Asymptotics beyond All Orders pp 15-28 | Cite as

# Computation of Transcendental Effects in Growth Problems: Linear Solvability Conditions and Nonlinear Methods-The Example of the Geometric Model

## Abstract

It is now well understood that in many moving interface problems, surface tension plays a singular role^{(1)}. The basic reason is that surface tension multiplies a term that contains the highest number of derivatives. This realization^{(2)} has made possible the analysis of several problems, including the velocity selection for a parabolic needle crystal moving in an undercooled melt^{(3)} and the shape selection of viscous fingers moving in a linear Hele-Shaw cell^{(4)}. The first general analytic method for treating this kind of singular perturbation problems was found by J. Langer^{(5)} in a simple case. He proposed to linearize the equations around a zero-surface tension solution while keeping in the linear equation terms with the highest number of derivatives although they were formally of higher order in the expansion parameter. Exponentially small terms could then be obtained from a linear solvability condition. It was however clear since its very proposal that Langer’s method could not be completely correct since the magnitude of the exponentially small term was about half the result obtained by solving the equation numerically^{(5)}. A more satisfactory way of handling the problem was described by M. Kruskal and H. Segur^{(6)} who generalized to nonlinear ODE a well-known idea^{(7)} in the linear case. It consists in extending the equation into the complex plane and analyzing it carefully in the neighborhhood of a singularity of the terms of the regular perturbation expansion. Contrary to Langer’s treatment it is not entirely analytic but reduces to solving a parameter free nonlinear problem numerically. Therefore the method of ref. (5) has continued to be utilized under one form or another by some authors, in spite of its somewhat unclear basis.

## Keywords

Asymptotic Series Singular Perturbation Problem Inhomogenous Term General Analytic Method Viscous Finger## Preview

Unable to display preview. Download preview PDF.

## References

- (1).Reviews can be found in D.A. Kessler, J. Koplik and H. Levine 1988, Adv. Phys.
**37**, 255;ADSCrossRefGoogle Scholar - (1a).Pelcé P.
*Dynamics of curved fronts*(Academic, Orlando, Fl, 1988)MATHGoogle Scholar - (2)a).McLean J.W., P.G. Saffman 1981, J.Fluid Mech.
**102**, 445ADSCrossRefGoogle Scholar - (2)b).Vanden-Broeck J.M. 1983, Phys.Fluids
**26**, 2033MathSciNetADSCrossRefGoogle Scholar - (2)c).Ben-Jacob E., N.D. Goldenfeld, B.G. Kotliar and J.S. Langer 1983 Phys.Rev.Lett.
**53**, 2110ADSCrossRefGoogle Scholar - (2)d).Kessler D. A., J. Koplik, H. Levine (1985) Phys. Rev.
**A31**, 1712ADSGoogle Scholar - (2)e).Pelcé P. and Y. Pomeau 1986, Stud. App. Math.
**74**, 245MATHGoogle Scholar - (3).Benamar M., Y. Pomeau 1986, Europhys. Lett.
**2**, 307ADSCrossRefGoogle Scholar - (3a).Barbieri A., D. Hong and J.S. Langer 1986, Phys.Rev.
**A35**, 1802ADSGoogle Scholar - (4)a).Shraiman B.I. 1986, Phys.Rev.Lett.
**56**, 2028;ADSCrossRefGoogle Scholar - (4)b).Hong D.C., J.S. Langer 1986 Phys.Rev.Lett.
**56**, 2032ADSCrossRefGoogle Scholar - (4)c).Combescot R., T. Dombre, V. Hakim, Y. Pomeau, A. Pumir 1986, Phys.Rev.Lett.
**56**, 2036ADSCrossRefGoogle Scholar - (5).Langer J.S. 1986, Phys.Rev.
**A33**, 435ADSGoogle Scholar - (6).M. Kruskal and H. Segur 1985, Aero. Res. Ass. Tech. Memo. 85–25Google Scholar
- (7).Pokrovsky V.L. and I.M. Khalatnikov 1961, Sov. Phys. J.E.T.P.
**13**, 1207Google Scholar - (7a).Meyer R.E. 1980 SIAM Review
**22**, 213MathSciNetMATHCrossRefGoogle Scholar - (8).Brower R., D.A. Kessler, J. Koplik, H. Levine 1984, Phys.Rev.
**A29**, 1335ADSGoogle Scholar - (9).Combescot R., V. Hakim, T. Dombre, Y. Pomeau, A. Pumir 1988, Phys.Rev.
**A37**, 1270MathSciNetADSGoogle Scholar - (10).Our exposition of Langer’s method differs slightly from the original one and follows closely Dombre T., V. Hakim, Y. Pomeau 1986, C. R. Acad. Sci. (Paris)
**302**, 803MATHGoogle Scholar - (11).This result was first obtained by a different method in Daschen R., D.A. Kessler, H. Levine, R. Savit 1986, Physica
**21D**371ADSGoogle Scholar - (12).Borel E., 1928
*Leçons sur les séries divergentes*(Gauthier-Villars, Paris)MATHGoogle Scholar - (13).This is in fact true only if B(x) has no other singularity inside the curve of equation (in polar coordinates) ρ
^{1/2}sin(θ/2)= -1 in the lower half plane. This is more restrictive that the absence of singularity inside the unit circle (ρ =1) that we have obtained and one would need a finer analysis of the asymptotic series to prove it. Nevertheless we will suppose that it is the caseGoogle Scholar