No.2 LIAO: Homotopy Analysis Method . .

95
References
[l] P. Hadley, M. R. Beasley and K. Wiesenfeld, Phase locking of Josephson junction
series array, Phys. Rev. B, 1988, 38, 8712-8719.
[2] Y. Kuramoto, Chemical Oscillators, Waves and Turbulence, Springer, New York,
1984.
[3] Minping Qian and Minghua Deng, Segmentation of connected image based on phase
locking, Commun. Nonlinear Sci. Numer. Simul., 1996, l(3), 32-36.
[4] Min Qian, Shu Zhu and Wenxin &in, Dynamics in a chain of overdamped pendula
driven by constant torques, SIAM J. Appl. Math., 1997, 57, 294-305
[5] Minping Qian and Dai Wang, On a system of hyperstable frequency locking persis-
tance under white noise, Preprint (Research Report of Institute of Mathematics and
School of Mathematical Sciences of Peking Univ. No.56), Beijing, 1996.
[6] Guanglu Gong and Minping Qian, The invariant measures, probability flux and cir-
culations of one-dimensional Markov process, LN in Math., 1983, 923, 188-198.
[7] R.N. Bhattacharya and E.C. Waymire, Stochastic Process with Applications, John
Wiley, New York, 1990, 432-436.
[8] A. Erdeiyi, Asymptotic Expansions, Dover, New York, 1956.
Homotopy Analysis Method: A New Analytical
Technique for Nonlinear Problems
Shijun LIAO
(Department of Naval Architecture & Ocean Engineering, Shanghai Jiao Tong Univrsity,
Shanghai 200030, China)
Abstract: In this paper, the basic ideas of a new kind of analytical technique, namely
the Homotopy Analysis Method (HAM), are briefly described. Different from perturbation
techniques, the HAM does not depend on whether or not there exist small parameters in
nonlinear equations under consideration. Therefore, it provides us with a powerful tool to
analyse strongly nonlinear problems. A simple but typical example is used to illustrate
the validity and the great potential of the HAM. Moreover, a pure mathematical theorem,
namely the General Taylor Theorem, is given in appendix, which provides us with some
rational knowledge for the validity of this new analytical technique.
Key Words: Nonlinear analytical technique, strong nonlinearity, homotopy, topology
Introduction
Although the rapid development of digital computers makes it easier and easier to nu-
merically solve nonlinear problems, it is still rather difficult to give analytical expressions
of them. Most of our nonlinear analytical techniques are unsatisfactory. For instance,
The paper was received on Apr. 28, 1997
96 Communications in Nonlinear Science & Numerical Simulation Vo1.2, No.2 (May 1997)
perturbation techniques are widely applied to analyse nonlinear problems in science and
engineering. Unfortunately, perturbation techniques are so strongly dependent on small pa-
rameters appearing in equations under consideration that they are restricted only to weakly
nonlinear problems. For strongly nonlinear problems which dont contain any small param-
eters, perturbation techniques are completely invalid. So, it seems necessary and worthwhile
developing a new kind of analytical technique independent of small parameters.
Liao [l-6] has made some attempts in this direction. Liao [1,2] proposed a new analytical
technique in his Ph. D. dissertation, namely the Homotopy Analysis Method (HAM). Based
on homotopy of topology, the validity of the HAM is independent of whether or not there
exists small parameters in the considered equations. Therefore, the HAM can overcome
the foregoing restrictions and limitations of perturbation techniques so that it provides us
with a possibility to analyse strongly nonlinear problems. Liao [3,4,5,6] used the HAM to
successfully solve some nonlinear problems, while he has been making unremitting efforts to
improve this method step by step. Moreover, Liao [7,8] and Liao & Chwang [9] applied the
basic ideas of the HAM to propose the so-called general boundary element method (GBEM),
which is valid even for those nonlinear problems whose governing equations and boundary
conditions dont contain any linear terms so that it greatly generalizes the traditional BEM.
Furthermore, Liao [lo] even applied the HAM to obtain a pure mathematical theorem,
namely the General Taylor Theorem, which can be applied to greatly enlarge the convergence
radius of a traditional Taylor series. All of these verify the validity and the great potential
of the HAM.
In this paper, we use a typical example to briefly introduce the basic ideas of the HAM
and to further show its validity and great potential.
1. Basic ideas of the Homotopy Analysis Method
Consider the following equation
u(t) + 2tu7t) = 0, u(0) = 1,
(1)
whose exact solution is
u(t) =
1
1+ t2
Assuming t is small, we can obtain by perturbation techniques the following power series
u(t) = 1 - t2 + t4 - t6 + . . . = E(-l)?,
(3)
k=O
whose convergence radius is however rather small, say, p = 1, so that the above perturbation
approximation is nearly worthless.
Now we apply the HAM to solve it. First, we construct such a continuous mapping
U(t,p, h) : [0, +oo) x [0, l] x !Re + R, governed by
(1 -p)au(t,PJ) = hp au(tYP>w
ot
[
at
+ 2tU2(t,p, Ii)
I
P E LO> 11, l-i # 0 (4
with boundary condition
WAP, fi> = 1 (5)
so that
qt,o, 7i) = 1, u(t, 1, A) = u(t) = &,
(6)
No.2 LIAO: Homotopy Analysis Method . . . 97
where !R = (-co, +co) and ?I& = (--00, O)U(O, +co). Note that we here let fi be a nonzero real
number. Clearly, as p increases from 0 to 1, U(t,p, A) varies continuously from U(t, 0, A) = 1
to the exact solution u(t) = (1 + t2)-. This kind of variation is named deformation in
topology so that we call Eqs. (4), (5) th e zeroth-order deformation equations. Second,
assuming that U(t,p, h) is so smooth that
$1 (t h) =
akw,P, h>
,
apk
p=o
namely the kth-order deformation derivatives (k 2 l), exists, and moreover, the correspond-
ing Maclaurin series of U(t,p, h) at p = 0, say,
converges at p = 1, then, we have by (6) that
Here, ur](t, R) (k 2 1) is governed by the kth-order deformation equations
a$$ h)
at
= 2At
(7)
(9)
a$] (t h)
at
(k 2 2) (10)
under boundary condition
u(yl(O
7
A) =o
(k 2 1L (11)
which are given by differentiating the zeroth-order deformation equations (4), (5) k times
with respect to p and then setting p = 0. The above linear first-order differential equation
can be rather easily solved. So, substituting these solutions into (8), we obtain
dt) = ,\y, 5 [(-l)kt2k] &,k(h),
k=O
(12)
where
k>m
( mAi) ( k+i-l)R, l<k<m (13)
kl0
namely the approaching function.
The approaching function @m&(h) has rather general meanings. Liao [lo] rigorously
proved that @m&(h) (1 _< k 5 m) has the following properties
tLk-(l + k)m-k,
(14)
98 Communications in Nonlinear Science & Numerical Simulation Vo1.2, No.2 (May 1997)
9
m+l,k@) - hz,k(fi) =
( 1
k 1 (-?=Q(1+ fi)m--k+l,
(15)
a,$(-1) = 1.
(16)
And moreover, for finite positive integer V, it holds
,:I& %n,@) = 1, (11 + ti] < 1).
(17)
The above properties hold for complex variable ti. Furthermore, Liao [lo] rigorously proved
the so-called General Taylor Theorem which is simply recited in Appendix of this paper.
Clearly, function f(z) = l/(1 + z) has two singularities [r = i and <z = -i. So, according
to the so-called General Taylor Theorem, expression (12) is valid in the region
Itl <
\i
--
,;, 17
(-2 < h < 0).
(18)
2. Discussions and conclusions
Now, let us compare the perturbation approximation (3) with the solution (12). The
former converges only in the region It] < 1, but the latter converges in a region which is
however a function of fL. First, according to (16) and (18), in case of h. = -1, (12) gives the
same result as (3), so that (3) is just a special member of the family of the approximations
(12). It means that (12) contains the perturbation approximation (3) in logic. This kind of
continuation in logic has proved rather important in both science and mathematics. Second,
as ?% increases from -1 to 0, the convergence radius of power series (12) becomes larger and
larger. In limit h + 0, (12) is valid in the whole real axis! However, as ti decreases
from -1 to -2, the convergence radius becomes smaller and smaller. So, the perturbation
approximation (3) is neither the best nor the worst: its a special but common member
among the family of approximations (12). At last, we emphasize that the whole approach
mentioned above does not need the small parameter assumption. In other words, the validity
of the HAM is independent of whether or not there exist small parameters in the problems
considered. In form, the HAM seems able to be applied to any nonlinear problems, although
there certainly should exist some restrictions for its applications which we currently dont
know clearly.
Different from the perturbation approximation (3), (12) denotes now a famiZy of approx-
imations. The convergence radius of (12) is determined by the value of ti which is introduced
to construct the so-called zeroth-order deformation equations (4) and (5). Clearly, different
values of fi correspond to different continuous mappings, or more precisely, different homo-
topies. In fact, Eqs. (4) and (5) construct a family of homotopies U(t,p, ti) dependent on
A. Certainly, some among them are better, some are worse. This can well explain why the
convergence radius of (12) is a function of ?i. The introducing of parameter h provides us
with larger freedom to get more and even better approximations.
The foregoing example verifies the validity and great potential of the Homotopy Analysis
Method, although it certainly needs more applications and further improvements.
Appendix
General Taylor Theorem Let v be finite positive integer and h. be complex number.
No.2 LIAO: Homotopy Analysis Method . . .
99
If complex function f(z) is analytical at z = 20, then
m
f(z) = ,;y,
Cl
Pp(z - zo)l] @m,k@)
k=O .
= lim C
m-t+CC
m+ Fp(z - z,)k] fPm,k-v@)
[
k=O .
m
=
lim C
m-t-t-~
1
!yd(z - zo)t] (Pm+v,k+v@)
k=O .
holds in the region
nl
1+ti-fi ;-To <l,
( )I
(I1 + w < 11,
kEI
where am,k(h) is defined by (13) and <k (k E I) denotes the set of all singular points of
f(z). Moreover, it, becomes just the traditional Taylor theorem in case fi = -1.
General Newtonian Binomial Theorem Let a, t and A be real numbers. The equality
m
(l+t)a = 1+,;mc
cr(a:-l)(cu-2)(a-L+1)tl, 9
k!
1
m,k (h) (a # 0,1,2,3, * 1
k=l
holds in the region
-1<t< ,;, 1,
--
(-2 < Ii < O),
where em,& (fi) is defined by (13).
References
[l] Liao, S. J., The Homotopy Analysis Method and its applications in mechanics. Ph. D.
Dissertation (in English), Shanghai Jiao Tong Univ., 1992.
[2] Liao, S. J., A kind of linear invariance under homotopy and some simple applications
of it in mechanics. Bericht Nr. 520, Institut fuer Schiflbau der Universitaet Hamburg,
1992.
[3] Liao, S. J., J. Appl. Mech., 1992, 14: 1173-1191.
[4] Liao, S. J., J. Ship Research, 1992, 36:30-37.
[5] Liao, S. J., A kind of approximate solution technique which does not depend upon
small parameters: a special example. Int. J. Non-linear Mech., 1995, 30:371-380.
[S] Liao, S. J., A kind of approximate solution technique which does not depend upon
small parameters (2): an application in fluid mechanics. Int. J. Non-linear Mech., in
press.
[7] Liao, S. J., Boundary Elements XVII. Computational Mechanics Publications,
Southampton, 1995,67-74.
[S] Liao, S. J., Int. J. Numer. Methods Fluids, 1996, 23: 739-751.
[9] Liao, S. J. and A. T. Chwang, Int. J. Numer. Methods Fluids, 1996, 23: 467-483.
100 Communications in Nonlinear Science & Numerical Simulation Vo1.2, No.2 (May 1997)
[lo] Liao, S. J, Homotopy Analysis Method and its applications in mathematics, J. Basic
Sci. Eng., 1997, 5 (2): 111-125.
[ll] Liao, S. J., The common ground of all numerical and analytical technique for solving
nonlinear problems. Commun. Nonlinear Sci. Nonlinear Simul. 1996, 1 (4): 26-30.
An Approach in Modeling Cavitating Flows with
Gravity Effect1T2
Zhengtong XIE & Yousheng HE+
(Institute of Mechanics, Chinese Academy of Sciences, Beijing 100080, China)
(Shanghai Jiao Tong University, Shanghai 200030, China)
Abstract: Numerical investigations on submerged cavitating slender axisymmetric bodies
with gravity effect are conducted in this paper. The developed cavity shape and the hy-
drodynamic characteristics on cavitating axisymmetric bodies are calculated. While these
cavities originate from both the shoulder and the tail of bodies, we focus our main attention
on the cavities originated from the shoulder. The boundary element method is used to solve
the potential flow problem, and a new iteration method named Adaptive Modified Newton
Iteration is developed to determine the cavity shape.
Key words: partial cavitation, axisymmetric bodies, iteration method, gravity effect
1. Introduction
The hydrodynamic characteristics on a submerged, fully cavitating bodies pose several
difficult problems. Many factors work together to make this problem more complex. Lemon-
nier, H. & Rowe, A.lll have studied cavitating flows in which allowance was made for the
presence of an axial gravity field, but no detailed exposition was provided.
A cavitating-flow calculation method is presented, which is an attempt at considering
the gravity effect. The cavitating flow around an axisymmetrical body is in general a three-
dimensional problem. The study is limited to the case of zero angle of attack and assumes
that the symmetry axis of the body is identical to the direction of the gravity.
2. Problem Formulation
Under the assumption of incompressible, irrotational and inviscid flow, the governing
equation for the potential function @toi is given by Laplaces equation. The potential func-
tion can be subdivided into two parts, one part Gs due to the free stream, and the other
part @ due to the perturbation caused by the body and the cavity. The governing equation
for the perturbation part can be written as follows:
v29=o
(1)
The paper was received on Feb. 26, 1997
Sponsored by the National Natural Science Foundation of China, the Scientific Research Foundation of
Doctoral Discipline of National Education Committee and the National Defence Foundation of Science and
Technology Pre-research.