• 検索結果がありません。

Some concrete shapes of degenerate Stokes curves of hypergeometric differential equations with a large parameter

N/A
N/A
Protected

Academic year: 2021

シェア "Some concrete shapes of degenerate Stokes curves of hypergeometric differential equations with a large parameter"

Copied!
4
0
0

読み込み中.... (全文を見る)

全文

(1)

」:8cんooZScご.五7η.1(〃zん̀こ ノノzごひ.47(2011)5‑8 理 工 学 部 研 究 報 告 第47号 5

超 幾何微分 方程 式 の退化 ス トー クス曲線 の具体形

青 木 貴 史*反 田 美 香**

Some concrete shapes of degenerate Stokes curves of hypergeometric differential equations with a large parameter

Takashi AOKI* and Mika TANDA**

Although Stokes curves of differential equations with a large parameter are transcendental objects in general, in some degenerate cases they become real algebraic or semialgebraic sets and the defining equations are explicitly given for hypergeometric differential equations with a large parameter. Classically well-known curves, such as lemniscate and limacon appear for special values of parameters. Some explicit shapes play a role in a topological classification of Stokes curves of hypergeometric differential equations.

Key words: Hypergeometric differential equation, Stokes curves

1. Introduction

We consider the classical hypergeometric differential equation:

2 w

(1) x(1 – x)d

x2—d+ {c – (a + b + 1)x}—dw dx – abw = 0, where a, b and c are complex parameters. We introduce a large parameter q by setting a = 1 /2+ qa, b = 1/ 2+ 77$ and c = 1 + qy with complex parameters a, fi and y. Since the hypergeometric differential equation contains essentially three parameters, our new parameters are redundant. We regard 77 being real positive and large. We eliminate the first-order term in (1) by taking

(2) = x0.-Fw)(1 – x)(1.-Fixa+,g--mw as unknown functions. Then (1) becomes

We assume a # /3. Then Qo has two zeros ao, al in the complex plane. We call them the turning points of (3). We also assume that ak # 0,1 for k = 0,1, namely, a +fl– y 0, y * 0. If ao # al, the turning points are said to be simple. If ao = al, we say that the turning point is double.

A Stokes curve of (3) emanating from a turning point T (r = ao or al) is an integral curve of the direction field

(7)

AtQc dx = 0

which start on r ([2], [5]). The turning points and Stokes curves of (1) are, by definition, those of (3).

If the turning point r is double, (7) can be integrated easily and we have the following form of defining equa- tion of the Stokes curves:

Hence in some special cases, this defines a semialgeb set or an algebraic set.

raic

2. Examples of algebraic Stokes curves 2.1. Let us consider the case where a = i/2, = y = –i/2.

Then our equation becomes

平 成 つ気 杢E6日2気 目 呼]理

*理 学 科 数 学 コー スDepartment of Mathematics

**総 合 理 工 学 研 究 科 理 学 専 攻Interdisciplinary Graduate School of Science and Engineering, Department of Mathematical Sciences

We have ao = al = 1 /2 and hence 1 /2 is a double turn- ing point of this equation. The Stokes curves of (8) are defined by

(9;

(2)

超 幾 何 微 分 方 程 式 の退 化 ス トー ク ス曲 線 の 具 体 形

The left-hand side turns out to be

Im (--i4log 4x(1x)).

Hence (9) reduces to

(10) 14x(1 — x)I = 1, or, equivalently,

(11) 2(u2 + v2)2 _ (u2 _ v2) = 0.

Here we set x = u+1/2+ iv (u, v E IR). This is a lemniscate (see Fig. 1).

3. Examples of semialgebraic Stokes curves 3.1 When a = 1, = y = —1, our equation turns out to be

d2 2 (X - (16)

dx (x(x — 1))2 Qi) = 0

which has a double turning point at a0 = al = 1/2. The Stokes curves are given by

x —

(17) Imf 2x(x — 1)dx = O.

Hence

2.2. If a = i, fi = y = 2i, our equation has the form

d2 712 (x 2)2 Qi) = 0.

(12) dx2 (x(x — 1))2

and we have a() = al = 2. The Stokes curves are defined by

(13) Imf 2i2x — 2x(x — 1)dx = O.

This implies

R 4(x — 1)

e log =0

x2

that is,

(14) 4(x

x2

— 1)1

= 1.

If we set x = u + iv (u, v E IR), we have (15) 16((u — 1)2 + v2) = (u2 + v2)2.

This is a limacon (Fig. 2).

(18)

or equivalently (19)

Setting x = u

(20) arg

If v = O. we lu

Im log 4x(1 — x) = 0,

Setting x = u + iv(u, v E IR) in (19), we have (20) arg (u —

If v = 0, we have u — u2 > 0, namely, 0 < u < 1. If u = 1/2, (20) holds since v2 + 1/4 > 0. Thus the Stokes curves consist of one straight line and one segment (Fig. 3).

arg x(1 — x) = 0.

v E IR) in (19), we have

Fig. 3 Line and segment 3.2Ifa= 206=y= 4,wehave

d2 2 (x — 2)2 (21)

axe(x(x-1))2 = 0.

a0 = al = 2 is a double turning point of this equation. The Stokes curves are

x — 2 (22) I

m1)dx = O.

f2x(x

This implies

(23) Im logx—= 0.

x21

(3)

理 工 学 部 研 究 報 告 第47号

Hence we have

If v = 0, u should be greater than 1. If u2 + v2 - 2u = 0, the real part of (25) is positive if u > 0. Thus the Stokes curves coincide with the semialgebraic set

(26) {xi v = 0, u> 1} U {XI (U - 1)2 + V2 = 1, U > 0}, which consists of a half line and a circle excluded the gin (Fig. 4).

ori-

type of non-degenerate Stokes graphs. A topological type of a non-degenerate Stokes graph is characterized by the sequence (n1, n2, n3), which we call the order sequence of the Stokes graph, of numbers n3 (j = 0, 1, 2) of Stokes curves which flows into the singularity bj, where we set bo = 0, b1 = 1 and b2 = co. By the definition, we have ni + n2 + n3 = 6. We define the set Hj (j = 0, 1, 2, 3) by 1/0 = {(a,/3, Y) E C3 I 0 < Rea < Rey, 0 < Rep < Rey}

U{(a,fl, y) E C3 I Rey < Rea, < 0, Rey < Re/3 < 0},

H1 = {(a )6, Y) E C3 I 0 < Rey < Rea, 0 < Rey < Ref}

U{(a,p, y) E C3 I Rea < Rey < 0, Rep < Rey < 0}

U{(a,fi, y) E C3 I Rea < 0, Re/3 < 0, Rey > 0}

U{(a,p, y) E C3 I Rea > 0, Rep > 0, Rey < 0},

H2 = {(a,fl, E C3 I 0 < Rey < Rea, Rep < 0)

Fig. 4 Circle and half-line 4. Application

Using examples given in the preceding section, we can establish a classification theorem of Stokes graphs of hy- pergeometric differential equations with a large parame- ter. (See [1], [2] for the definition of Stokes graphs of Fuchsian differential equations of second order.) Let Ej

(j = 0, 1, 2) be the sets of the parameters (a,fi, y) defined by

E0 = [(a,fl, y) E C3 I

Y (a - )3) - (a - (fi -

x(a +13 - y) = 0), E1 = {(a,fi, y) E C3 I

Re a - Rep - Re(y - a) Re(y -13) = 0), E2 = {(a,fi, y) E C3 I

Re(a - p) Re(a + - y) Re y = 0).

The second author has obtained the following

Theorem 1. ([4], Theorem 3.1) Suppose that (a, y) is not contained in Ecs.

(i) If two distinct turning points are connected by a Stokes curve, then (a,/3, y) belongs to El.

(ii) If a Stokes curve forms a closed curve with a single turning point as the base point, then (a, 13,y) belongs to E2.

Hence each connected component of the complement of E0 U El U E2 in C3 should correspond to a topological

E C3 U{(a,p, y) E C3 U{(a,fl, E C3

Rea, < Rey < 0, Rep > 0) Rey < Re/3, Rea, < 0) Rep < Rey < 0, Rea, > 0),

H3 = {(a , 13, y) E C3 I

Rea Rep Re(y - a) Re(y (3) < 0).

Then the set C3 - (E1 U ((a,/3, y) I Re y = OD can be expressed by a disjoint union of four components:

H0UH1UH2UH3.

The following theorem is announced in [4] as a claim obtained by observation through numerical experiments.

Now we can give a proof by using our examples.

Theorem 2. Let H.; be as above and let us suppose (a,fl, y) is not contained in E0 U E2. Let n = (ni, n2, n3) be the order sequence of the Stokes graph of (3).

(i) If (a, 13, y) E Ho, then n = (4, 1, 1).

(ii) If (a,/3, E Ill, then n= (1, 4,1).

(iii) If (a,/3, y) E H2, then n. (1, 1, 4).

(iv) If (a,fl, y) E H3, then n = (2, 2, 2).

Outline of the proof. As is pointed out in [4], if (a, /3, y) is a point in C3 - E0 U Ei U E2, then the Stokes geometry is non-degenerate and hence any small perturbation of the parameters a, /3, y does not make degeneracy of the Stokes geometry. Let (ao,fio, Yo) be a point in E2 - E0 U El. If we take a sufficiently small neighborhood SI of (ao,fio, yo) in C3, the order sequence n = (no, ni, n2) of the Stokes graph is the same for every (ail, y) E S2 - E2. Hence each connected component of H3 corresponds to a topo- logical type of Stokes graphs. Thus we look at the bound- ary between Hj and Hk for j # k or that of connected components of Hi. For example, the triplet of parameters

y) = (2, 4, 4) E E0 given in 3.2 belongs to the bound- ary between 1/0 and H3. Small perturbations of (a,/3, y)

(4)

超 幾 何 微 分 方 程 式 の退 化 ス トー ク ス曲 線 の 具 体 形

make non-degenerate Stokes geometries and the order se- quences n of them can be found by using arguments given in § 4 of [3]. That is, (a,i3,y) = (2, 4 + E, 4) E. H3 with a small e > 0 yields n = (2, 2, 2) (Fig. 5). On the other hand, (2, 4 — e, 4) E H0 and n = (4,1, 1) (Fig. 6).

[5] Voros, A., The return complex WKB method, 39(1983), 211-338.

of quartic oscillator, The Ann. Inst.Henri Poincare,

rig. b n = (4, 1, 1)

Hence in the connected components of H3 (resp. H0) con- taining (2, 4+e, 4) (resp. (2, 4—e, 4)) we have n = (2, 2, 2) (resp. n = (4, 1, 1)). Other parts of boundary can be dis- cussed similarly. A complete proof of this theorem will be given in our forthcoming paper.

References

[1] Aoki, T. and Iizuka, T., Classification of Stokes graphs of second order Fuchsian differential equa-

tions of genus two, Publ. RIMS, Kyoto Univ.„ 43

(2007), 241-276.

[2] Kawai, T. and Takei, Y., Algebraic Analysis of Singu- lar Perturbation Theory, Translation of Mathematical

Monographs, vol. 227, AMS, 2005.

[3] Sasaki, S., On the classification of Stokes graphs for second order Fuchsian equations, Master's Thesis,

RIMS, Kyoto University, 2010.

[4] Tanda, M., Exact WKB analysis differential equations, to appear in

Bessatsu.

of hypergeometric RIMS Kokyfiroku

参照

関連したドキュメント

Functional and neutral differential equations play an important role in many applications and have a long and rich history with a substantial contribution of Hungarian

Trujillo; Fractional integrals and derivatives and differential equations of fractional order in weighted spaces of continuous functions,

[9, 28, 38] established a Hodge- type decomposition of variable exponent Lebesgue spaces of Clifford-valued func- tions with applications to the Stokes equations, the

This article is devoted to establishing the global existence and uniqueness of a mild solution of the modified Navier-Stokes equations with a small initial data in the critical

Using the theory of nonlinear semigroups, we prove existence results for strong and weak solutions1. Examples are

The existence and uniqueness of adapted solutions to the backward stochastic Navier-Stokes equation with artificial compressibility in two-dimensional bounded domains are shown

The numerical tests that we have done showed significant gain in computing time of this method in comparison with the usual Galerkin method and kept a comparable precision to this

We show the uniqueness of particle paths of a velocity field, which solves the compressible isentropic Navier-Stokes equations in the half-space R 3 + with the Navier