Everand Logo

Welcome to Everand!

  • Discover millions of ebooks, audiobooks, and so much more with a free trial

    Only $11.99/month after trial. Cancel anytime.

    Singular Integral Equations: Boundary Problems of Function Theory and Their Application to Mathematical Physics
    Singular Integral Equations: Boundary Problems of Function Theory and Their Application to Mathematical Physics
    Singular Integral Equations: Boundary Problems of Function Theory and Their Application to Mathematical Physics
    Ebook993 pages8 hours

    Singular Integral Equations: Boundary Problems of Function Theory and Their Application to Mathematical Physics

    By N. I. Muskhelishvili

    3/5

    ()

    Read preview

    About this ebook

    Singular integral equations play important roles in physics and theoretical mechanics, particularly in the areas of elasticity, aerodynamics, and unsteady aerofoil theory. They are highly effective in solving boundary problems occurring in the theory of functions of a complex variable, potential theory, the theory of elasticity, and the theory of fluid mechanics.
    This high-level treatment by a noted mathematician considers one-dimensional singular integral equations involving Cauchy principal values. Its coverage includes such topics as the Hölder condition, Hilbert and Riemann-Hilbert problems, the Dirichlet problem, inversion formulas for arcs, and many other areas. Intended for graduate students, applied and pure mathematicians, engineers, physicists, and researchers in a variety of scientific and industrial fields, this text is accessible to students acquainted with the basic theory of functions of a complex variable and the theory of Fredholm integral equations.
    LanguageEnglish
    PublisherDover Publications
    Release dateFeb 19, 2013
    ISBN9780486145068
    Singular Integral Equations: Boundary Problems of Function Theory and Their Application to Mathematical Physics

    Related to Singular Integral Equations

    Titles in the series (100)

    View More
    View More

    Related ebooks

    Mathematics For You

    View More
    View More

    Related podcast episodes

    Related articles

    • Article

      The Sacred, Spherical Cows of Physics: Theoretical physics milks symmetry to power its newest tool.

      May 1, 2014

      Early in their training, many physics students come across the idea of spherical cows. Cows in the real world—even at their most plump and well-fed—are hardly spherical, and this makes it tricky to calculate things like, say, how their volume or surf

      11 min read

    • Article

      Physicists Attack Math’s $1,000,000 Question

      Apr 4, 2017

      Physicists are attempting to map the distribution of the prime numbers to the energy levels of a particular quantum system.

      4 min read

    • Article

      Cloudy With a Chance of War

      Mar 7, 2019

      The burial detail, which had come for the corpses in the pigpen, was surprised. The “dead” were getting up and speaking English. Qu’est-ce que c’est? Ah, they were an ambulance crew. British volunteers, in the trenches with the French Army on the Wes

      14 min read

    • Article

      Cloudy With a Chance of War: His weather forecasts changed the world. Could his predictions of war?

      Jul 31, 2014

      The burial detail, which had come for the corpses in the pigpen, was surprised. The “dead” were getting up and speaking English. Qu’est-ce que c’est? Ah, they were an ambulance crew. British volunteers, in the trenches with the French Army on the Wes

      14 min read

    • Article

      A Supermassive Lens On The Constants Of Nature

      Nov 25, 2020

      The 2020 Nobel Prize in Physics went to three researchers who confirmed that Einstein’s general relativity predicts black holes, and established that the center of our own galaxy houses a supermassive black hole with the equivalent of 4 million suns

      9 min read

    • Article

      A Universe Made of Tiny, Random Chunks: The space-time that makes up our universe is inherently uncertain.

      Jun 6, 2013

      One of science’s most crucial yet underappreciated achievements is the description of the physical universe using mathematics—in particular, using continuous, smooth mathematical functions, like how a sine wave describes both light and sound. This is

      6 min read

    • Article

      The Scientific Problem That Must Be Experienced: To understand turbulence we need the intuitive perspective of art.

      Jul 3, 2014

      When the German physicist Arnold Sommerfeld assigned his most brilliant student a subject for his doctoral thesis in 1923, he admitted that “I would not have proposed a topic of this difficulty to any of my other pupils.” Those others included such g

      16 min read

    • Article

      A Short History Of atoms

      Jun 3, 2020

      A Short History Of atoms
      5 min read

    • Article

      The First City on Mars

      Feb 16, 2023

      OFF-WORLD IDEAS “There will be towns and then cities on Earth's Moon and on Mars sooner than you think,” predicts Hollander in this fantastic book on engineering a future for humans on Mars. And if that's true, he argues, there is much to consider to

      1 min read

    • Article

      The First City on Mars

      Feb 16, 2023

      OFF-WORLD IDEAS “There will be towns and then cities on Earth's Moon and on Mars sooner than you think,” predicts Hollander in this fantastic book on engineering a future for humans on Mars. And if that's true, he argues, there is much to consider to

      1 min read

    • Article

      The Math That Takes Newton Into the Quantum World

      Feb 28, 2019

      In my 50s, too old to become a real expert, I have finally fallen in love with algebraic geometry. As the name suggests, this is the study of geometry using algebra. Around 1637, René Descartes laid the groundwork for this subject by taking a plane,

      9 min read

    • Article

      How to Weigh the World

      Mar 6, 2017

      How to Weigh the World
      4 min read

    • Article

      From Newton To String Theory

      Jun 20, 2019

      In his work Principia, Isaac Newton ties together Galileo's work on gravity, Kepler's laws of planetary motion and the phenomenon of tides in one law covering gravity throughout the universe. James Clerk Maxwell discovers that electricity, magnetism

      1 min read

    • Article

      Monster Science

      Dec 6, 2018

      Monster Science
      11 min read

    Related categories

    Reviews for Singular Integral Equations

    3/5

    1 rating0 reviews

    What did you think?

    Tap to rate

    Review must be at least 10 words

      Book preview

      Singular Integral Equations - N. I. Muskhelishvili

      INDEX

      PREFACE

      This book is intended for an extensive field of readers, in particular for those interested in applications to the theory of elasticity, hydromechanics and other branches of mathematical physics. The book is accessible to those acquainted with the basic theory of functions of a complex variable and the theory of Fredholm integral equations. To facilitate the reading of the book, theorems, the method of proof of which is not of essential independent interest, have been printed in italics, so that the proofs may be omitted without affecting an understanding of the nature of the matter. In addition, wherever possible, the Parts and their chapters devoted to different applications have been made independent of one another. It is hoped that the methods studied in this book may be effectively employed in the solution of many problems of an applied character. Some simple applications to potential theory, the theory of elasticity and hydromechanics are given in this book.

      The idea of writing this book resulted from the Author’s lectures in a seminar at the Mathematical Institute, Tiflis, in 1940—1942. Under the influence of a series of results obtained by members of the seminar (chiefly due to the excellent work of 1. N. Vekua) the range of problems which the Author proposed to examine was essentially altered ; the Author may state with much satisfaction that a large proportion of the contents of this book must be considered as the result of the collective work of the young members of the Tiflis Mathematical Institute of the Academy of Sciences of the Georgian S.S.R. together with I. N. Vekua and the Author himself.

      Tiflis, Autumn 1944

      N. I. MUSKHELISHVILI

      In adding the last corrections to the book the Author desires to use this opportunity to express his deep gratitude to the publishing house which always willingly complied with the Author’s suggestions. The difficult and responsible task of proof-reading was considerably eased for the Author by the extraordinary kindness of L. I. Bokshitski on the staff of the press, to whom he extends his sincere gratitude just as to all the other members of the staff of the 16th Printing Establishment of the State publishers for their prompt and efficient work.

      Moscow, Autumn 1945

      N. I. M.

      INTRODUCTION

      In recent years the theory of singular integral equations has assumed increasing importance in applied problems.

      In this book only one-dimensional (i.e.. where the range of integration is one-dimensional, i.e., a line) singular equations involving Cauchy principal values will be examined, since the theory of multi-dimensional equations of corresponding form is still far from completion. (Some references to the literature dealing with the latter are given in § 59).

      The fundamentals of the theory of one-dimensional singular integral equations of the type described were included in the work of Poincaré and Hilbert, almost directly after the development of the classical theory of integral equations by Fredholm. However, the theory of singular integral equations did not receive the attention of mathematicians for some time. On the other hand many problems of an applied character naturally reduced to singular equations, e.g. problems of the theory of elasticity, etc.; thus often in practice these equations were arrived at by ordinary methods and this did not always lead to satisfactory results.

      However, during recent years, the theory of one-dimensional singular integral equations has advanced considerably and it can now be presented in a finished form.

      This theory appears to be particularly simple and effective, if the solution of a boundary problem of the theory of functions of a complex variable, to be called the Hilbert problem, is considered. Therefore the theory of singular equations is here closely linked with the above boundary problem. The solution of the latter will be used for the development of the theory of singular equations; afterwards this theory will be applied to the solution of other more complicated boundary problems, in particular, to problems encountered in potential theory, the theory of elasticity and in hydromechanics.

      Having in mind the implications for different problems of mathematical physics, some restrictions will be imposed upon the unknown and the given functions appearing in the integral equations under consideration or in the boundary conditions of the problems considered, which will largely simplify the investigation, but not affect the final theory.

      The fundamental tools of the investigation are Cauchy integrals, the elementary theory of which is given in Part I together with a number of direct simpler applications.

      As regards the contents of the remaining Parts nothing will be said here, since a preliminary idea can be gained by means of the sufficiently detailed list of contents at the beginning of the book.

      PART I

      FUNDAMENTAL PROPERTIES OF CAUCHY INTEGRALS

      CHAPTER 1

      THE HÖLDER CONDITION

      All functions considered will be defined on smooth lines and satisfy the Holder condition. In this chapter some properties of such functions will be given.

      § 1. Smooth and piecewise smooth lines.

      In the sequel only lines lying in one and the same plane will be considered. A coordinate system Oxy in this plane will always be a right-handed cartesian system.

      Unless otherwise stated, lines are always assumed to be simple, i.e., not to intersect.

      Curves will be called smooth arcs, when they can be represented in the form

      (1.1)

      where sa, sb are finite constants, x(s) and y(s) functions, continuous in the interval of definition, with the following, properties:

      1°. They have continuous first derivatives x′(s), y′(s) within the interval (sa, sb), including the end points, and these derivatives are never simultaneously zero; x′(s), y′(s) at the ends of the interval are to be interpreted as x′(sa + 0,) y′(sa + 0) and x′(sb − 0), y′(sb − 0) respectively.

      2°. The equalities x(s1) = x(s2), y(s1) = y(s2) are incompatible for

      sa ≦ s1, s2 ≦ sb, s1, ≠ s2.

      The first condition shows that the arc L is smooth, i.e., that the direction of its tangent varies continuously with the position of its point of contact (see later). The second condition states that the arc is simple.

      The points a and b, corresponding to the values sa and sb, are the ends of the arc. In agreement with (1.1) the ends a, b belong to the arc L; when it is necessary to stress this fact, L will be called a closed arc. When the end points do not belong to the arc, i.e., when sa < s < sb in (1.1), the arc will be called open.

      For each smooth arc considered a definite positive direction is chosen, namely that direction which corresponds to an increase of the parameter s. The arc with end points a, b will always be denoted by ab, where the order of letters is to indicate that the positive direction is from a to b.

      The smooth arc L is obviously rectifiable; therefore it is possible to take as parameter s the length of arc from any fixed point of L, where the sign of s depends on the direction of displacement. This is done in all subsequent calculations. Hence

      (1.2)

      In this case the parameter s is called the arc coordinate of a point on the arc L. The point which corresponds to s will be denoted by t(s) or simply t, and points corresponding to coordinates si will be denoted by t(si) or simply ti.

      By the tangent to an arc L is always understood the positive tangent, i.e., the tangent in the increasing direction of s. If θ is the angle between the tangent at the point t and the axis Ox, measured from the latter counter-clockwise, then

      (1.8)

      These formulae show that the angle between the tangent and any fixed direction is a continuous function of s. Conversely, it is clear that, if θ varies continuously with s, x′(s) and y′(s) are continuous functions of s.

      Curves will be termed smooth contours L, if they differ from smooth arcs only in that, in condition 2°, the equalities x(s1) = x(s2), y(s1) = y(s2) are incompatible for sa < s1, s2 < sb, s1 ≠ s2, but x(sb) = x(sa) = y(sa) and x′(sb − 0) = x′(sa + 0), y′(sb − 0) = y′(sa + 0). The first of these equalities indicates that the curve L is closed, and the second that the direction of the tangent changes continuously on passing the point of contact corresponding to the values sa, sb of the arc coordinate. This point will sometimes be called the point of discontinuity of the arc-coordinate; it does not differ essentially from any other point of L, since it only depends on the choice of the parametric representation and clearly can be any point of the contour L.

      In the following, without stating it explicitly in many cases, if any arc ab is under consideration consisting of portion of a contour L, the point of discontinuity of the arc coordinate is assumed moved on L so that it is outside or at one of the endpoints of the arc ab.

      The union of a finite number of non-intersecting smooth arcs or contours will be called a smooth line L. Thus, by definition, a line may consist of several disconnected parts, while arcs or contours consist of one continuous portion. The main text of the book will refer to smooth lines only. However, many of the results obtained below may be applied directly or almost directly to piece-wise smooth lines. (Some of the necessary modifications are given in Appendices 1 and 2.)

      Arcs or contours will be called piecewise smooth, if they are continuous, simple, and consist of a finite number of separate smooth curves with contiguous ends. By a piece-wise smooth line is to be understood the union of a finite number of piece-wise smooth arcs (and contours) which do not intersect one another.

      § 2. Some properties of smooth lines.

      In the sequel, the following easily proved properties of smooth lines are needed: (For the proofs see Appendix 1)

      Let L be a given smooth line, α0 any arbitrary acute angle (i.e., 0 < α) and let R0 = R0(α0) be a positive number depending on α0, but not on the position of the point t on L, such that:

      1 °. The part of L, lying within a circle Γ with radius R ≦ R0 with its centre at any point t on L, consists of a single arc ab. (If L consists only of a closed contour, the ends a, b of this arc lie always on the circumference of Γ. If an arc is part of L, one or both of its end points may be inside Γ.)

      2°. The non-obtuse angle between the tangents at any two points of the arc ab does not exceed α0.

      From the above the following properties of the arc ab result immediately:

      3°. The non-obtuse angle between the chord joining any two points of the arc ab and the tangent at an arbitrary point of this arc does not exceed α0.

      In fact, there is always a point on the arc ab at which the corresponding tangent is parallel to the chord, so that the statement is reduced to 2°.

      4°. Let β0 be any angle satisfying the condition α0 < β, also let Δa, Δb be two parallel lines through a and b which make a non-obtuse angle β ≧ β0with the tangent at any point t on ab. Then every line Δ, parallel to the straight lines Δa, Δb and lying between them, cuts ab at one point only.

      That Δ cuts ab in at least one point is obvious; that it does not intersect ab at any other point follows from 8° and from the condition β ≧ β0 > α0.

      5°. Let Δ be a straight line through any point t on ab, making a non-obtuse angle not less than β0 > α0 with the tangent at t, and let t′ be any other point on ab. Then the non-obtuse angle between Δ and the chord, connecting t and t′, is not less than ω0 = β0 − α0 > 0 ¹).

      For the given angle α0, R0 = R0(α0) is called the standard radius, the circle Γ0 of radius R0. the standard circle, the arc ab, cut from the line L by the circle Γ0 drawn at any point on it, the standard arc.

      Let L = ab be a standard arc, t0 a fixed point on L (which may coincide with a or b), t a variable point on L and r the length of the chord joining t and t0. Then

      (2.1)

      where s is the arc coordinate of the point t and a the acute angle between the chord tt0 and the tangent at t, the upper sign referring to the part t0b and the lower sign to the part at, it follows from (2.1) that r is a monotonic function of s over each of the parts at0, t0b (decreasing on the former, increasing on the latter part of ab), and consequently that the position of t on each of these parts is uniquely defined for any given r.

      A frequently used inequality follows from (2.1), namely

      (2.2)

      where K is a positive constant independent of the position of t0 on ab.

      Consider one of the arcs at0, t0b, integrate both sides of (2.1) between the limits s1, s2 and apply the mean value theorem; then for any pair of points t1, t2 on at0 or t0b

      (2.8)

      where k0 is a constant and r1 r2 are the distances of t1, t2 from t0.

      Taking t1 as t0 gives

      (2.4)

      where r12 is the distance between t1 and t2 and σ12 = | s1 − s2 | is the length of that part of L which lies between t1 and t2.

      The relation (2.4) is easily seen to remain valid in the case when L is an arbitrary arc or contour, provided that in the case of a contour the part of shorter length is taken as the part included between t1 and t2.

      § 3. The Hölder Condition (H condition).

      Let there be given on the arc L a function of position ϕ(t) (which is in general complex). In future let t denote both the point t(x, y) and the corresponding complex number t = x + iy.

      The function ϕ(t) will be said to satisfy a Holder condition (O. Hölder) on L, if for any two points t1, t2 of L

      (8.1)

      where A and µ are positive constants. A is called the Hölder constant and µ the Holder index.

      A function which satisfies a Holder condition will be said to obey the H condition or, when it is necessary to specify the index µ, the H(µ) condition. The value of the constant A is generally of no interest.

      A function satisfying the H condition on the arc L is clearly continuous on L. Further, if the function ϕ(t) satisfies the H(µ) condition with some constant A for any pair of points t1, t2 for which the distance r12 between them does not exceed some positive constant δ, then it will satisfy the H(µ) condition over the whole arc L.

      If (8.1) holds when r12 ≦ δ, then for the entire arc L

      where A′ is the larger of the numbers A , and M is the upper limit of |ϕ(t)| on L.

      From (2.4) the H(µ) condition is equivalent to

      (8.2)

      where σ12 = |s2 − s1| is the length of that part of the arc L included between t1 and t2. (In case L is a contour take the shorter of the two parts into which L is divided).

      Therefore, because of (2.8), the H(µ) condition is equivalent to the condition that on any standard arc ab belonging to L

      (8.8)

      where r1 = |t1 − a|, r2 = |t2 − a|.

      In the inequalities (3.2), (3.3) as in (3.1), A is some positive constant; in all these three inequalities A may be considered as having the same value, for if necessary A may be changed to a higher value in some of the inequalities.

      Further note that for μ , and hence ϕ = const. Since this case is of no interest, it will always be assumed that

      0 < μ ≦ 1.

      Clearly, if ϕ(t) satisfies the H(μ) condition, it satisfies the condition H(v) for all ν ≦ μ.

      § 4. Generalization to the case of several variables.

      Naturally the concept of the Holder condition can be generalized to the case of functions of several variables. For example, the function ϕ(u, v), where u and v lie in certain intervals, will be said to satisfy the H(μ, ν) condition with 0 < μ ≦ 1, 0 < ν ≦ 1, if for any given pair of values (u1, v1), (u2, v2)

      (4.1)

      where A and B are constants.

      It is clear that, if ϕ(u, v) satisfies the H(μ, v) condition, then it also satisfies a condition denoted by H(λ):

      (4.2)

      where λ is the lesser of the numbers μ, ν and C is a constant. Clearly, if ϕ(u, v) satisfies the condition H(μ, ν), then it satisfies the condition H(μ) for the variable u uniformly with respect to v, and the condition H(v) for v uniformly with respect to u. (This shows that the Holder constants in the conditions H(μ) and H(v) may be chosen independently, respectively, of v and u.) That the converse is also true follows from the inequality

      |ϕ(u2, u2)−ϕ(u1, v1)|≦|ϕ(u2, u2)−ϕ(u1, v2)|+|ϕ(u1, v2)−ϕ(u1, v1)|.

      § 5. Two auxiliary inequalities.

      The two following well known inequalities for any positive numbers σ1 and σ2, 0 ≦ μ ≦ 1 will be needed:

      (5.1)

      (5.2)

      In proving these inequalities it is possible, without loss of generality, to suppose σ1 ≧ σgives the above inequalities the form:

      These latter inequalities follow in a completely elementary manner by finding the maximum of the functions of σ on the left.

      § 6. Sufficient conditions for the H condition to be satisfied.

      In this section some of the simplest criteria for fulfilment of the H condition will be given. Throughout this section L is taken to mean a smooth arc or contour.

      1°. Let the point t0 divide the arc L = ab into two parts at0 and t0b. If a function ϕ(t) is continuous on L and satisfies the H(μ) condition on both parts at0 and t0b separately, then it satisfies this condition for the whole of the arc L.

      In fact, let t1 and t2 be two points on the arc L. It t1 and t2 lie on the same side of t0, then the inequality (8.2) is satisfied from the data. If, however, t1 and t2 lie on different sides of t0, then, denoting by σ1 and σ2 the lengths of the arcs t1t0 and t0t2 and using (5.1),

      which proves the statement.

      2°. If ϕ(t) and ψ(t) satisfy on L the H(μ) and H(v) conditions respectively, then the functions ϕ(t) + ψ(t) and ϕ(t). ψ(t) satisfy the H(λ) condition on L, where λ is the smaller of the numbers μ, ν.

      For example, for the product ϕ(t). ψ(t) the proof is as follows:

      where M and N are the upper bounds of | ϕ(t) | and | ψ(t) |on L. Hence the result follows.

      3°. If ϕ(t) satisfies the H(μ) condition on L and ϕ(t) ≠ 0 everywhere on Lalso satisfies the H(μ) condition.

      The proof is obvious.

      4°. Let t and t0 be variable and fixed points respectively on L.

      The function of t

      rμ = |t−t0|μ, 0 < μ ≦ 1

      satisfies the H(μ) condition on L, because by (5.2)

      |r2μ−r1μ| ≦ |r2−r1|μ.

      A similar result is obtained for t fixed and t0 variable. Hence the function | t−t0 |μ of the two variables t and to satisfies the H(μ) condition on L.

      5°. Let ϕ(t) satisfy the H(μ) condition on L, and let 0 ≦ λ < μ ≦1. Then the function of t

      where t0 is a fixed point on L, satisfies the H (μ − λ) condition on L.

      Without loss of generality it is clearly possible to assume that t lies on a standard arc ab, cut from L by a standard circle with centre at t0; also, from 1°, this may be restricted to the case in which t lies, for example, on the segment t0b. The position t on t0b is determined by r = |t−t0| | (cf. § 2); sometimes ϕ(r) and ψ(r) will be written for ϕ(t) and ψ(t). Taking h > 0 (which does not affect the generality) and writing ϕ(t)−ϕ(t0) = ω(r), one obtains

      Since

      |ω(r + h)−ω(r)| ≦ Ahμ, |ω(r)| = |ϕ(t)−ϕ(t0)| ≦ Arμ,

      it follows that

      |ψ(r+h)−ψ(r)| ≦ ∆1 + ∆2,

      where

      Further

      so that ∆1 satisfies the required condition.

      For ∆2 consider the two possible cases r ≦ h and r > h. In the first case (r ≦ h), using the inequality

      (cf. 4°), it is seen that

      while in the second case (r > h), using the inequality ²)

      it is seen that

      and thus the statement is verified.

      6°. It is clear that the above inequalities do not depend on the position of t0 on L; in addition, it is possible to interchange the parts played by t and t0. Therefore the function of the two variables t and t0

      satisfies the H(μ − λ) condition on L, if ϕ(t) satisfies the H(μ) condition there and if 0 ≦ λ < μ ≦ 1.

      7°. Now consider the function ϕ(t, τ), where t lies on L and τ is a parameter varying in some region T. Let ϕ(t, τ) satisfy the condition H(μ) (with respect to t and to τ), where t lies on L and τ in T (cf. § 4). It will be shown that the function of three variables

      (where t0 as well as t lie on L) satisfies the H(μ − λ) condition for all three variables.

      With respect to t0 and t this has been proved already. In order to prove the statement for τ, put

      For | t − t0 | ≦ | h |,the first expression for ∆ gives

      | Δ| ≦ 2A |t− t0 |μ−λ ≦ 2A |h|μ−λ

      For |t − t0 |≧ |h| , the second expression gives

      and the statement is proved. A direct consequence of the above is that, if the function ϕ(t, t0) of two points on L satisfies the H(μ) condition, then the function

      satisfies the H (μ − λ) condition.

      8°. Finally, the following obvious property should be noted.

      For simplicity suppose u(s) is a real function which satisfies the H(μ) condition over some interval s1 ≦ s ≦ s2; let ƒ(u) be a function (in general complex) defined for values u = u(s) with s1 ≦ s ≦ s2 and having a bounded derivative ƒ′(u). Then ƒ(u) satisfies the H(μ) condition.

      § 7. Sufficient conditions for the H condition to be satisfied (continued).

      The following frequently used proposition will now be proved. Let ϕ(t) satisfy the H(μ) condition on the smooth arc or contour L and let ω(t) be a bounded function on L having a derivative ¹) with respect to t, except possibly at t = t0, such that

      (7.1)

      where C is a constant and to is some fixed point on L. Then the function of t

      ψ(t) = [ϕ(t) − ϕ(t0)] ω(t)

      also satisfies the H(μ) condition on L.

      (t1, t points on L, where s is the arc coordinate.

      It may be supposed, without loss of generality, that t lies on a standard arc having the point to as one of its ends. Then the condition (7.1) is equivalent to

      (7.2)

      where C0 is a constant and s, s0 are the arc coordinates corresponding to t and t0. Further, it may be assumed that ω(t) takes only real values (otherwise the reasoning may be applied separately to the real and imaginary parts).

      Replacing ϕ(t), ψ(t), ω(t) by ϕ(s), ψ(s), ω(s), one has

      ψ(s+h)−ψ(s) = [ϕ(s+h)−ϕ(s0)]ω(s+h)−[ϕ(s)−ϕ(s0)]ω(s)= = [ϕ(s+h )−(s)]ω(s+h)+[ϕ(s)−ϕ(s0)][ω(s+h)−ω(s)).

      The suppositions s − s0 ≧ 0, h ≧ 0 do not affect the generality. Consequently, from the boundedness of ω(s), the first term in the last line does not exceed in magnitude C1hμ, where C1 is a constant. The same is obviously true for the second term, provided s − s0 ≦ h, while for s − s0 ≧ h and 0 < θ < 1

      and the proposition is proved.

      Some simple applications of this result will now be given:

      1°. Let t be a variable, t0 a fixed point on L. The function of t

      ψ(t) = |t − t0|μ log |t − t0|, 0 < μ ≦ 1

      satisfies the H(μ − ε) condition on L, where e is any positive number less than μ.

      In fact, the last result may be applied to ψ(t) by putting ϕ(t) = |t − t0|μ−8 and ω(t) = |t − t0 |⁸ log |t − t0|. It is clear that t and t0 may interchange roles and that |t − t0|μ log |t − t0| satisfies the H(μ − ε) condition for both variables t and t0.

      2°. Let t be a variable, t0 a fixed point on L. Denote by θ(t0, t) = arg (t − to) makes with any fixed direction (arg z stands for the argument of the complex number z). This angle is defined uniquely apart from an additive term of the form 2nπ, where n is an integer. Suppose that θ(t0, t) varies continuously as t moves on L without passing through t0. If t0 is not an end of L, this angle must change discontinuously by an amount (2n − 1)π, n an integer, when t passes through t0.

      It will be shown that the function ω(t) = ϑ(t0, t) satisfies the condition (7.1). To prove this it is obviously sufficient to consider only the neighbourhood of the point t0. Take as the axis Ox the tangent to L at t0 and let the origin of the coordinate system be t0. Let x = x(s), y = y(s) be the parametric representation of L and measure the arc coordinate from t0 = 0. Then on either side in the neighbourhood of the point t0

      But from the choice of the coordinate system: x(0) = y(0) = 0, x′(0) = 1, y′(0) = 0, and hence

      x(s) = s • x′(θ1s), y(s) = s • y′(θ2s), 0 < θ1, θ2 < 1.

      When | s | is sufficiently small x′(θ1s) > k, where k is a positive constant. Hence it is seen that

      and the proposition is proved.

      Any function ƒ(ϑ) satisfies an analogous inequality, provided dƒ/dϑ is bounded.

      As |t − t0lμ satisfies the H(μ) condition for 0 < μ ≦ 1, it follows from the above that the function

      (7.8)

      where ϑ = ϑ(t0, t) = arg (t − t0), satisfies the H(μ) condition on L. Similarly the function

      ψ(t) = |t−t0|εeαϑ, 0< e ≦ 1, (α any constant)

      (7.8a)

      satisfies the H(ε) condition on L. The function

      (7.4)

      where 0 < μ ≦ 1 and ν any real constant, likewise satisfies the H(μ) condition. For μ = 0, ν ≠ 0, the function

      (7.5)

      obviously does not satisfy the H condition in the neighbourhood of t = t0; it is not even continuous there. However, the function

      (7.6)

      satisfies the H(ε) condition for all arbitrarily small values of ε > 0.

      Clearly, as before, t and t0 may interchange roles.

      3⁰. Let t be a variable, t0 a fixed point on L. Consider the function

      in the neighbourhood of t0. It is easy to prove directly that ω(t) satisfies the condition (7.1) or, what is the same thing, (7.2).

      Hence it may be stated, for example, that

      satisfy the H(ε) condition. Similarly it is easy to prove that

      μ an arbitrary constant, satisfy the H(ε) condition.

      § 8. Sufficient conditions for the H condition to be satisfied (continued).

      Finally another simple result will be given.

      Let the function ƒ(s) of the real variable s, defined in the interval s1 ≦ s ≦ s2, have a continuous n-th derivative ƒ(n)(s) in this interval. Put

      (8.1)

      then the function of the two variables s0. s

      is continuous for s1 ≦ s, s0 ≦ s2. If, further, ƒ(n)(s) satisfies the H(μ) condition, then the preceding function satisfies the H(μ) condition for both variables.

      In fact, it follows from

      (8.2)

      that

      (8.3)

      This will be applied to a simple example.

      Let L = ab be a smooth arc. In addition, assume that the signed angle θ(t) between the tangent to L at t and any fixed direction satisfies the H condition. Then clearly the derivatives x′(s), y′(s) satisfy the H condition.

      Let t = t(s) and t0 = t(s0) respectively be variable and fixed points on L and let ϑ(t0, t) = arg (t − tand some fixed direction and assume that ϑ varies continuously with t, provided t does not pass through t0.

      It will be proved that under the given conditions ϑ satisfies the H condition with respect to t (clearly an interchange of the roles played by t and t0 is again possible) on each of the arcs at0, t0b separately and that

      (8.4)

      where K(t0, t) satisfies the H condition (with respect to t and t0) on L and λ is some real number less than 1.

      Without loss of generality, L = ab may be supposed to be a standard arc. If the Ox axis is taken parallel to the tangent at some point on this arc, x′(s) ≠ 0 for s ≠ s0, x(s) − x(s0) ≠ 0.

      Put

      From what has been proved at the beginning of this section the functions X(s0, s) and Y(s0, s) are continuous on L and satisfy the H condition; in addition, X(s0, s) ≠ 0. By

      will be understood a branch which varies continuously on L. Then

      where C remains constant, while t lies on either of the arcs at0, t0b and changes discontinuously by some odd multiple of π when t passes through t0.

      From the above formula ϑ(t0, t) satisfies the H condition with respect to t, if t lies on one of the arcs at0, t0b. Further, assuming for convenience that the angles θ and ϑ are reckoned from the Ox axis,

      (8.5)

      where r = |t − t0| , and (8.4) follows easily from § 6 (5⁰), since obviously sin (θ − ϑ) satisfies the H condition and is zero at t = t0.

      Next assume that L has the continously changing curvature

      Then, clearly, the second derivatives x"(a) and y"(s) are continuous and it follows from the above that

      are also continuous. Hence

      will also be a continuous function of s (and of swill also satisfy the H condition.

      CHAPTER 2

      INTEGRALS OF THE CAUCHY TYPE

      § 9. Definitions.

      In what follows, L is assumed to be a smooth line (for a generalization of the results to the case in which L is piecewise smooth see Appendix 2), i.e., the union of a finite number of contours or arcs which do not intersect one another and are of finite length.

      A certain direction on L will be defined as positive. If arcs form parts of L, the ends of these arcs will be called ends of the line L.

      Around each point t0, lying on L and not coinciding with an end, a circle may be drawn with radius so small that it is divided by L into two parts, lying respectively to the right and left of L, when viewed in the positive direction of L. Accordingly they will be regarded as the left and right neighbourhoods of the point to.

      If L consists of contours and is the boundary of some connected region, the positive direction of L will be selected so that, if L is described in this direction, that region is always on the left or on the right. That part of the plane which lies on the left will always be denoted by S+ and that to the right by S−.

      A part of L is always to be understood as a portion, consisting of a finite number of contours or arcs.

      § 10. The Cauchy integral.

      Let ϕ(t) be a function of the point t on L, bounded everywhere on L with the possible exception of a finite number of points ck where, however,

      (10.1)

      c stands for any of the points ck, and C, α are positive constants, α < 1.

      In addition, let the function ϕ(t) be integrable with respect to s (the arc coordinate of t) in the Riemann sense on any part of L (cf. the end of § 9) which does not include points ck; from the condition (10.1) ϕ(t) and |ϕ(t)| will be integrable everywhere on L. Consider the integral

      (10.2)

      where z is any point of the plane; this integral is called a Cauchy integral. (Cauchy integrals may naturally be defined for much larger classes of lines L and functions ϕ(t). Cf. for example I. I. Privalov [1], [2], [3].) ϕ(t) will be called the density function.

      Below (§ 12) a definite meaning will be given to this integral in certain cases where the point z lies on L (but not at one of its ends); meanwhile it will be assumed that the point z does not lie on L.

      It is clear that the function Φ(z) is holomorphic in the entire region excluding L and that for large | z | .

      (since L is of finite length).

      § 11. Connection with logarithmic potential. Historical remarks.

      Cauchy integrals are closely related to the potentials of simple and double layers, distributed along the line of integration L. To show this connection, ϕ(t) will be considered as a real function, since the general case follows directly from this.

      Assume that z does not lie on L and put

      (11.1)

      Further, put

      (11.2)

      where r = |t − z| and ϑ = ϑ(z, t) = arg (t − z). Taking the logarithmic derivative of (11.2) (for variable t and constant z),

      (11.8)

      Substituting this in (11.1) and separating real and imaginary parts leads to

      (11.4)

      (11.5)

      where s is the arc coordinate of the point t, n the normal at this point, directed to the left of L, and (r, nand n (Fig. 1). The transformation in (11.4) used the relation

      (11.6)

      which is one of the Cauchy Riemann equations for a system of axes consisting of the positive tangent T and the normal n (this system is right-handed as is the Oxy system) for the function log (t − z) = log r + iϑ (for constant z and variable t).³)

      Fig. 1.

      It follows from (11.4) that U(x, y) represents the potential of a double layer with moment-density ϕ/2π.

      After an integration by parts (assuming for simplicity that ϕ has an integrable derivative with respect to s; this assumption may be replaced by a more general one, e.g. that ϕ. is only continuous, if use is made of the Stieltjes integral) (11.5) may be written in the form

      (11.7)

      where the ck are the ends of L (if it contains arcs) and the rk are the distances between the points (x, y) and ck. If L consists only of contours,

      (11.7a)

      The last formula shows that in the case of contours V(x, y) represents the potential of a simple layer

      (11.8)

      with the density

      (11.9)

      If L contains arcs, the potentials due to point masses, concentrated at the ends ck, must be added to the potential (11.7a). However, it must be noted that, although on the one hand the function V(x, y) defined by (11.5) is a generalization of the potential of a simple layer (for in that formula the function ϕ(t) is not assumed differentiable), in the case of a differentiable ϕ(t) it does not give the potential of a simple layer in its usual form. More specifically, the potential defined by (11.5) corresponds to the ordinary potential of a simple layer, if the masses

      distributed over the individual contours Lk which enter into the composition of L are zero; this is clear from (11.9). Hence, in future, an expression of the form (11.5) will be referred to as a modified potential of a simple layer.

      From the above it is clear that the study of Cauchy integrals may lead to the consideration of logarithmic potentials of double and simple layers. A. Harnack’s work [1] was essentially in this direction and it was one of the first important investigations devoted to Cauchy integrals (1885). However, the direct study of these integrals leads to more general results which are very important from the applied point of view. The first such investigation, which however did not attract due attention, came from G. Morera [1]. (1889). The next important results were those of J. Plemelj [1] published in a short note in 1908 which also did not receive much attention until later. Further very general and important results on Cauchy integrals are due to I. I. Privalov; they are set out in his books [2] and [8]; see also his textbook [1].

      § 12. The values of Cauchy integrals on the path of integration.

      Returning to the direct investigation of Cauchy integrals the case will be considered in which in (10.2) the point z, now to be denoted by t0, lies on L. Write purely formally

      (12.1)

      The integral on the right, in general, has no meaning from the ordinary point of view. However, for an extensive and important class of functions ϕ(t), this integral may be given a definite sense, if the concept of the principal value of a Cauchy integral is introduced. This is as follows.

      Let t0 not coincide with either of the ends of L. Describe about t0 as centre a circle with so small a radius e that it intersects L in the two points t′ and t", and consider the integral

      (12.2)

      where l stands for the arc t′t′′. If for ε → 0 the preceding integral tends to a definite limit, then this limit is called the principal value of the Cauchy integral. It is obvious that, if the integral (12.1) exists in the ordinary (i.e., Riemann) sense, then the principal value also exists (but not conversely). (The integral (12.1) exists in the ordinary sense, if the integral (12.2) tends to a definite limit whatever the arc l cut off around t0 may be, as long as the length of this arc tends to zero; it is essential for the definition of the principal value that the ends t′ and t" of the arc l lie at equal distances from t0. See below). Therefore

      Enjoying the preview?
      Page 1 of 1