AAS 03-284

THE ACCURATE ELEMENT METHOD: A NOVEL METHOD FOR

INTEGRATING ORDINARY DIFFERENTIAL EQUATIONS
Paul Cizmas and Maty Blumenfeld

This paper presents the development of a new methodology for numerically solving ordinary differential equations. This methodology,
which we call the accurate element method (AEM), breaks the connection between the solution accuracy and the number of unknowns.
The differential equations are discretized by using finite elements,
like in the finite element method (FEM). A key feature of the AEM
is the methodology developed for eliminating unknowns inside the
element by using the relations provided by the governing equations.
The results of the AEM applied to a second-order ordinary differential equation (ODE) show that the degree of the local approximation
function does not affect the number of unknowns used to discretize
the differential equation. For a second-order ODE it is shown that the
discretized solution has the same number of unknowns, whether the
local approximation (or interpolation) function is a third-degree or
a nineteen-degree polynomial. The implication of this result is that
higher-degree interpolation functions can be used without increasing
the number of unknowns. For a specified accuracy, the usage of elements with high-degree interpolation functions leads to a reduction
in the necessary number of elements. Numerical results using the
AEM, the shooting method and the relaxation method are presented
and compared for several ODE boundary value problems.

INTRODUCTION
Numerical solution of differential equations is of paramount importance for most of the fields of
engineering and science because it provides a way to investigate the governing equations of various
physical phenomena. As the field of numerical methods for solving differential equations matured,
some of the algorithms were preferred for various reasons and are currently part of the classical
methodologies. If one restricts this discussion to ordinary differential equations, it is easy to identify
the most popular methods of the moment: (1) Runge-Kutta methods, (2) Richardson extrapolation
and its particular implementation as Bulirsch-Stoer method, and (3) predictor-corrector methods, 24
p.702. The most general and widely used methods for partial differential equations include the
finite difference method, finite volume method and finite element method (FEM). The common

Assistant Professor, Department of Aerospace Engineering , Texas A&M University, College Station, TX 778433141 . E-mail: cizmas@tamu.edu.

Professor Emeritus, Chair of Strength of Materials, PolitehnicaUniversity Bucharest, Romania, Email:mblum@itcnet.ro.

goal of these methods is to approximate as well as possible the solution of the governing equations
describing various physical phenomena.
Several options are currently used to improve the accuracy of these methods: (1) increasing
the degree of the polynomials that locally approximate the solution, that is, p-type refinement; (2)
decreasing the size of the elements which discretize the computational domain in order to reduce
the variation of the unknowns over the element, that is, h-type refinement; (3) using anisotropic
elements, that is, having different orders of expansion for each spatial direction in an element; (4)
using non-conforming elements with either h-type or p-type non-conformities; (5) using adaptive
refinement strategies.
High-order finite element methods9, 26 and spectral element methods17, 20, 23, 27 are commonly
used to produce high accuracy solutions. Spectral element methods combine the high accuracy of
spectral methods6, 12 and the flexibility of the finite element method. The Galerkin spectral element
method, a subset of the spectral element method, is similar to the high-order finite element method
in the sense that both are weighted residual techniques. A basic difference between the two methods
is the fact that the weighting functions used in the Galerkin spectral element method are either
Chebyshev or Legendre polynomials.5, 16
Anisotropic and non-conforming elements are also used to improve the solution accuracy of
certain problems. The non-conforming elements can be either h-type or p-type non-conforming.
Various special formulations have been developed for anisotropic and non-conforming elements,
such as the constrained approximation method,8, 22, 25 the mortar element method,1, 2 the FETI
method,19 the transition element method13, 21 and the iterative patching method.10, 15
Another option for producing high accuracy solutions is to use adaptive refinement strategies.
Adaptive methods have the ability to automatically adjust themselves to improve the solution of a
problem.22 The adaptive methods are mainly used for: (1) problems for which a properly resolved
initial mesh is not available; and (2) unsteady problems for which the initial mesh is not good
enough at some instants of the solution. The local error estimation and the adaptive strategy are
two important aspects of the adaptive methods.7, 14, 28
For all the methods currently used to solve differential equations, the outcome of increasing solution accuracy is an increase in the number of unknowns. To minimize the number of unknowns,
each method tries to optimize the relation between the degree of the approximation function and
the size of the element, such as in the h p adaptive finite element or spectral element methods.
This paper presents a new methodology for the numerical solution of differential equations which
breaks the connection between the solution accuracy and the number of unknowns. Consequently,
the number of unknowns corresponding to high-accuracy solutions obtained using high-order approximation polynomials will be equal to the number of unknowns corresponding to low-order
approximation polynomials. This new methodology, which we call the accurate element method
(AEM), will be applied herein to numerically solve linear and nonlinear ODEs. The results shown in
this paper indicate that the degree of the local approximation function does not affect the number
of unknowns used to discretize the differential equation.

THE ACCURATE ELEMENT METHOD (AEM)

The AEM will be presented by comparing it to the FEM. The strategies of both the AEM and
the FEM can be summarized by the following three steps: (1) discretization of the domain D on
which the governing equations must be integrated; (2) local approximation of the solution of the
governing equations, in which the information given by each element is concentrated at the nodes
of the element; and (3) reconstruction of the domain D, obtained by writing nodal equations that
bring together the information given by the elements adjacent to each node. The solution of the
2

system of equations obtained in this way represents the nodal unknowns.

The first and third steps of the AEM are identical to those of the FEM. The two methods differ
only in how the second step of their common strategy is accomplished. Note that the second step
is crucial in obtaining an accurate solution of the governing equations. As will be shown herein,
the AEM uses a much smaller number of elements than the FEM to obtain a similar accuracy.
In step two, the local approximation of the solution in the FEM is done by using interpolation
functions (IF). Usually these interpolation functions are low degree polynomials that lead to poor
approximations. Consequently, the number of elements used in the FEM must be large in order
to obtain a result with reasonable accuracy. The local approximation in the AEM will be done
by high-degree polynomial functions. Because the functions used for local approximation are highdegree interpolation functions (HIF), only a small number of elements is needed to obtain an
accurate solution in the AEM. The example presented in the next section shows that the accuracy
obtained in the AEM by using two elements with a seventh-degree polynomial HIF is better than
the accuracy obtained in the FEM by using 100 elements with a linear IF.
The degree of the polynomials used in the AEM is usually higher than seven, since, as will be
shown herein, the computational effort in the AEM is not affected by the degree of the polynomial.
This is an important difference between the AEM and the FEM. High-degree polynomials may be
used for the IF in the FEM, but this increases the number of unknowns of the discretized system
of equations. In the AEM, the degree of the interpolation function used in the local
approximation does not affect the number of unknowns of the discretized system of
equations. This is an important advantage of the AEM compared to the finite element, finite
volume or finite difference methods.
To illustrate how the AEM can be implemented to solve ODEs, three cases will be presented:
(1) a linear ODE with constants coefficients; (2) a linear ODE with variable coefficients; (3) a
nonlinear ODE. All cases shown herein are second-order ODEs.

AEM Applied to Linear ODEs

To introduce the methodology of the AEM, let us start by presenting a simple but yet meaningful
case, the solution of a linear, second-order ordinary differential equation:
d2
+ c(x) + d(x) = 0,
dx2

(1)

where c is a constant and d(x) is a function of the independent variable x. Let us consider the
following boundary conditions
1 := (x = 0) = 0
2 := (x = 1) = 0.

(2)

Let us consider a HIF that is a seventh-degree polynomial:

(x) = C0 + C1 x + C2 x2 + C3 x3 + + C7 x7

(3)

and let us write using a natural coordinate = x/L, where L is the length of the element.
Equation (3) can be written in vectorial form using the natural coordinate :
() = [
][CL],

(4)

where
[
] = [1 2 . . . 7 ]
[CL] = [C0 C1 L C2 L2 . . . C7 L7 ]T .
The first goal is to write the local interpolation function (x) as a function of the values of (x)
and its derivatives at nodal locations. The values of the function and its derivatives at both ends
of the element can be written as:

(0)
1

(1)

L1 1 0 0 0 0 0
0
0
2 (2)
0
1
0
0
0
0
0
0
L 1

0 0 2 0 0 0
0
0

L3 (3)
0 0 0 6 0 0
0
0

1
=
[CL]
(5)

(0) 1 1 1 1 1 1
1
1

0 1 2 3 4 5
2
6
7

0 0 2 6 12 20 30 42
L(1)

2
2 (2)
0 0 0 6 24 60 120 210
L 2
(3)
L3 2
or

[B] = [A][CL].

(6)
(k)

(0)

The notation i := i and i

[B] matrix can be rearranged as:

0
1 0
10
00 00 "
#

(0)
0
1

+L
[B] = 00 01
0

(0)

2
0
0
0

0
0 0
0
0 0

:=
0
0
0
0
0
1
0
0

d k i
,i
dxk

[1, 2] and k 1 has been used in equation (5). The

"
#

(1)
1
2
+L

(1)
2

0
0
1
0
0
0
0
0

0
0
0
0
0
0
1
0

"
#

(2)
1
3
+L

(2)
2

0
0
0
1
0
0
0
0

0
0
0
0
0
0
0
1

"
#

(3)
1
(7)

(3)
2

and the inverse of the [A] matrix needs to be computed only once. The vector of coefficients [CL]
can be written using equations (6) and (7) as:
#
"
#
"
#
"
#
"
(1)
(2)
(3)
(0)
1
1
1
1
3
1
2
+ L [F 83]
(8)
[CL] = [A] [B] = [F 80]
+ L [F 81]
+ L [F 82]
(1)
(2)
(3)
(0)
2
2
2
2
where

[F 80] =

1
0
0
0
0
0
0
0
35
35
84 84
70
70
20 20

, [F 81] =

0
0
0
0
1
0
0
0

0
0
1
0

0
0
0
0
,
[F
82]
=

20 15
5
2 10
45
39
20 14
36 34
15
13
10
10
4 4

[F 83] =
6

0
0
0
0
0
0
1
0 .
4 1

6
3
4 3
1
1

By substituting the [CL] vector from equation (8) into equation (4), the local approximation
function (x) is written as a function of the nodal values of the function and its derivatives.
Let us derive a relation similar to but more general than the stiffness matrix relation. This
(k)
(0)
relation should provide a link between the variables i and the variable derivatives i , k 1.
This relation, which is a generalization of the stiffness matrix relation, will be called the complete
transfer relation (CTR).3 The derivation of the CTR can be done based on the physical phenomenon
that is described by the governing equations or by using the Galerkin method. The latter approach
will be used herein, because of its generality.
Let us use the following weighting function
h
i
x
Lx

1 + 2 = [ N ]
.
w(x) =
2
L
L

(9)

Using the Galerkin projection one obtains

Z

L
0

]T d dx + c
[N
dx2

]T (x)dx +
[N

]T d(x)dx = [0].
[N

(10)

After integrating by parts the first term, equation (10) becomes

"
#
"
#
Z L h
h
i
(1)
1 h 1 1 i (0)
1 Lx i
1
1
0
1
+

c
(x)dx =
(0)
(1)
x
1
0 1
L 1
2
2
0 L
=

L
0

1 h Lx i
d(x)dx.
x
L

(11)

Let us now replace the function of the third term of the left-hand side by the expression given
in equation (4), where the vector of coefficients [CL] is written as in equation (8). Equation (11)
becomes
"
#
"
"
"
#
#
#
h
i
(1)
(0)
(1)
1 h 1 1 i (0)

1
0
2
1
1
1
1
+ 0 1
cL[G0F 0]8
cL [G0F 1]8

(0)
(1)
(0)
(1)
1
L 1
2
2
2
2
3

cL [G0F 2]8

"

(2)

1
(2)
2

cL [G0F 3]8

"

(3)

1
(3)
2

L
0

1 h Lx i
d(x)dx
x
L

(12)

where
1 h 17 10 i
1 h 13 5 i
,
[G0F
1]
=
8
5 13
17 ,
36
252 10
1 h 7 5 i
1 h 5 4 i
[G0F 2]8 =
,
[G0F
3]
=
.
8
1008 5 7
15120 4 5
Note that the third term of left-hand side of equation (11) is responsible for introducing errors in
FEM, because the function in the integral is approximated by a low degree polynomial. This
[G0F 0]8 =

term is responsible for introducing errors in the FEM. Unlike the FEM, the AEM uses a high degree
polynomial to approximate (x). The result of this approach is that equation (12) has 8 unknowns
(3)
(0) (0)
1 , 2 , . . . , 2 . There are, however, only 4 available relations: 2 equations (12) and 2 boundary
conditions (18). Consequently, 4 additional relations are necessary.
Usually, when more information is needed, this information is obtained either from some relations with other elements or by accepting a reasonable approximation. Neither of these options
are used in the AEM. Instead, the necessary information is obtained accurately from inside the
element, that is, from the governing equations themselves. Consequently, the second and third
derivatives are
#
# "
#
"
"
(0)
(0)
(2)
d1
1
1
(13)

= c
(0)
(0)
(2)
d2
2
2
and
"

(3)

1
(3)
2

= c

"

(1)

1
(1)
2

"

(1)

d1
(1)
d2

(14)

where
(1)

(0)

(0)

d1 := d(x = 0), d2 := d(x = L), d1 :=

d
d
(1)
d(x) |x=0 , d2 :=
d(x) |x=L .
dx
dx

Consequently, the equation (12) becomes

"
#
"
#
(0)
(1)
1
1
A
+B
=C,
(0)
(1)
2
2

(15)

where
1 h 1 1 i
2 3
1 cL [G0F 0]8 + c L [G0F 2]8
L 1
h
i
B = 10 10 cL2 [G0F 1]8 + c2 L4 [G0F 3]8
"
#
"
#
Z L h
(0)
(1)
1 Lx i
d1
d1
3
4
d(x)dx cL [G0F 2]8
cL [G0F 3]8
C=
(0)
(1)
x
d2
d2
0 L
A=

(2)

(2)

(3)

(16)

(3)

The unknowns 1 , 2 , 1 and 2 , which we call the apparent unknowns,4 have been eliminated by using the governing equations.
To illustrate the methodology, let us consider that in equation (1) c = 4 and d(x) = 4x 3
2
8x 18x 8, that is
d2
+ 4 4x3 8x2 18x 8 = 0
dx2

(17)

with the boundary conditions

1 := (x = 0) = 0
2 := (x = 1) = 0.

(18)

The analytical solution of equation (17) with the specified boundary conditions is:
(x) = [(cos(2) 7)/ sin(2)] sin(2x) cos(2x) + x3 + 2x2 + 3x + 1.

(19)

The values of the A, B and C matrices corresponding to the seventh- and nineteen-degree interpolation function are obtained from equations (16):
HIF8
A=

1/3 1.476190476190476
1.476190476190476
1/3

0.7354497354497356
0.1544973544973545
0.1544973544973545 0.7354497354497356
h
i
6.915343915343915
C = 11.62751322751323
B=

HIF20

A=

0.310606670618465 1.417281845711744
1.417281845711744 0.310606670618465

0.7082522309668606
0.153520002220184
0.153520002220184 0.7082522309668606
i
h
6.571622875498251
C = 11.13461085637016
B=

For comparison, let us show the A, B and C matrices obtained using a linear interpolation
function (IF2)

1/3
5/3
A = 5/3 1/3
B=
C=

1
0
0 1

23.6/3
12.8

.
(1)

Table 1 shows the variation of 1 as a function of the degree of the polynomial in the HIF.
(1)
This solution is based on one element. For comparison, the value of 1 calculated using the
(1)
FEM with an IF2 is also included. The exact value is obtained from the solution (19) as 1 =
-13.31181749284520. The variation of the error shows that an accurate solution can be obtained
with only one AEM element with HIF12.
(1)

IF2
HIF4
HIF8
HIF12
HIF16
HIF20
exact

1
-7.86666
-12.85714285714286
-13.31157287382312
-13.31181747484904
-13.31181749284484
-13.31181749284521
-13.31181749284520

Error [%]
-40.90
-3.41
-1.83e-3
-1.35e-7
-2.72e-12
4.00e-14

Table 1 Solution based on one element.

(1)

The variation of 1 as a function of the number of elements is shown in Table 2. Results are
shown for the FEM with IF2 and the AEM with HIF4 and HIF8. The advantage of using the AEM
7

and high-degree polynomials is obvious: two AEM elements with HIF8 produce more accurate
results than 100 FEM elements with IF2. Four AEM elements with HIF8 produce a solution with
an error of the order of 108 %.
Number Elem.
1
2
4
6
10
20
50
100

IF2
(1)
1
-7.86666
-11.632291666
-12.85655835183100
-13.10604588668801
-13.23707977753565
-13.29306202953536
-13.30881341149338
-13.31106635773660

Error [%]
-40.90
-12.62
-3.42
-1.55
-0.56
-0.14
-2.26e-2
-5.60e-3

HIF4
(1)
1
-12.85714285714286
-13.28681646362417
-13.31028065459640
-13.31151472702216
-13.31177830616024
-13.31181504502042
-13.31181743019060
*

Error [%]
-3.41
-1.87e-1
-1.15e-2
-2.27e-3
-2.94e-4
-1.83e-5
-4.70e-7
*

HIF8
Error [%]
-1.83e-3
-4.74e-6
-1.84e-8
*
*
*
*
*

(1)

Table 2 Variation of the value of 1 as a function of the number of elements and degree of the
HIF/IF.

AEM Applied to Nonlinear ODEs

This section presents the methodology for solving nonlinear ODEs using the AEM. In order
to solve nonlinear ODEs using the AEM, a methodology for solving linear ODEs with variable
coefficients must be developed first. For this reason, the first part of this section is devoted to solving
linear ODEs with variable coefficients. The second part of this section presents the methodology
for solving nonlinear ODEs using the AEM.
Let us consider the following second-order ODE with variable coefficients
a(x)

d2
d
+ b(x)
+ c(x)(x) + d(x) = 0.
2
dx
dx

(20)

The CTR corresponding to equation (20) will be derived using the Galerkin method. For this
reason, it is convenient to rewrite equation (20) in a transformed form

d
d
d
da(x)
db(x) d2 a(x)
a(x)
+
b(x)
+ c(x)
+
+ d(x) = 0.
(21)
dx
dx
dx
dx
dx
dx2
Let us use for the Galerkin projection the weighting function w(x) from equation (9). The result
of the Galerkin projection applied to equation (21) is:

Z
Z
T d
T d
d
da(x)

N
a(x)
dx +
b(x)
dx +
N
dx
dx
dx
dx
L
L

Z
Z
T
T
db(x) d2 a(x)

d(x)dx = 0.
c(x)
N
N
(22)
+
dx
+
dx
dx2
L
L

For brevity, only the derivation of the first term of equation (22) will be presented herein. The
same methodology is applied for the derivation of the other terms in equation (22). The first term
of equation (22) will be calculated using integration by parts:
T

T d
d N
d T
d
d 2

[T1 ] =
N
a(x)
+ N a(x) .
a(x)
dx =
dx
dx
dx
dx
dx 1
L
L
Z

(23)

The first term of the right-hand side is denoted [T1a ] and the second term is denoted [T1b ]. Taking
8

 ] from equation (9), the term [T1a ] becomes

into account the expression of [N
[T1a ] =

T
Z

d N
d
d
1 h 1 i
a(x) dx.
a(x) dx =
1
dx
dx
L
dx
L

Integrating by parts again, [T1a ] becomes

Z
1 h 1 i
da(x)
[T1a ] =
dx + (a(L)2 a(0)1 ) .

1
L
L dx
(k)

(k)

d a
d a
:= dx
Using the notations introduced in equation (5) and a1 := dx
k (0) and a2
k (L), the term [T1 ]
becomes
"
#"
# "
#"
#
Z
(0)
(0)
(0)
(0)
(1)
1 h 1 i
da(x)
1
a2
a1
1
0
a1
1
[T1 ] =
dx

.
(0)
(0)
(0)
(1)
1
L
L a(0)
a2
2
0 a2
2
L dx
1

A similar approach is used for the other terms in equation (22), such that the general form of the
CTR is
"
#"
# "
#"
#
(0)
(0)
(1)
(0)
(0)
(0)
(1)
a1 /L + b1 a1
a2
1
a1
0
1
+
+
(0)
(0)
(0)
(1)
(0)
(0)
(1)
a1 /L
a2 /L b2 a2
2
0
a2
2

Z
Z
Z
T
da(x)
db(x) d2 a(x) T
1 h 1 i
dx.(24)
b(x) 2
dx
c(x)
+
N
N dx =
1
2
L
dx
dx
dx
L
L
L

Having derived the CTR for linear ODEs with variable coefficients, the next step is to produce
an algorithm for solving nonlinear ODEs. To present the methodology for solving nonlinear ODEs,
let us choose the following second-order, nonlinear ODE
(2) + (3(0) + 4(1) 14)(0) + (8 40x + 85x2 62x3 + 2x4 + 12x5 3x6 ) = 0
(0)

(1)

(25)

with the boundary conditions 2 := (x = L) = 0 and 2 := d

dx (x = L) = 0. For simplicity,
let us assume L = 1. The nonlinearity is due to the 3(0) (0) and 4(1) (0) terms. In order to
eliminate the nonlinearity, the terms in the parentheses (3(0) + 4(1) 14) will be approximated
using HIF. To simplify the algebra, a third-degree polynomial HIF will be used in this example.
The derivation of the vector of coefficients [CL] is similar to the derivation presented in section ,
where [CL] is given by equation (8). For a third-degree polynomial HIF, the expression of the vector
of coefficients [CL] is
#
"
#
"
(0)
(1)
1
1
+ L [F 41]
,
(26)
[CL] = [F 40]
(0)
(1)
2
2
where
[F 40] =

"

" 0
0 #
1
0 #
0
0
and [F 41] = 21 10 .
3
3
1
1
2 2

Using equation (4), the HIF turns out to be the well known Hermite function
(x) =

L3 3Lx2 + 2x3 (0) 3Lx2 2x3 (0) L2 x 2Lx2 + x3 (1) Lx2 + x3 (1)
1 +
2 +
1 +
2 . (27)
L3
L3
L2
L2

Replacing the terms in the parenthesis (3(0) + 4(1) 14) by the HIF from equation (27), equation (25) becomes
(2) + (K0 + K1 x + K2 x2 + K3 x3 )(0) + (8 40x + 85x2 62x3 + 2x4 + 12x5 3x6 ) = 0, (28)
where
(0)

(0)

(1)

(0)

(1)

(0)

(1)

K0 = 71 14, K1 = 241 131 , K2 = 151 + 61 , K3 = 61 + 31 .

(29)

As a result, the nonlinear ordinary differential equation (25) has been replaced by a linear ODE
with variable coefficients (28). The CTR of equation (28) is derived using the general form obtained
in equation (24). Having b(x) = 0 and L = 1, equation (24) becomes
#
"
#
"
h
i
h
i
(0)
(1)
1
1
1 1
1
0
+ 0 1
+
(0)
(1)
1
1
2 = 0
2 = 0
Z
Z
T

T
2
3
dx.

d(x) N
(30)
K0 + K1 x + K2 x + K3 x N (x)dx =
L

After substituting the HIF, (x), from equation (27), the Ki coefficients from equation (29) and
d(x) from equation (25), the CTR equation (30) becomes

i (1)
i (0) h

h
(0) 2
2/5
89/210 (0) (1) +
1
0
1 1
1
1

+
+
+
1
1
1
0 1
1
1
17/35
23/210
0
0

2
(0)
(1)
(1)
31/840
(31)
+ 49/10 1 + 17/10 1 1357/840 = 0,
1
11/10
7/15
333/840
1/120
(0)

(1)

or using the notation u := 1 and v := 1

336u2 356uv 31v 2 + 4956u + 1428v = 1357

408u2 + 92uv + 7v 2 + 924u + 392v = 999

(32)

Two sets of real solutions are obtained by solving the nonlinear system of equations (32)
u = 2,

v = 5 and

u = 15.28,

v = 160.13.

Only the first set of solutions leads to an acceptable solution which is

(x)

(0)

(1)

(2x3 3x2 + 1)1 + (x3 2x2 + x)1 = (2x3 3x2 + 1)(2) + (x3 2x2 + x)(5) =
x3 4x2 + 5x 2.
(33)
(1)

(0)

The second set of solutions (u := 1 = 15.28, v := 1 = 160.13) produces the solution

(x) = 129.57(x 1)2 (x 0.117929)
which does not satisfy the governing equation (25).

10

AEM vs. Existing ODE Solvers

To compare the AEM against existing ODE solvers, let us consider the second-order ODE with
variable coefficients:
(1 + 2x + 3x2 + 3x3 + x4 )

d2
d
+ (2 5x + x2 + 3x3 + 4x4 ) + (1 + x + 2x2 + 3x3 + 4x4 ) + f (x) = 0,
dx2
dx

where f (x) has been chosen to be f (x) = 5 + 37x + 8x2 + 166x3 59x4 28x5 298x6 725x7
1144x8 1053x9 1056x10 447x11 160x12 16x13 , such that the solution of the ODE is the
ninth-degree polynomial (x) = 3 + 5x 4x2 + 2x3 6x4 + x5 + 2x6 + 2x7 + x8 + 4x9 . Various
integration lengths have chosen with various two-point boundary conditions, as shown in Table 3.
The AEM has been compared against the shooting method and the relaxation method. All the
methods were implemented in FORTRAN. The shooting method used the Runge-Kutta method
implemented in the odeint routine.24 One hundred years after they have been introduced,18 the
Runge-Kutta methods are still the workhorse for solving ODEs,24 p. 703. There were several
improvements of the Runge-Kutta method, such as Runge-Kutta-Gill11 and (3/8) Runge-Kutta,
but these modifications did not fundamentally changed the basic idea of the method. Other method
for solving ODEs, such as the Bulirsch-Stoer method and the predictor-corrector method, can be
faster than the Runge-Kutta in certain circumstances, but the differences are rather limited. Given
these facts and due to its robustness, the fourth-order Runge-Kutta method has been selected as
one of the methods to be compared against the AEM.
The relaxation method used the solvde routine.24 The computations were done on an SGI
Indigo2 computer, with an R8000 processor. The results shown in Fig. 1 indicate that for the same
computational time, the accuracy of the AEM results with a 13th-degree polynomial (HIF14) was
approximately five orders of magnitude better than the accuracy of the shooting and relaxation
methods. In addition, for the same error level, the AEM with HIF14 interpolation functions
produced solutions at least two orders of magnitude faster than the shooting and relaxation
methods. The relaxation method diverged when the integration was equal or exceeded x = 2.
Case
1
2
3
4

Domain
xlef t xright
0
1
0
2
0
3
0
4

Boundary Values
lef t right
3
10
3
2637
3
90918
3
1154647

Table 3 Integration cases.

CONCLUSIONS
A new method, which we call the accurate element method, has been presented herein and
applied to linear and nonlinear ODEs. A key feature of the AEM was the methodology developed
for eliminating unknowns inside the element by using the relations provided by the governing
equations. The results of the AEM applied to a second-order ODE showed that the order of the local
approximation function did not affect the number of unknowns used to discretize the differential
equation. This is a breakthrough in the paradigm used to approach the numerical solution of
differential equations. For a second-order ODE it was shown that the discretized solution has
the same number of unknowns, whether the local approximation (or interpolation) function is a
11

0<=x<=1

6
7
AEM, CF10
AEM, CF14
Relax, eps=5e6
Relax, eps=5e7
Relax, eps=5e9
Shoot, eps=1e11, h1=1e8
Shoot, eps=1e11, h1=1e6
Shoot, eps=1e8, h1=1e6

8
9
10
11
12
2.5

1.5

0.5
0
0.5
log10(time) [sec]

0<=x<=2

log10(error) [%]

log10(error) [%]

1.5

6
7
8
AEM, CF10
AEM, CF14
Shoot, eps=(1e8,1e14), h1=1e6

9
10
11
2

12
2.5

2.5

1.5

0<=x<=3

log10(error) [%]

log10(error) [%]

6
7
8

9
AEM, CF10
AEM, CF14
Shoot, eps=(1e12,1e8), h1=1e8

0.5
0
0.5
log10(time) [sec]

1.5

2.5

10

1.5

2.5

10

12
2.5

1.5

0<=x<=4

11

(ii)

(i)
1

0.5
0
0.5
log10(time) [sec]

AEM, CF10
AEM, CF14
Shoot, eps=(1e12,1e8), h1=1e8

11
1.5

2.5

(iii)

12
2.5

1.5

0.5
0
0.5
log10(time) [sec]

(iv)

Figure 1 AEM vs. Relaxation and Shooting Methods

12

third-degree or a nineteen-degree polynomial. The implication of this result is that higher-degree

interpolation functions can be used without increasing the number of unknowns.
The AEM has been used to solve linear ODEs with constant and variable coefficients and
nonlinear ODEs. For a linear ODE with variable coefficients, the AEM has been compared against
the relaxation method and the shooting method. The accuracy of the AEM results with a 13thdegree polynomial (HIF14) was approximately five orders of magnitude better than the accuracy
of the shooting and relaxation methods. In addition, for the same error level, the AEM with HIF14
interpolation functions produced solutions at least two orders of magnitude faster than traditional
ODE solvers.

REFERENCES
1. C. Bernardi, Y. Maday, C. Mavriplis, and A. T. Patera. The mortar element method applied
to spectral discretizations. In T. J. Chung and G. R. Karr, editors, Finite Element Analysis
in Fluids, The Seventh Int. Conf. on Finite Element Methods in Flow Problems. UAH Press,
1989.
2. C. Bernardi, Y. Maday, and A. T. Patera. A new nonconforming approach to domain decomposition: The mortar element method. In H. Brezis and J. L. Lions, editors, Nonlinear
Partial Differential Equations and Their Applications, number 299 in Pitman Research Notes
in Mathematics. Wiley, 1994.
3. M. Blumenfeld. A New Method for Accurate Solving of Ordinary Differential Equations. Editura Tehnica, Bucharest, 2001.
4. M. Blumenfeld and P. G. A. Cizmas. The accurate element method: A new paradigm for
numerical solution of ordinary differential equations. submitted for publication to the Journal
of the Romanian Academy, January 2003.
5. J. P. Boyd. Chebyshev and Fourier Spectral Methods. Springer, New York, 1989.
6. C. Canuto, M. Y. Hussaini, A. Quarteroni, and T. A. Zang. Spectral Methods in Fluid Dynamics. Springer, New York, 1987.
7. G. F. Carey and A. I. Pehlivanov. Computer methods in applied mechanics and engineering.
Computer Methods in Applied Mechanics and Engineering, 150(1-4):125131, Dec. 1997.
8. R. Y. Chang and C. H. Hsu. A variable-order spectral element method for incompressible viscous flow simulation. International Journal for Numerical Methods in Engineering,
39(17):28652887, 1996.
9. L. Demkowicz, J. T. Oden, W. Rachowicz, and O. Hardy. Toward a universal h p adaptive
finite element strategy: Part 1: Constrained approximation and data structure. Computer
Methods in Applied Mechanics and Engineering, 77(1-2):79112, 1989.
10. D. Funaro, A. Quarteroni, and P. Zanoli. An iterative procedure with interface relaxation for
domain decomposition methods. SIAM Journal on Numerical Analysis, 25(6):12131236, Dec.
1988.
11. S. Gill. A process for the step-by-step integration of differential equations in an automatic
digital computing machine. Proc. Cambridge Philos. Soc., 1951.
12. D. Gottlieb and S. A. Orszag. Numerical Analysis of Spectral Methods: Theory and Applications. SIAM-CMBS, Philadelphia, 1977.
13

13. A. K. Gupta. A finite element method for transition from a fine to a coarse grid. Int. J.
Numer. Meth. Eng., 12:395411, 1978.
14. R. D. Henderson. Dynamic refinement algorithms for spectral element methods. Computer
Methods in Applied Mechanics and Engineering, 175(3-4):395411, 1999.
15. R. D. Henderson and G. E. Karniadakis. Hybrid spectral element-low order methods for
incompressible flows. J. Sci. Comp., 6(2):79, 1991.
16. G. E. Karniadakis and S. J. Sherwin. Spectral/hp Element Methods for CFD. Oxford University
Press, Oxford, 1999.
17. K. Z. Korczak and A. T. Patera. An isotropic spectral element method for solution of the
navier-stokes equations in complex geometry. Journal of Computational Physics, 62(2):361
382, Feb. 1986.
18. W. Kutta. Beitrag zur naherungsweisen integration totaler differentialgleichunge. Zeit. Math.
Physik, 46:435453, 1901.
19. C. Lacour and Y. Maday. Two different approaches for matching nonconforming grids: The
mortar element method and the feti method. BIT, 37(3):720738, 1997.
20. Y. Maday and A. T. Patera. Spectral element methods for the incompressible navier-stokes
equations. In A. K. Noor and J. T. Oden, editors, State of the Art Surveys in Computational
Mechanics, chapter 3. 1989.
21. D. J. Morton, J. M. Tyler, and J. R. Dorroh. A new 3D finite element for adaptive hrefinement in 1-irregular meshes. International Journal for Numerical Methods in Engineering,
38(23):39894008, Dec. 1995.
22. T. J. Oden and L. Demkowicz. Advances in adaptive improvement: A survey of adaptive finite
element methods in computational mechanics. In A. K. Noor and J. T. Oden, editors, State
of the Art Surveys in Computational Mechanics, chapter 13. 1989.
23. A. T. Patera. A spectral method for fluid dynamics: Laminar flow in a channel expansion.
Journal of Computational Physics, 54(3):468485, 1984.
24. W. H. Press, W. T. Vetterling, S. A. Teukolsky, and B. P. Flannery. Numerical Recipes in
FORTRAN - The Art of Scientific Computing. Cambridge, second edition, 1992.
25. W Rachowicz, J. T. Oden, and L. Demkowicz. Toward a universal h p adaptive finite
element strategy: Part iii: Design of h p meshes. Computer Methods in Applied Mechanics
and Engineering, 77(1-2):181212, 1989.
26. J. N. Reddy. An Introduction to the Finite Element Method. McGraw-Hill, New York, second
edition, 1993.
27. E. M. Ronquist and A. T. Patera. Spectral element methods for unsteady navier-stokes equations. In Proceedings of the Seventh GAMM Conference on Numerical Methods in Fluid Mechanics, page 318. Viewig, 1988.
28. W. Wang. Special quadratic quadrilateral finite elements for local refinement with irregular
nodes. Computer Methods in Applied Mechanics and Engineering, 182(1-2):109134, Feb. 2000.

14