition for the interaction operator ^
that is independent of . However, because of (4.2), we cannot dene its domain
I
without rst dening the domain . Thus, we choose to use the equivalent
denition (4.5a) so as to emphasize the role of the interaction domain
I
.
4.2. Nonlocal integral theorems
Denitions 3.1 and 4.2 lead to integral theorems of the nonlocal vector calcu-
lus that mimic the basic integral theorems of the vector calculus for dierential
operators.
Theorem 4.1. (Nonlocal integral theorems) Assuming the notations and deni-
tions found in Denitions 3.1 and 4.2, we have
m
_
T()dx =
_
I
^()dx, (4.6a)
_
(()dx =
_
I
o()dx, (4.6b)
_
(()dx =
_
I
T ()dx. (4.6c)
Proof. The proof is contained in (4.3) because, by (4.5a), the last term is equal to
I
^()dx. In a similar manner, (4.6b) and (4.6c) can be derived. Alternatively,
they can be derived from (4.6a); one simply chooses = b and = b ,
respectively, in that equation, where b is a constant vector; one also has to associate
and with .
From (4.4), we have that
_
I
^()dx is the nonlocal ux from into
I
. Thus,
in words, (4.6a) states that the integral of the nonlocal divergence of over is
equal to the total ux out of into
I
.
n
Similar interpretations of (4.6b) and (4.6c)
hold.
m
The nonlocal integral theorems (4.6) are analogous to the classical dierential integral theorems
given by
Z
vdx =
Z
v ndx,
Z
vdx =
Z
vndx, and
Z
vdx =
Z
n vdx
for functions v and v dened on R
n
, with n = 3 for the third one, for which the integrals are well
dened.
n
This observation is analogous to the observation for the classical Gauss theorem
R
v =
R
v ndA that, by (2.1), the integral of the local divergence of v over is equal to the total
ux out of .
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
512 Q. Du et al.
The nonlocal integral theorems (4.6a)(4.6c) are actionreaction principles. For
example, from (4.6a) and the assumptions made, we have that
0 =
_
T()dx
_
I
^()dx
=
_
_
R
n
( +
) dydx +
_
I
_
I
( +
) dydx
=
_
I
( +
) dydx +
_
I
_
( +
) dydx, (4.7)
which is exactly of the form (2.4c) with
1
= ,
2
=
I
, and = ( +
) ;
(4.7) simply states that the ux out of into
I
is equal and opposite to the ux
from
I
into .
The nonlocal integral theorems (4.6a)(4.6c) have the simple consequences listed
in the following corollary that can be viewed as providing nonlocal integration by
parts formulas.
Corollary 4.1. (Nonlocal integration by parts formulas) Adopt the hypotheses of
Theorem 4.1 and let the nonlocal adjoint operators be given as in Proposition 3.1.
Then, given the point functions u(x) : R
n
R, v(x) : R
n
R
k
, and w(x) : R
n
R
3
, we have
o
_
uT()dx
_
I
_
I
T
(u) dydx =
_
I
u^()dx, (4.8a)
_
v (()dx
_
I
_
I
(
(v)dydx =
_
I
v o()dx, (4.8b)
_
w (()dx
_
I
_
I
(
(w) dydx =
_
I
w T ()dx. (4.8c)
Proof. Let = u; then, from (3.11a) and (3.15), we have
T() =
_
R
n
(u( +
) + (u
u)
)dy
= u
_
R
n
( +
) dy +
_
R
n
(u
u)
dy
o
If and are scalar-valued functions and for free space, a version of the integration by parts
formula (4.8a) appears in Lemma 2.1 of Ref. 13.
The classical analog of (4.8a) is given by, for a scalar function u(x) and vector function v(x),
Z
u vdx +
Z
v udx =
Z
uv ndA
and similarly for (4.8b) and (4.8c).
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
= uT() +
_
R
n
(u
u)
dy
= uT() +
_
I
(u
u)
dy x , (4.9)
where the last equality follows from (4.1). Also, from (3.15) and (4.5a), we have
^() =
_
I
(u( +
) + (u
u)
)dy
= u^()
_
I
(u
u)
dy x
I
. (4.10)
Then,
0 =
_
T()dx
_
I
^()dx
=
_
uT()dx
_
I
u^()dx
+
_
I
(u
u)
dydx +
_
I
_
I
(u
u)
dydx
=
_
uT()dx
_
I
u^()dx +
_
I
_
I
(u
u)
dydx,
where the rst equality follows from (4.6a), the second from (4.9) and (4.10), and
the third from the linearity of the integration operator. Then, (4.8a) follows from the
antisymmetry of , the fact that, similar to (3.17), we have that
_
I
_
I
(u
u)(
) dydx = 0, and (3.14a).

In a similar manner, (4.8b) and (4.8c) can be derived from (4.6a) along with
(3.11b) and (3.11c), respectively. Alternatively, (4.8b) and (4.8c) easily follow by
setting, for an arbitrary constant vector b, = b and = b , respectively, in
(4.8a) and also associating and with .
4.3. Nonlocal Greens identities
Nonlocal (generalized) Greens identities are simple consequences of Corollary 4.1.
For the following two corollaries, we carry over the notations, denitions, and results
obtained above.
Corollary 4.2. (Nonlocal (generalized) Greens rst identities) Given the scalar
point functions u(x) : R
n
R and v(x) : R
n
R and the two-point second-order
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
514 Q. Du et al.
tensor function (x, y) : R
n
R
n
R
kk
, then
p
_
uT( T
(v))dx
_
I
_
I
T
(u) ( T
(v))dydx
=
_
I
u^( T
(v))dx. (4.11a)
Given the vector point functions u(x) : R
n
R
k
and v(x) : R
n
R
k
and the two-
point scalar function (x, y) : R
n
R
n
R, then
_
v (((
(u))dx
_
I
_
I
(
(v)(
(u)dydx
=
_
I
v o((
(u))dx. (4.11b)
Given the vector point functions u(x) : R
n
R
3
and w(x) : R
n
R
3
and the
two-point second-order tensor function (x, y) : R
n
R
n
R
33
, then
_
w (( (
(u))dx
_
I
_
I
(
(w) (( (
(u))dydx
=
_
I
w T ( (
(u))dx. (4.11c)
Proof. The results (4.11a)(4.11c) follow by setting = T
(u) in (4.8a),
= (
(u) in (4.8b), and = (
(u) in (4.8c), respectively.

Corollary 4.3. (Nonlocal (generalized) Greens second identities) We assume the
same notation as in Corollary 4.2 and we also assume that the tensor appearing
in (4.11a) and (4.11c) is symmetric. Then,
_
uT( T
(v))dx
_
vT( T
(u))dx
=
_
I
u^( T
(v))dx
_
I
v^( T
(u))dx (4.12a)
p
We have that (4.11a) and (4.12a) are the nonlocal analogs of the local classical (generalized) rst
Greens identity
Z
u (C v)dx +
Z
u (C v)dx =
Z
un (C v)dA
and of the local classical (generalized) second Greens identity
Z
u (C v)dx
Z
v (C u)dx =
Z
un (C v)dA
Z
vn (C u)dA,
respectively, and similarly for (4.11b), (4.12b), (4.11c), and (4.12c). Here, C(x) denotes a second-
order tensor function with C(x) symmetric for the second identity. These are generalizations of
the classical local Greens identities for which C is the identity tensor.
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
_
v (((
(u))dx
_
u (((
(v))dx
=
_
I
v o((
(u))dx
_
I
u o((
(v))dx (4.12b)
_
w (( (
(u))dx
_
u (( (
(w))dx
=
_
I
w T ( (
(u))dx
_
I
u T ( (
(w))dx. (4.12c)
Proof. The result (4.12a) is obtained by reversing the roles of u and v in (4.11a)
and then subtracting the result from (4.11a). The results (4.12b) and (4.12c) are
obtained from (4.11b) and (4.11c), respectively, in a similar manner.
4.4. Special cases of the vector calculus
We consider some special cases of the vector calculus developed in Secs. 4.14.3.
4.4.1. The free space vector calculus
As was mentioned previously, the vector calculus allows for the choice = R
n
in
which case we also simply set
I
= in all denitions and results. The resulting
nonlocal integral theorems have already been stated; see (3.12). Corresponding to
the nonlocal divergence operator T, we have the nonlocal generalized Greens rst
identity
_
R
n
uT( T
(v))dx
_
R
n
_
R
n
T
(u) T
(v)dydx = 0
and the generalized nonlocal Greens second identity
_
R
n
uT( T
(v))dx
_
R
n
vT( T
(u))dx = 0,
corresponding to (4.11a) and (4.12a), respectively, and similarly for the other
operators.
4.4.2. The vector calculus for interactions of innite extent
In case points in interact with all points in R
n
, we have that
I
= R
n
and
all denitions and results remain unchanged. In this case, ^() = T(), o() =
((), and T () = (() on
I
= R
n
.
4.4.3. The vector calculus for localized kernels
An important special case is that of localized kernels for which we have that
(x, y) = 0 if [y x[ (4.13)
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
516 Q. Du et al.
Fig. 2. For a localized kernel, the domain and the interaction regions
I
and
whose
thicknesses are given by the horizon .
and similarly for and ; here, > 0 denotes a cut-o or interaction or horizon
q
parameter that is not necessarily small and denes the extent of interactions. We
then have that
I
= y R
n
: [y x[ < for x .
We also dene the domain
:= x : [y x[ < for y R
n
= x : [y x[ < for y
I
,
where the last equality follows from (4.13). See Fig. 2 for an illustration.
From (4.1), (4.5a), and (4.13), we have that
_
I
^()dx =
_
I
_
I
( +
) dydx =
_
I
_
( +
) dydx
=
_
I
( +
) dydx =
_
I
( +
) dydx.
The rst term is the nonlocal ux from into
I
; according to the last term, only
points in
contribute to that ux. Similar observations can be made for the

operators o and T .
Of course, all the denitions and results of Secs. 3.24.3 hold for the case of
localized kernels.
5. Relations Between Nonlocal and Dierential Operators
In this section, we choose particular kernel functions , , and that lead to iden-
tications of the nonlocal operators with their dierential counterparts. In Sec. 5.1,
the identication is made in a distributional sense whereas, in Sec. 5.2, we demon-
strate circumstances under which the nonlocal point operators are weak represen-
tations of the divergence, gradient, and curl dierential operators. For the sake
of brevity, we mostly consider the nonlocal point divergence operator T and its
adjoint operator T
. However, analogous results also hold for the nonlocal gradient

and curl operators ( and (, respectively, and their adjoints. We then consider, in
q
The terminology horizon is introduced in the context of cut-o parameters in peridynamic
models for continuum mechanics; see, e.g. Refs. 21 and 24.
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
Sec. 5.3, a connection between the nonlocal and local Gauss theorems made with
the help of two results given in Ref. 17.
Throughout this section, we set k = n in Proposition 3.1, Theorem 3.1, and all
subsequent results.
5.1. Identication of nonlocal operators with dierential
operators in a distributional sense
We begin with the identication of the nonlocal point divergence operator T with
the divergence dierential operator . This identication is subject to the under-
standing that the former operates on two-point functions whereas the latter operates
on point functions. Where there might be ambiguity in applying dierential opera-
tors to two-point function, we identify explicitly the variable of dierentiation, e.g.
y
(x, y) denotes the divergence of (x, y) with respect to y.
We consider special cases of localized kernels (see Sec. 4.4.3) for which 0. As
such, the results and proofs provided in this subsection are purely formal except-
ing Proposition 5.1; a limiting process involving sequences of kernels similar to that
presented in the proof of Proposition 5.1 would make all other proofs rigorous.
We rst relate the nonlocal divergence with the local dierential divergence. Let
(x) be a standard Gaussian with variance .

Proposition 5.1. Let [(
0
(R
n
R
n
)]
n
, i.e. the components of belong
to the space of compactly supported innitely dierentiable functions. Select the
(antisymmetric) distribution
(x, y) =
y
(y x), (5.1)
where
y
denotes the dierential gradient with respect to y. Then,
lim
0
T()(x) = (x, x) x R
n
, (5.2)
where denotes the dierential divergence operator.
Proof. First, by the chain rule, we have
(x, x) = (
x
(x, y))[
y=x
+ (
y
(x, y))[
y=x
= (
y
(y, x))[
y=x
+ (
y
(x, y))[
y=x
= (
y
((y, x) +(x, y)))[
y=x
.
Then, given the convergence of
to the Dirac delta measure in the sense of

distributions, we obtain
(x, x) = (
y
((y, x) +(x, y)))[
y=x
= lim
0
_
R
n
y
((y, x) +(x, y))
(y x)dy
= lim
0
_
R
n
((y, x) +(x, y))
y
(y x)dy
which gives (5.4) by the denition given in (3.11a).
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
518 Q. Du et al.
For any x R
n
, given that in the sense of distributions
y
(y x) converges
to
y
(y x) with the Dirac delta measure, we may let
(x, y) =
y
(y x), (5.3)
and rewrite (5.2) formally as
T()(x) = (x, x) x R
n
. (5.4)
We may further extend (5.4) in the sense of distributions to a more general function
space for , again via a passage to limit or density argument.
Along the same spirit, the following proposition relates T
formally to =
()
.
Proposition 5.2. Let u (
0
(R
n
) and select as in (5.3). Then, for any
[(
(R
n
R
n
)]
n
,
_
R
n
T
(u) dy = (x, x) u(x) x R

n
. (5.5)
Proof. Starting with (3.14a) and (5.3), we have that
_
R
n
T
(u) dy =
_
R
n
(u(y) u(x))(x, y)
y
(y x)dy
=
_
R
n
(y x)
y
((x, y)(u(y) u(x)))dy
=
_
R
n
(y x)(x, y)
y
u(y)dy
= (x, x) u(x) = (x, x) ()
u(x).
Choosing to be the natural Cartesian unit vectors in R
n
, we have, from (5.5),
that
_
R
n
T
(u)dy = u(x) = ()
u(x) x R
n
, (5.6)
i.e. the average of the action of T
on a point function is related, in a distributional

sense, to the action of on that function.
r
We next relate, in a distributional sense, the interaction operator ^ with the
normal component along the boundary.
Proposition 5.3. Assume the hypotheses of Proposition 5.2. Then, for all u(x)
(
0
(R
n
),
_
I
u^()dx =
_
u(x)(x, x) ndA, (5.7)

r
We see that (x, y) as given by (5.3) is antisymmetric. Had we chosen to be a symmetric
distribution or a distribution having a nonvanishing symmetric part, we would not have obtained
the results (5.4) and (5.5), and (5.6). This provides at least partial justication for the assumption,
made at the end of Sec. 3.1, that (x, y) should be antisymmetric.
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
where n denotes the unit outward-pointing normal vector along . Thus, ^() is
a delta measure concentrated on with coecient (weight) (x, x) n.
Proof. With chosen as in (5.3) and using the denition (3.14) for the operator
T
, we have that
I
_
I
T
(u) dydx
=
_
I
_
I
u(y) u(x))(x, y) (x, y)dydx
=
_
I
_
I
(u(y) u(x))(x, y)
y
(y x)dydx
=
_
I
_
y
(u(y) u(x)) (x, y)(y x)dydx
+
_
I
_
I
(u(y) u(x))(y x)
y
(x, y)dydx
I
_
(
I
)
(u(y) u(x))(y x)(x, y) ndydx
=
_
I
_
y
u(y) (x, y)(y x)dydx
=
_
I
u(x) (x, x)dx,
where the fourth equality follows because u(y) u(x) = 0 whenever y = x and
because
y
u(x) = 0. Also, with chosen as in (5.3) and using (5.4), we have that
_
uT()dx =
_
u(x) (x, x)dx

=
_
u(x)(x, x) ndA
_
u(x) (x, x)dx.

Substituting the last two results into (4.8a) results in
_
I
u^()dx =
_
uT()dx
_
I
_
I
T
(u) dydx
=
_
u(x)(x, x) ndA
+
_
I
u(x) (x, x)dx
_
u(x) (x, x)dx. (5.8)

Because is a singular measure, the measure of
I
vanishes, i.e. it becomes a
lower-dimensional domain, in fact, just the boundary of , so that the last two
terms in (5.8) cancel each other out and we obtain (5.7).
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
520 Q. Du et al.
5.2. Relations between weighted nonlocal operators and weak
representations of dierential operators
The classical dierential calculus involves only operators mapping point functions
to point functions; on the other hand, the nonlocal operators dened earlier map
two-point functions to point functions or, in case of the nonlocal adjoint operators,
point functions to two-point functions. To further demonstrate that the nonlocal
operators correspond to nonlocal versions of the classical divergence, gradient, and
curl dierential operators, we use the nonlocal operators T, (, and ( to dene,
in Sec. 5.2.1, corresponding nonlocal weighted operators that map point functions
to point functions. We also show that the adjoint operators corresponding to the
weighted operators are weighted integrals of the nonlocal adjoint operators T
, (
,
and (
. Then, in Sec. 5.2.2, the weighted operators are rigorously shown to be

nonlocal versions of the corresponding dierential operators.
5.2.1. Nonlocal weighted operators
The nonlocal point operators dened in (3.11) induce new operators, referred to as
weighted operators. Throughout this section, we assume that = = . Deni-
tions 5.1 and 5.2 and Proposition 5.4 hold for general k and m although, when we
apply them in Sec. 5.2.2, we set k = m = n.
Denition 5.1. Let (x, y) : R
n
R
n
R denote a non-negative scalar-valued
two-point function. Let the operators T, (, and ( be dened as in (3.11). Then,
given the point function u(x) : R
n
R
k
, the weighted nonlocal divergence operator
T
(u) : R
n
R is dened by its action on u by
T
(u)(x) := T((x, y)u(x)) for x R

n
. (5.9a)
Given the point function v(x) : R
n
R, the weighted nonlocal gradient operator
(
(v) : R
n
R
k
is dened by its action on u by
(
(v)(x) := (((x, y)v(x)) for x R

n
. (5.9b)
Given the point function w(x) : R
n
R
3
, the weighted nonlocal curl operator
(
(u) : R
n
R
3
is dened by its action on w by
(
(w)(x) := (((x, y)w(x)) for x R

n
. (5.9c)
The following result shows that the adjoint operators corresponding to the
weighted nonlocal operators are determined as weighted integrals of the correspond-
ing nonlocal two-point adjoint operators (3.14).
Proposition 5.4. Let (x, y) : R
n
R
n
R denote a non-negative scalar two-
point function and let T
, (
, and (
denote the nonlocal adjoint operators given

M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
in (3.14). The action on u(x) of the adjoint operator T
(u)(x) : R
n
R
k
corre-
sponding to the weighted nonlocal divergence operator T
is given by
T
(u)(x) =
_
R
n
T
(u)(x, y)(x, y)dy for x R

n
(5.10a)
for scalar point functions u(x) : R
n
R. The action on v(x) of the adjoint operator
(
(v)(x) : R
n
R corresponding to the weighted nonlocal gradient operator (
is
given by
(
(v)(x) =
_
R
n
(
(v)(x, y)(x, y)dy for x R

n
(5.10b)
for vector point functions v(x) : R
n
R
k
. The action on w(x) of the adjoint oper-
ator (
(w)(x) : R
n
R
3
corresponding to the weighted nonlocal curl operator (
is given by
(
(w)(x) =
_
R
n
(
(w)(x, y)(x, y)dy for x R

n
(5.10c)
for vector point functions w(x) : R
n
R
3
.
Proof. We have that
(T
(u), u)
R
n =
_
R
n
u(x)T((x, y)u(x))dx
=
_
R
n
_
R
n
T
(u)(x, y) ((x, y)u(x))dydx

=
_
R
n
__
R
n
(x, y)T
(u)(x, y)dy
_
u(x)dx,
where (5.9a) is used for the rst equality and (3.13) and (3.14a) for the second.
But, by denition, the adjoint operator T
() corresponding to the operator T
()
satises
(u, T
(u))
R
n = (T
(u), u)
R
n.
Comparing the last two results yields (5.9a). The conclusions (5.9b) and (5.9c) are
derived in a similar fashion.
The denition given in (5.9a) and the result (5.10a) can be extended to tensors
and vectors, respectively.
Denition 5.2. Let (x, y) : R
n
R
n
R denote a non-negative scalar two-
point function and let T be given as in (3.18b). Given the tensor point function
U(x) : R
n
R
k
, the weighted nonlocal divergence operator T
(U) : R
n
R
for
tensors is dened by
T
(U)(x) := T((x, y)U(x))(x) for x R

n
. (5.11)
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
522 Q. Du et al.
Corollary 5.1. The adjoint operator T
(u)(x) : R
n
R
k
corresponding to the
operator T
is given by
T
(u)(x) =
_
R
n
T
(u)(x, y)(x, y)dy for x R

n
(5.12)
for vector point functions u: R
n
R
.
Similar results hold for the operator (
v
.
5.2.2. Relationships between weighted operators and dierential operators
We have seen that the nonlocal operators satisfy many properties that mimic their
dierential counterparts. In particular, we established, in Sec. 5.1, that the nonlocal
point operators can be identied with the corresponding dierential operators in a
distributional sense. Here, we establish rigorous connections between the weighted
operators and their dierential counterparts for the case of special localized kernels
(see Sec. 4.4.2), i.e. we introduce the horizon parameter > 0 and analyze what
occurs as 0, that is, in the local limit. To this end, we dene
B
(x) := y R
n
: [y x[ <
and let denote a positive radial function satisfying the normalization condition
_
B
(x)
[y x[
2
([y x[)dy = n, (5.13)
where n denotes the space dimension. We then choose
s
_
_
(x, y) =
y x
[y x[
for x ,= y,
(x, y) =
_
[y x[([y x[) y B
(x),
0 otherwise.
(5.14)
We have that is an antisymmetric function
t
whereas is a symmetric function.
We then have, for a scalar function u, that the components of (
(u) and T
(u)
s
Analogous to the results of Sec. 5.1, if we instead choose the singular distribution
(|z|)
j
(z) = z
j
(|z|) =
j
(z) for j = 1, 2, 3,
where
j
u denotes the weak derivative of u with respect to x
j
, we obtain
d
j
=
j
and d
j
=
j
.
t
Again, the antisymmetry of (x, y) is necessary for the results of this section, giving further
justication for the assumption, made at the end of Sec. 3.1, that (x, y) should be antisymmetric.
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
given by (5.9b) and (5.10a), respectively, are given by, for j = 1, . . . , n,
d
j
u(x) :=
_
B
(0)
(u(x +z) + u(x))z
j
([z[)dz, (5.15a)
d
j
u(x) :=
_
B
(0)
(u(x +z) u(x))z
j
([z[)dz, (5.15b)
where z
j
denotes the jth component of z.
The following result ensues.
Lemma 5.1. Assume that and are dened as in (5.14). Let u : R
n
R and
let d
j
u(x) and d
j
u(x) denote the jth components of (
(u) and T
(u), respectively.
Then,
d
j
u = d
j
u. (5.16)
Proof. From (5.15) we have
d
j
u(x) + d
j
u(x) = 2u(x)
_
B
(0)
z
j
([z[)dz = 0
from which the result directly follows.
Based on this lemma and the denition of the weighted operators, we have the
following results.
Corollary 5.2. Under the same conditions as in Lemma 5.1, we have
T
= (
, (
= T
, and (
= (
. (5.17)
Thus, under the hypotheses of this corollary, the identities in (5.17) mimic their
counterparts in the vector calculus for dierential operators. Identities such as those
in Proposition 3.2 do not generally hold for the weighted operators due to their
nonlocal properties.
u
The following proposition demonstrates that the components of the weighted
gradient operator (
or, by (5.17), T
, converge, as 0, to the corresponding

spatial derivatives.
Proposition 5.5. Let be dened as in (5.14) with satisfying (5.13). Then, for
j = 1, . . . , n, the weighted operators d
j
and d
j
dened by (5.15) are bounded linear
operators from H
1
(R
n
) to L
2
(R
n
). Moreover, if u H
1
(R
n
), then as 0,
|d
j
u
j
u|
L
2
(R
n
)
0, (5.18a)
|d
j
u +
j
u|
L
2
(R
n
)
0, (5.18b)
u
The identities in Lemma 5.1 and Corollary 5.2 also hold for more general choices of and ,
namely, those such that is antisymmetric and is a radial function with support in B
(x).
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
524 Q. Du et al.
where
j
u denotes the weak partial derivative of u with respect to x
j
. Moreover, if
_
B
(0)
[z[
1+s
([z[)dz < for some 0 s 1, (5.18c)
then, for j = 1, . . . , n, the weighted operators d
j
and d
j
are bounded linear operators
from H
t
(R
n
) to H
ts
(R
n
) for any t 0.
Proof. The conclusion (5.18a) is equivalent to |
d
j
u
j
u|
L
2
(R
n
)
0, where

d
j
u
and

j
u denote the Fourier transforms of d
j
u and
j
u, respectively.
Under the condition (5.18c) for s [0, 1], if u H
t
(R
n
) for t 0, then
d
j
u =
_
B
(0)
(e
iy
+ 1) u()y
j
([y[)dy
= i
_
B
(0)
sin(y ) u()y
j
([y[)dy
= i
_
B
(0)
_
_
sin(y
j
j
) cos
_
_
k=j
y
k
k
_
_
+ cos(y
j
j
) sin
_
_
k=j
y
k
k
_
_
_
_
u()y
j
([y[)dy
= i u()
_
B
(0)
sin(y
j
j
) cos
_
_
k=j
y
i
k
_
_
y
j
([y[)dy,
where the second and fourth equalities hold because of the symmetry of the domain
of integration and the antisymmetry of the integrand with respect to the integration
variable, respectively. Because for any 0 s 1, [ sin(y
j
j
) cos(
k=j
y
i
k
)[
[y
j
j
[
s
, we have
[
d
j
u[ [
s
u()[
_
B
(0)
[y
j
[
1+s
([y[)dy,
so that

d
j
u H
ts
(R
n
). In particular, for t = s = 1, we obtain that, under the
condition (5.13), the weighted operators d
j
, j = 1, . . . , n, dened in (5.15a) are
bounded linear operators from H
1
(R
n
) to L
2
(R
n
).
If u H
1
(R
n
), we have that
[
d
j
u
j
u[ [
d
j
u[ +[
j
u[
_
B
(0)
sin(y
j
j
) cos
_
_
k=j
y
k
k
_
_
u()y
j
([y[)
dy +[
j
u[
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
[
j
u[
_
B
(0)
y
2
j
([y[)dy +[
j
u[
2[
j
u[ L
2
(R
n
),
where the third inequality holds because [sin(x)[ [x[ and [cos(x)[ 1. By Taylors
theorem, we have
d
j
u = i
_
B
(0)
(y
j
j
+ y
3
j
3
j
cos(
1
)/6)
_
_
_
1 +
_
_
k=j
y
k
k
_
_
2
cos(
2
)/2
_
_
_
uy
j
([y[)dy
= i
j
u
_
B
(0)
y
2
j
([y[)dy + I
() = i
j
u + I
(),
where, for some
1
and
2
,
I
() :=
i
6
_
B
(0)
y
4
j
cos(
1
)([y[)dy
3
j
u
+
i
2
_
B
(0)
y
2
j
_
_
k=j
y
k
k
_
_
2
cos(
2
)([y[)dy
j
u
+
i
12
_
B
(0)
y
4
j
_
_
k=j
y
k
k
_
_
2
cos(
1
) cos(
2
)([y[)dy
3
j
u.
Because u is bounded a.e., we see that, for any ,
[I
()[
2
[
3
j
u[ +
2
_
_
k=j
k
_
_
2
j
u
+
4
_
_
k=j
k
_
_
2
3
j
u
0 as 0.
Hence, we have
d
j
u
j
u a.e. for R
n
as 0.
Then, by the dominated convergence theorem, we obtain
|d
j
u
j
u|
L
2
(R
n
)
0 as 0.
Lemma 5.1 implies that the same results hold for the operator d
j
.
The above lemma implies that if satises (5.18c) for s = 0, then the weighted
operators are bounded operators from L
2
(R
n
) to L
2
(R
n
). More generally, for
satisfying (5.18c) with positive s, the operators d
j
and d
j
actually map a subspace
of L
2
(R
n
), for instance the fractional Sobolev space H
s
(R
n
) to L
2
(R
n
), or even
map L
2
(R
n
) to H
s
(R
n
). We refer to Ref. 7 for related work.
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
526 Q. Du et al.
A direct consequence of Proposition 5.5 is the following result.
Corollary 5.3. Under the condition of Proposition 5.5, the weighted operators T
,
(
, and (
and their adjoint operators T
, (
, and (
are bounded linear operators

from H
t
(R
n
) to H
ts
(R
n
) for 0 s 1, where n = 3 for the weighted curl
operators. Moreover, if u H
1
(R
n
) and u [H
1
(R
n
)]
n
, then
T
(u) u, T
(u) u,
(
(u) u, (
(u) u,
(
(u) u, (
(u) u,
(5.19)
where the convergence as 0 is with respect to L
2
(R
n
).
Proof. We prove the convergence of the operator T
(u) to the divergence dier-

ential operators. First, note that
T
(u) =
n
i=1
d
i
u
i
and u =
n
i=1
i
u
i
,
so that
|T
(u) u|
L
2
(R
n
)

n
i=1
|d
i
u
i

i
u
i
|
L
2
(R
n
)
0 as 0.
The remaining results can be proved in a similar fashion.
For the purpose of discussing nonlocal equations, we also need to consider com-
binations of the weighted operators such as T
(C
1
T
(u)), where C
1
(x) is a
constitutive tensor point function that, for example, describes a point property
of a material. Note that the two-point property is involved in the denition of the
weighted operators. In the next corollary, we illustrate that whenever the horizon
goes to zero, the combinations of the weighted operators converge to their local
counterparts.
Corollary 5.4. Let u H
1
(R
n
), u [H
1
(R
n
)]
n
, C
1
: R
n
R
n
R
n
in L
(R
n
R
n
), and c
2
: R
n
R in L
(R
n
). Then,
T
(C
1
T
(u)) (C
1
u),
(
(c
2
(
(u)) (c
2
u),
(
(C
1
(
(u)) (C
1
(u)),
(5.20)
where the convergence as 0 is with respect to H
1
(R
n
).
Proof. Using Proposition 5.5 and a similar method of proof as for Lemma 5.3, the
results are obtained.
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
We also have the following result.
Corollary 5.5. Let L denote a linear operator that commutes with the dieren-
tial and nonlocal operators. Then, if Lu [H
1
(R
n
)]
n
, Lu H
1
(R
n
), and Lu
[H
1
(R
3
)]
3
we have
T
(Lu) Lu, T
(Lu) Lu,
(
(Lu) Lu, (
(Lu) Lu,
(
(Lu) Lu, (
(Lu) Lu,
(5.21)
where the convergence as 0 is with respect to L
2
(R
n
).
If L is selected as a dierential operator with constant coecients, or its formal
inverse, we can observe convergence of the weighted operators in either stronger or
weaker norms.
5.3. A connection between the nonlocal and local Gauss theorems
Two interesting lemmas found in Ref. 17
v
allow us to make another connection
between nonlocal and local operators or, more precisely, between the nonlocal Gauss
theorem (4.6a) and the classical Gauss theorem.
Given a two-point function : R
n
R
n
R
k
, we dene the vector-valued point
function q(x) : R
n
R
k
by
q(x) :=
_
R
n
(y x)(x, y x)dy, (5.22)
where, with p(x, y) = ( +
) and z = y x, the function : R

n
R
n
R is
given by
(x, z) =
_
1
0
p(x + z, x (1 )z)d.
Then, Lemmas 1 and 2 in Ref. 17 state that
q(x) =
_
R
n
( +
) dy x R
n
(5.23)
and
_
q(x) ndA =
_
_
R
n
\
( +
) dydx, (5.24)
respectively, where n denotes the outward pointing unit normal vector along .
From (3.11a) and (5.23) we then have that
q = T() (5.25)
v
See Ref. 15 for an English translation of Ref. 17.
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
528 Q. Du et al.
and from (4.5a) and (5.24), using the by now familiar sequence of steps, we have
that
_
q ndA =
_
_
R
n
\
( +
) dydx =
_
R
n
\
_
( +
) dydx
=
_
I
_
( +
) dydx =
_
I
_
I
( +
) dydx
=
_
I
^()dx. (5.26)
Then, the nonlocal Gauss theorem (4.6a) for , (5.25), and (5.26) imply that
0 =
_
T()dx
_
I
^()dx =
_
qdx
_
q ndA,
i.e. the classical Gausss theorem for the vector-valued function q. Thus, we have
shown that the nonlocal Gausss theorem (4.6a) for the nonlocal vector two-point
function (x, y) formally implies the classical Gauss theorem for the local vector
point function q(x) derived from through (5.22). Evidently, the Gauss theorem
can be given by a meaning without the notions of the divergence operator, unit
normal vector, or surface.
6. Examples of Nonlocal Volume-Constrained Problems
The nonlocal point operators, the corresponding nonlocal adjoint operators, and
the corresponding nonlocal interaction operators given in (3.11), (3.14), and (4.5),
respectively, can be used to dene nonlocal problems that are analogous to classical
boundary-value problems for partial dierential equations. Here, we merely state
problems involving scalar and vector second-order operators so that we are in
the setting of elliptic problems. Specically, we dene nonlocal problems that are
analogous to the second-order dierential boundary-value problems
_
_
_
(C
2
u) = b in ,
u = g
d
on
d
,
(C
2
u) n = g
n
on
n
,
(6.1a)
_
_
_
(C
4
: u) = b in ,
u = g
d
on
d
,
(C
4
: u) n = g
n
on
n
,
(6.1b)
_
_
_
(C
2
u) (c
0
u) = b in ,
u = g
d
on
d
,
n (C
2
u) n + c
0
( u)n = g
n
on
n
,
(6.1c)
respectively, where =
d

n
denotes the boundary of with
d

n
= ;
d
and
n
are the parts of the boundary on which Dirichlet and
Neumann boundary conditions are applied, respectively. In (6.1a)(6.1c), C
4
, C
2
,
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
and c
0
denote fourth-order tensor, second-order tensor, and scalar point functions,
respectively.
Each of the problems (6.1a)(6.1c) for an unknown function, e.g. u in (6.1a),
consists of a partial dierential equation and boundary conditions, where the latter
may be viewed as constraints placed on possible solutions of the partial dierential
equation. Thus, in the case of local operators, constraints are applied along the
boundary of the domain on which the partial dierential equation is applied.
A consequence of the nonlocality of operators is that constraints analogous to the
boundary conditions in (6.1a)(6.1c) are applied instead over sets with positive
measure in R
n
. Thus, in the case of nonlocal operators, constraints are applied on the
interaction domain
I
that has positive volume and that corresponds to the domain
on which the nonlocal operator equation is applied. As a result, we refer to nonlocal
problems with constraints imposed on volumes as volume-constrained problems in
contrast to local problems with constraints imposed on boundary surfaces that are
universally referred to as boundary-value problems.
We now describe the nonlocal problems analogous to (6.1a)(6.1c). In the dif-
ferential equation setting we had to divide the boundary in the two parts
d
and
n
over which we applied Dirichlet and Neumann conditions, respec-
tively. Similarly, in the nonlocal case, we divide the interaction domain
I
into
two subdomains that we use to apply Dirichlet-like and Neumann-like volume con-
straints. Let
I
=
Id

In
, where
Id

In
= although either
Id
or
In
may be empty. Over
Id
we specify function values, e.g. for a given function g
d
,
we set
u(x) = g
d
(x) for x
Id
. (6.2)
This is, of course, a straightforward generalization of the Dirichlet boundary con-
dition u(x) = g
d
(x) for x
d
for the partial dierential equation case; see, e.g.
(6.1a).
The Neumann boundary condition in (6.1a) implies that q n = (C
2
u) n =
g
n
so that it species the negative of the ux density at a point x ; see
Sec. 2.1. To follow the local case, in the nonlocal case, we similarly specify the
negative of the nonlocal ux density at a point x . Thus, from the discussion
in Sec. 4.1 about the operator ^, we specify, for a given function g
n
(x) and tensor
function
2
(x, y), we set ^
In
()(x) = g
n
(x) for x
In
and =
2
T
(u)
so that
^
In
(
2
T
(u)) = g
n
for x
In
. (6.3)
Thus, to dene volume-constrained problems analogous to the boundary-value
problems (6.1a)(6.1c), we let
4
,
2
, and
0
denote fourth-order tensor, second-
order tensor, and scalar two-point functions, respectively, where the tensors are
symmetric in the function and matrix senses and are positive denite. Using
(6.2) and (6.3), the nonlocal volume-constrained problems corresponding to the
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
530 Q. Du et al.
boundary-value problems (6.1a)(6.1c) are then given by
_
_
_
T(
2
T
(u)) = b in ,
u = g
d
in
Id
,
^
In
(
2
T
(u)) = g
n
in
In
,
(6.4a)
_
_
_
T(
4
: T
(u)) = b in ,
u = g
d
in
Id
,
^
In
(
4
: T
(u)) = g
n
in
In
,
(6.4b)
_
_
_
((
2
(
(u)) +((
0
(
(u)) = b in ,
u = g
d
in
Id
,
T
In
(
2
(
(u)) +o
In
(
0
(
(u)) = g
n
in
In
,
(6.4c)
respectively.
The nonlocal Greens rst identities given in Corollary 4.2 are useful for dening
variational formulations of the nonlocal volume-constrained problems. For example,
from (4.11a) and (6.4a), we have that u(x) satises u = g
d
on
Id
, and for suitable
test functions v(x) that vanish on
Id
,
_
I
_
I
T
(u) (
2
T
(v))dydx =
_
vbdx +
_
In
vg
n
dx. (6.5)
The kernel function dictates the choice of function spaces for the trial and test
functions u and v, respectively, so that the problem (6.5) is well-posed.
A special form of the nonlocal volume-constrained problem (6.4a) correspond-
ing to a scalar-valued solution is studied in Ref. 12 by appealing to a variational
formulation. Well-posedness results are provided in Ref. 12 for the case in which
the natural energy space (used to dene the variational problem) is equivalent to
L
2
(), the space of square integrable functions. Although the notion of a nonlocal
vector calculus is not introduced, the recent book Ref. 2 describes, using semi-
group analysis, substantial recent work on nonlocal diusion and its relationship
to (6.1a). The report Ref. 5 extends these previous results to the case when the
natural energy space is equivalent to a fractional Sobolev space, a proper subspace
of L
2
(). In this case, the variational problem possesses smoothing properties akin
to those for elliptic partial dierential equations but with possibly reduced order of
smoothing.
7. Local and Nonlocal Balance Laws with Applications
A balance law postulates that the rate of change of the amount of a quantity in any
domain is given by the rate at which that quantity is produced within the domain
minus the rate at which the quantity exits the domain; the latter is referred to as
the ux out of the domain. In the classical dierential equation setting, the ux
is local, i.e. the quantity exits or enters the subdomain through its boundary; see
Sec. 2.1. Diculties arise, however, in classical settings due to, e.g. discontinuities
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
and corner singularities, all of which are troublesome when dening an appropriate
notion for the ux through the boundary of a subdomain.
The nonlocal vector calculus we developed has an important application to non-
local balance laws for which subdomains not in direct contact may have a nonzero
interaction, i.e. there is a nonzero ux between the subdomains. This is accom-
plished by dening the ux in terms of interactions between open regions of posi-
tive measure that are possibly a nite distance apart; recall Sec. 2.2. An important
feature of nonlocal balance laws is that the signicant technical details associated
with determining normal and tangential traces along boundaries of suitable regions
is obviated when uxes are induced through interactions between volumes. Our
nonlocal calculus, then, provides an alternative to standard approaches for circum-
venting the technicalities associated with lack of sucient regularity in local balance
laws such as measure-theoretic generalizations of the GaussGreen theorem (see,
e.g. Refs. 3 and 19) or the use of the fractional calculus (see, e.g. Refs. 1 and 25).
7.1. Abstract ux operators
In this section, we dene an abstract ux operator that serves as a proxy for inter-
actions between two domains and then derive several useful properties that such
operators satisfy.
Denition 7.1. For any two not necessarily disjoint open domains
1
R
n
and
2
R
n
, a ux operator is any operator T(
1
,
2
) satisfying
T(
, ) = T(,
) = 0
R
n
, (7.1a)
T(
1

2
,
3
)
= T(
1
,
3
) +T(
2
,
3
) T(
1

2
,
3
)
T(
3
,
1

2
)
= T(
3
,
1
) +T(
3
,
2
) T(
3
,
1

2
)
_
_

1
,
2
,
3
R
n
, (7.1b)
and the alternating or no self-interaction property
T(
) = 0
R
n
. (7.1c)
The following result shows that the dening relation (7.1c) can be replaced with
an antisymmetry property.
Proposition 7.1. Assume that the ux operator T(, ) satises (7.1a) and (7.1b).
Then, the antisymmetry property or actionreaction principle
T(
1
,
2
) +T(
2
,
1
) = 0
1
,
2
R
n
(7.2)
is equivalent to the alternating property (7.1c).
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
532 Q. Du et al.
Proof. The fact that (7.2) implies (7.1c) is obvious; simply set
1
=
2
=

in
the former.
On the other hand, through repeated applications of (7.1a)(7.1c), we obtain
0 = T(
1

2
,
1

2
)
= T(
1
,
1

2
) +T(
2
,
1

2
) T(
1

2
,
1

2
)
= T(
1
,
1
) + T(
1
,
2
) T(
1
,
1

2
)
+T(
2
,
1
) +T(
2
,
2
) T(
2
,
1

2
)
T(
1

2
,
1
) T(
1

2
,
2
) +T(
1

2
,
1

2
)
= T(
1
,
2
) + T(
2
,
1
) T(
1
,
1

2
) T(
1

2
,
1
)
T(
2
,
1

2
) T(
1

2
,
2
)
= T(
1
,
2
) + T(
2
,
1
)
T(
1
(
1

2
),
1

2
) T(
1

2
,
1

2
)
T(
1

2
,
1
(
1

2
)) T(
1

2
,
1

2
)
T(
2
(
1

2
),
1

2
) T(
1

2
,
1

2
)
T(
1

2
,
2
(
1

2
)) T(
1

2
,
1

2
)
= T(
1
,
2
) + T(
2
,
1
)
T(
1
(
1

2
),
1

2
) T(
1

2
,
1
(
1

2
))
T(
2
(
1

2
),
1

2
) T(
1

2
,
2
(
1

2
)). (7.3)
Now, assume that the regions
1
and
2
are disjoint, i.e.
1
2
= . Then, (7.2)
immediately follows from (7.3). Thus, we have proven (7.2) for disjoint regions.
If
1

2
,= , we have that
1
(
1

2
) and
1

2
are disjoint regions as
are
2
(
1
2
) and
1

2
. Then, from (7.2) for disjoint regions, which we just
proved, and (7.3), we obtain (7.2) for the case
1

2
,= .
The following results hold only for disjoint regions.
Then, T(, ) is bilinear for disjoint regions, i.e.
T(
1

2
,
3
) = T(
1
,
3
) +T(
2
,
3
)
T(
3
,
1

2
) = T(
3
,
1
) +T(
3
,
2
)
_

1
,
2
,
3
R
n
such
that
1

2
= .
(7.4)
Proof. The results immediately follow from (7.1a) and (7.1b).
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
Then, T(, ) is additive for disjoint regions, i.e.
T(
1

2
, R
n
(
1

2
)) = T(
1
, R
n
1
) +T(
2
, R
n
2
)

1
,
2
R
n
such that
1

2
= , (7.5)
if and only if T(, ) is symmetric, i.e. if and only if T(, ) satises (7.2).
Proof. We have that if T(, ) satises (7.2), then
T(
1

2
, R
n
(
1

2
))
= T(
1
, R
n
(
1

2
)) +T(
2
, R
n
(
1

2
))
= T(
1
, R
n
(
1

2
)) +T(
2
, R
n
(
1

2
)) +T(
1
,
2
) +T(
2
,
1
)
= T(
1
, (R
n
(
1

2
))
2
) +T(
2
, (R
n
(
1

2
))
1
)
= T(
1
, R
n
1
) +T(
2
, R
n
2
),
i.e. T(, ) satises (7.5). Here, we have used (7.4) for the rst and third equalities,
(7.2) for the second, and simple set theory for the fourth.
If, on the other hand, if T(, ) satises (7.5), then reversing the above steps
0 = T(
1
, R
n
1
) +T(
2
, R
n
2
) T(
1

2
, R
n
(
1

2
))
= T(
1
, R
n
(
1

2
)) +T(
2
, R
n
(
1

2
))
T(
1

2
, R
n
(
1

2
)) +T(
1
,
2
) +T(
2
,
1
)
= T(
1
,
2
) +T(
2
,
1
),
where the rst equality follows from (7.5), the second from simple set theory, and
the third from (7.4).
7.2. Local and nonlocal balance laws
In this section, we pose an abstract balance law, discuss local and nonlocal balance
laws, and show how the classical vector calculus and the nonlocal vector calculus
are used to derive local and nonlocal eld equations, respectively. The process is
illustrated in the context of heat conduction; another illustration is given in Sec. 7.3
where the peridynamic model of continuum mechanics is considered.
7.2.1. Abstract balance laws
We start with an abstract balance law for an open subset R
n
given by
/(
; q) = T(
) T(
, R
n
; q)
, (7.6)
which postulates that /(
; q) (the rate of change of the amount of the intensive

quantity q in any open subdomain

) is equal to T(
) (the rate at which the

M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
534 Q. Du et al.
intensive quantity is produced in the subdomain by sources) minus T(
, R
n
; q)
(the rate at which the intensive quantity exits the subdomain). In (7.6), q(x, t) may
be a scalar or vector time-dependent point function and we assume that T(
), the
rate of production of q in any subdomain

, is given, e.g. through the specication
of an external source function. Here, we take the usual choices
/(
; q) =
d
dt
_
e
q(x, t)dx and T(
) =
_
e
b(x, t)dx (7.7)

for the rate of change and production terms, respectively, where b(x, t) denotes a
given source density function. If we assume the general case for which points in
only interact with points in
I
R
n
, then (7.6) reduces to
/(
; q) = T(
) T(
,
I
; q)
. (7.8)
The assumption (7.1c) postulates that there are no self-interactions. The result
(7.2) is an abstract actionreaction principle, i.e. the ux from
1
into
2
is equal
and opposite to the ux from
2
into
1
. The result (7.4) states that the total
interaction or ux from two disjoint regions
1
and
2
into a region
3
is simply
the sum of the individual uxes from
1
into
3
and from
2
into
3
; (7.4) also
states that the total ux from a region
3
into two disjoint regions
1
and
2
is the
sum of the individual uxes from
3
into the two subdomains. Finally, (7.5) implies
the following proposition that shows that the abstract balance law is additive for
disjoint regions.
Proposition 7.4. Assume the ux operator T(, ) satises (7.1c)(7.1b) and that
1
and
2
are disjoint. Then, the balance law (7.6) and (7.7) is additive in the
following sense:
/(
1

2
; q) T(
1

2
) +T(
1

2
, R
n
(
1

2
); q)
= /(
1
; q) T(
1
) +T(
1
, R
n
1
; q)
+/(
2
; q) T(
2
) +T(
2
, R
n
2
; q). (7.9)
Proof. The additivity of /(; q) and T() given by (7.7) is clear by the linearity of
the integral. Combining with (7.5) then yields (7.9).
7.2.2. Local balance laws
For disjoint open regions
1
,
2
R
n
, we assume that the ux operator T(
1
,
2
; q) is given by (2.1) so that
T
loc
(
1
,
2
; q) =
_
12
q ndA for
1

2
= with
12
=
1

2
(7.10)
for some vector q, where
12
denotes the common boundary of
1
and
2
and n
the unit normal along
12
pointing out of
1
; q is related to the intensive variable
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
q through a constitutive equation; see below for an example. We have that
if
12
= , then T
loc
(
1
,
2
; q) = 0.
The ux operator given in (7.10) satises (7.4) and, by (2.2), is antisymmetric.
From (7.6), (7.7), (7.10), and the Gauss theorem, we have that
_
e
_
q
t
+ q b
_
dx = 0

from which, because

is arbitrary in , we obtain the local eld equation
q
t
+ q = b x , (7.11)
corresponding to the balance law (7.6) and the ux operator (7.10).
To obtain a eld equation in terms of the intensive variable q, a constitutive
equation must be postulated relating the ux vector q to q. For example, consider
the case of heat conduction for which the variable q denotes the temperature. Then,
a constitutive equation relating the heat ux vector q to the temperature q is given
by the Fourier heat law q = q, where denotes the thermal diusivity, assumed
here to be a constant. Then, from (7.11), we have the eld equation
q
t
= q + b (7.12)
for the temperature q, i.e. the heat equation.
7.2.3. Nonlocal balance laws
From (2.5), in the nonlocal case, we dene the ux operator
w
T
nonloc
(
1
,
2
; q) :=
_
1
_
2
( +
) dydx
1
,
2
R
n
(7.13)
that gives the nonlocal ux from
1
into
2
; here, the vector (x, y) has to be
related to q through a constitutive relation. The ux operator given in (7.13) satis-
es (7.4), i.e. is bilinear for disjoint regions, and, because (x, y) is antisymmetric,
is alternating and therefore satises the actionreaction principle.
In stark contrast to the ux operator T
loc
(, ; q) given in (7.10), the ux induced
by T
nonloc
(, ; q) dispenses with the need for determining a unit normal to an ori-
entable surface and instead considers volume interactions among regions. Further-
more, the local ux operator T
loc
(, ; q) vanishes whenever
1

2
= whereas,
in general, the nonlocal ux operator T
nonloc
(, ; q) does not.
w
Comparing F
loc
(
1
,
2
; q) and F
nonloc
(
1
,
2
; q), we see that the role of the ux density q n
for the local balance is now assumed by the nonlocal ux density N().
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
536 Q. Du et al.
We derive a nonlocal eld equation analogous to the classical heat equation.
Due to (7.1c) with
1
=

and
2
=

I
, we have from (7.13) that
T
nonloc
(
I
; q) =
_
e
_
e
I
( +
) dydx =
_
e
I
_
e
( +
) dydx
=
_
e
I
_
e
I
( +
) dydx =
_
e
I
^()dx
,
(7.14)
where

I

I
denotes the interaction domain corresponding to

.
x
Substituting
(7.14) along with (7.7) into (7.6), we obtain
_
e
q
t
dx +
_
e
I
^() =
_
e
bdx
.
Then, using the nonlocal Gauss theorem (4.6a), we obtain
_
e
_
q
t
+T() b
_
dx = 0

from which it follows that, because

is arbitrary in ,
q
t
+T() = b x .
The interaction vector is related to the intensive quantity q through a constitutive
relation such as = T
(q), where the adjoint operator T
is dened in (3.14a).
We then obtain the eld equation for q given by
q
t
+ TT
(q) = b x ,
which may be viewed as a nonlocal heat equation analogous to the classical local
heat equation (7.12). Note that that the operator TT
() is exactly that which

appears in the steady-state volume constrained problem (6.4a) with
2
= I, where
is constant and I denotes the identity tensor.
7.3. The peridynamics nonlocal theory of continuum mechanics
We now demonstrate that the nonlocal calculus and the related nonlocal balance
laws developed in the earlier sections can be used to formulate the equations of the
peridynamic nonlocal theory of continuum mechanics.
The abstract balance law (7.6) and the discussion that ensues generalize, in
straightforward manner, to vector-valued intensive quantities. In particular, we
obtain the eld equation
m
t
+T() = b x , (7.15)
x
See also p. 85 in Ref. 24 and p. 29 in Ref. 18.
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
where the operator T is dened in (3.18a). In particular, consider the case of
m denoting the momentum density for a continuum material so that m =
(x)

t
u(x, t), where () denotes the material density and u(x, t) the displacement.
A constitutive equation relating (x, y) to u(x) is given by (x, y) =
1
2
(T
(u))
T
,
where the adjoint operator T
() is dened in (3.18b). We then have, from (7.15),

that
2
u
t
2
+
1
2
T((T
u)
T
) = b x . (7.16)
Substituting the denitions (3.18a) and (3.18b) for T and T
, respectively, we
obtain, invoking (4.1),
T((T
u)
T
) = 2
_
I
() (u
u)dx, (7.17)
so that, from (7.16), we have that
2
u
t
2
=
_
I
() (u
u)dx +b x . (7.18)
A mechanical perspective indicates that (T
u)
T
describes the deformation of u and
that the constitutive relation maps the deformation to the force density given by
the integral operator. Because the integrand of (7.17) is antisymmetric with respect
to the arguments x and y, the operator (7.17) induces an interaction, i.e. that of
forces between subdomains.
y
Equation (7.18) is a generalization of the conservation of momentum equation in
the linearized bond-based peridynamic theory introduced in Ref. 21. In fact, if the
null space of the integral operator in (7.16) is required to contain rigid displacement
elds, i.e. that
T((T
(Ax +c))
T
) = 0 for all constant skew-symmetric tensors A and all
constant vectors c,
then a sucient condition is that (x, y) = (y x)([y x[), where : R
+
R.
Then, from (7.18), we have that
2
u
t
2
=
_
I
(y x) (y x)
([y x[)
(u
u)dx +b x , (7.19)
where :=
2
, which is the linearized peridynamic bond-based model of Ref. 21.
In Ref. 7, analytical conditions on describing the amount of smoothing associated
with the integral operator are given and the well-posedness of the balance of linear
y
In the local case, (7.10) is an abstraction of Cauchys postulate in mechanics. There, the intensive
variable q is the vector momentum density and q is the stress tensor with q n then being the
stress force density at a point of a surface. Then, Cauchys postulate states that the interaction
between two abutting regions occurs at the common interface between the two regions and is given
by the integral of the stress force along that interface. This is exactly a word description of (7.10).
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
538 Q. Du et al.
momentum is discussed. That paper also demonstrates that, for the case of localized
kernels, as 0,
1
2
T((T
u)
T
) (u) 2( u),
which is the Navier operator of linear elasticity for Poisson ratio one-quarter; see
also Ref. 9. The subsequent paper Ref. 27 considers volume-constrained problems
on bounded domains in R and squares in R
2
.
The state-based peridynamic theory (see Ref. 23) requires the consideration of
both volumetric and shear deformations. Similar to the above discussion for bond-
based peridynamic models, we may also formulate the state-based peridynamic
model in terms of the nonlocal operators. Let and be given by (5.14). Then,
the linear state-based peridynamic integral operator (see Ref. 22) is given by
T((T
u)
T
) T
(tr(T
u)I), (7.20)
where and are material constants and T, T
, T
, and T
are given by (3.18a),

(3.18b), (5.11), and (5.12), respectively. The scalar tr(T
u) measures the volu-

metric change, or dilatation, in the material so that tr(T
u)I is a diagonal tensor

representing volumetric stress. This allows us to readily apply the nonlocal calcu-
lus to study the well-posedness of both free-space and volume-constrained linear
peridynamic state-based balance laws, and also suggests why, in the limit as 0,
the operator given by (7.20) leads to the linear Navier operator of elasticity for
linear isotropic materials with general Poisson ratios, e.g. not relegated to a value
of one-quarter; see Ref. 6 for details.
Acknowledgments
Q.D. was supported in part by the US Department of Energy grant DE-SC0005346
and the US National Science Foundation grant DMS-1016073. M.G. was supported
by the US Department of Energy grant number DE-SC0004970 and the US National
Science Foundation grant DMS-1013845. Also supported in part while in residence
at Yonsei University (Republic of Korea) as part of the WCU (World Class Univer-
sity) program under award number R31-2008-000-10049-0 of the National Research
Foundation of Korea. Sandia is a multiprogram laboratory operated by Sandia Cor-
poration, a Lockheed Martin Company, for the US Department of Energy under
contract DE-AC04-94AL85000. Supported in part by the US Department of Energy
grant number FWP-09-014290 through the Oce of Advanced Scientic Computing
Research, DOE Oce of Science. K.Z. was supported in part by the US Depart-
ment of Energy grant DE-SC0005346 and the US National Science Foundation
grant DMS-1016073.
References
1. R. Almeida, A. Malinowska and D. Torres, A fractional calculus of variations for mul-
tiple integrals with application to vibrating string, J. Math. Phys. 51 (2010) 033503.
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
2. F. Andreu, J. Mazon, J. Rossi and J. Toledo, Nonlocal Diusion Problems, Mathe-
matical Surveys and Monographs, Vol. 165 (Amer. Math. Soc., 2010).
3. G. Chen, M. Torres and W. Ziemer, GaussGreen theorem for weakly dierentiable
vector elds, sets of nite perimeter, and balance laws, Commun. Pure Appl. Math.
62 (2009) 242304.
4. T. Coulhon and A. Grigoryan, Random walks on graphs with regular volume growth,
Geom. Funct. Anal. 8 (1998) 656701.
5. Q. Du, M. Gunzburger, R. Lehoucq and K. Zhou, Analysis and approximation of
nonlocal diusion problems with volume constraints, Technical report 2011-3168J,
Sandia National Laboratories, Albuquerque (2011), to appear in SIAM Rev.
6. Q. Du, M. Gunzburger, R. Lehoucq and K. Zhou, Analysis of the volume-constrained
peridynamic Navier equations of linear elasticity, Technical Report SAND 2012-5553J,
Sandia National Laboratories, Albuquerque, 2012.
7. Q. Du and K. Zhou, Mathematical analysis for the peridynamic nonlocal continuum
theory, Math. Model. Numer. Anal. 45 (2011) 217234.
8. L. Ehrenpreis, On the theorem of kernels of Schwartz, Proc. Amer. Math. Soc. 7
(1956) 713718.
9. E. Emmrich and O. Weckner, On the well-posedness of the linear peridynamic model
and its convergence towards the Navier equation of linear elasticity, Commun. Math.
Sci. 5 (2007) 851864.
10. H. Gask, A proof of Schwartzs kernel theorem, Math. Scand. 8 (1960) 327332.
11. G. Gilboa and S. Osher, Nonlocal operators with applications to image processing,
Multi. Model. Simulat. 7 (2008) 10051028.
12. M. Gunzburger and R. Lehoucq, A nonlocal vector calculus with application to non-
local boundary value problems, Multi. Model. Simulat. 8 (2010) 15811620.
13. L. Ignat and J. Rossi, Decay estimates for nonlocal problems via energy methods,
J. Math. Pures Appl. 92 (2009) 163187.
14. X. Jiang, L.-H. Lim, Y. Yao and Y. Ye, Statistical ranking and combinatorial Hodge
theory, Math. Prog. 127 (2011) 203244.
15. R. Lehoucq and O. von Lilienfeld, Translation of Walter Nolls Derivation of the
fundamental equations of continuum thermodynamics from statistical mechanics,
J. Elast. 100 (2010) 524.
16. O. Lezoray, V.-T. Ta and A. Elmoataz, Partial dierences as tools for ltering data
on graphs, Pattern Recogn. Lett. 31 (2010) 22012213.
17. W. Noll, Die Herleitung der Grundgleichungen der Thermomechanik der Kontinua aus
der statistischen Mechanik, Indiana U. Math. J. 4 (1955) 627646; original publishing
journal was the J. Rational Mech. Anal. See the English translation of J. Elast. 100
(2010) 524.
18. W. Noll, Thoughts on the concept of stress, J. Elast. 100 (2010) 2532.
19. F. Schuricht, A new mathematical foundation for contact interactions in continuum
physics, Arch. Rational Mech. Anal. 184 (2007) 495551.
20. L. Schwartz, Espaces de fonctions dierentiables `a valerus vectorielles, J. Anal. Math.
4 (1954)(1955) 88148.
21. S. Silling, Reformulation of elasticity theory for discontinuities and long-range forces,
J. Mech. Phys. Solids 48 (2000) 175209.
22. S. Silling, Linearized theory of peridynamic states, J. Elast. 99 (2010) 85111.
23. S. Silling, M. Epton, O. Weckner, J. Xu and E. Askari, Peridynamic states and con-
stitutive modeling, J. Elast. 88 (2007) 151184.
24. S. Silling and R. Lehoucq, Peridynamic theory of solid mechanics, Adv. Appl. Mech.
44 (2010) 7369.
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
January 4, 2013 9:31 WSPC/103-M3AS 1250054
540 Q. Du et al.
25. V. Tarasov, Fractional vector calculus and fractional Maxwells equations, Ann. Phys.
323 (2008) 27562778.
26. D. Zhou and B. Scholkopf, Regularization on discrete spaces, in Pattern Recognition,
Proc. 27th DAGM Symp., eds., W. G. Kropatsch, R. Sablatnig and A. Hanbury,
Lecture Notes in Computer Science, Vol. 3663 (Springer, 2005), pp. 361368.
27. K. Zhou and Q. Du, Mathematical and numerical analysis of linear peridynamic mod-
els with nonlocal boundary conditions, SIAM J. Numer. Anal. 48 (2010) 17591780.
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.
This article has been cited by:
1. Qiang Du, Max Gunzburger, R. B. Lehoucq, Kun Zhou. 2012. Analysis of the Volume-
Constrained Peridynamic Navier Equation of Linear Elasticity. Journal of Elasticity .
[CrossRef]
2. Raymond A. Wildman, George A. Gazonas. 2012. A perfectly matched layer for
peridynamics in two dimensions. Journal of Mechanics of Materials and Structures 7:8-9,
765-781. [CrossRef]
M
a
t
h
.

M
o
d
e
l
s

M
e
t
h
o
d
s

A
p
p
l
.

S
c
i
.

2
0
1
3
.
2
3
:
4
9
3
-
5
4
0
.

D
o
w
n
l
o
a
d
e
d

f
r
o
m

w
w
w
.
w
o
r
l
d
s
c
i
e
n
t
i
f
i
c
.
c
o
m
b
y

U
N
I
V
E
R
S
I
T
Y

O
F

M
I
L
A
N
O

-

B
I
C
O
C
C
A

B
I
B
L
I
O
T
E
C
A

D
I

A
T
E
N
E
O

o
n

0
3
/
2
8
/
1
3
.

F
o
r

p
e
r
s
o
n
a
l

u
s
e

o
n
l
y
.

January 4, 2013 9:31 WSPC/103-M3AS 1250054

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

January 4, 2013 9:31 WSPC/103-M3AS 1250054

Uploaded by

Copyright:

Available Formats

January 4, 2013 9:31 WSPC/103-M3AS 1250054

Mathematical Models and Methods in Applied Sciences

Department of Scientic Computing,

:= (y, x), := (x, y),

) are dened in the usual manner. For

u vdx for u(x), v(x) L

dydx for (x, y), (x, y) L

((x, y) + (y, x))dydx

((x, y) + (y, x))dydx = 0

denotes the space of distributions. The corresponding duality

, and, for U V and (U V )

) ; note that the integrability of does not

) is an antisymmetric two-point function so that, by (2.4b),

, the negative of the adjoint of the nonlocal gradient oper-

serve to dene two forms for the nonlocal

u) (where u denotes a one-point function)

maps a vector point function to a second-

acting on a point tensor function

= (), these identities can be written

can also be viewed as nonlocal analogs of the

(w) as the nonlocal

(w) as the nonlocal

) dy to dene the operator we

) dy for this purpose

) dydx = 0, and (3.14a).

(u) in (4.8b), and = (

(u) in (4.8c), respectively.

contribute to that ux. Similar observations can be made for the

. However, analogous results also hold for the nonlocal gradient

(x) be a standard Gaussian with variance .

to the Dirac delta measure in the sense of

(u) dy = (x, x) u(x) x R

on a point function is related, in a distributional

u(x)(x, x) ndA, (5.7)

u(x) (x, x)dx

u(x) (x, x)dx.

u(x) (x, x)dx. (5.8)

. Then, in Sec. 5.2.2, the weighted operators are rigorously shown to be

(u)(x) := T((x, y)u(x)) for x R

(v)(x) := (((x, y)v(x)) for x R

(w)(x) := (((x, y)w(x)) for x R

denote the nonlocal adjoint operators given

(u)(x, y)(x, y)dy for x R

(v)(x, y)(x, y)dy for x R

(w)(x, y)(x, y)dy for x R

(u)(x, y) ((x, y)u(x))dydx

() corresponding to the operator T

(U)(x) := T((x, y)U(x))(x) for x R

(u)(x, y)(x, y)dy for x R

, converge, as 0, to the corresponding

and their adjoint operators T

are bounded linear operators

(u) to the divergence dier-

) and z = y x, the function : R

; q) (the rate of change of the amount of the intensive

) (the rate at which the

q(x, t)dx and T(

b(x, t)dx (7.7)

(q), where the adjoint operator T

() is exactly that which

() is dened in (3.18b). We then have, from (7.15),

are given by (3.18a),