Potential Theory - Glossary Analytic function A complex function ๐ค = ๐(๐ง) is said to be analytic at a point ๐ง0 if ๐ is differentiable at ๐ง0 and at every point in some neighbourhood of ๐ง0 . A function is analytic in a domain ๐ท if it is analytic at every point in ๐ท. Note: the term analytic has been used to refer to a function that is in fact holomorphic (i.e. differentiable) rather than expressible as a power series. However, this is not a problem since a complex function is holomorphic if, and only if it is analytic. Complex potential Let ๐(๐ฅ, ๐ฆ) be a harmonic conjugate of a potential function ๐(๐ฅ, ๐ฆ). Then the analytic function Ω(๐ง) = ๐(๐ฅ, ๐ฆ) + ๐๐(๐ฅ, ๐ฆ) is called the complex potential corresponding to the real potential ๐. The function ๐ is called the stream function. The level curves ๐(๐ฅ, ๐ฆ) = ๐1 and ๐(๐ฅ, ๐ฆ) = ๐2 of ๐ and ๐, respectively, are orthogonal families. Conformal Mapping Let ๐ค = ๐(๐ง) be a complex mapping defined in a domain ๐ท, and let ๐ง0 be a point in ๐ท. Then ๐ค is a conformal mapping (or conformal) at ๐ง0 if for every pair of smooth oriented curves ๐ถ1 and ๐ถ2 in ๐ท intersecting at ๐ง0 , the angle between ๐ถ1 and ๐ถ2 is equal to the angle between the image curves ๐ถ1′ and ๐ถ2′ at ๐(๐ง0 ) in both magnitude and sense (direction). If ๐ค = ๐(๐ง) maps a domain ๐ท onto a domain ๐ท′ and if ๐ค is conformal at all points in ๐ท, then ๐ค is a conformal mapping of ๐ท onto ๐ท′ . Critical Point Let the complex function ๐(๐ง) be analytic in a domain ๐ท containing the point ๐ง0 . If ๐ ′ (๐ง0 ) = 0, then the mapping ๐ค = ๐(๐ง) is not conformal at ๐ง0 , and ๐ง0 is called a critical point of ๐. Dipole A pair consisting of a source and a sink is called a dipole (or doublet). Dirichlet's principle Let ๐ท be a domain in the plane bounded by a simple closed curve ๐ถ. Let ๐ be a real-valued smooth function defined on ๐ถ, and let ๐น be the family of real-valued functions with smooth derivatives on ๐ท and ๐ถ and boundary values ๐. If inf {โฌ (๐ข๐ฅ2 + ๐ข๐ฆ2 ) ๐๐ด: ๐ข ∈ ๐น} ๐ท is attained by ๐ ∈ ๐น, then ๐ is the solution of the Dirichlet problem for ๐ท with boundary values ๐. Dirichlet problem Let ๐ท be a domain in the plane bounded by a finite number of nonintersecting simple closed contours ๐ถ๐ , and let ๐๐ be ๐ continuous functions defined on the ๐ contours. The problem of finding a function ๐ข(๐ฅ, ๐ฆ), which satisfies Laplace's equation in ๐ท and which takes on the values ๐๐ on ๐ถ๐ is called a Dirichlet problem. Green’s function Let ๐ง0 = ๐ฅ0 + ๐๐ฆ0 be a point in a region of the plane bounded by a circle ๐ถ of radius ๐ . Then a function of the form 1 ln|๐ง − ๐ง0 | + ๐ป(๐ฅ, ๐ฆ, ๐ฅ0 , ๐ฆ0 ) 2๐ where ๐ป is harmonic for |๐ง| < ๐ and ๐บ(๐ฅ, ๐ฆ, ๐ฅ0 , ๐ฆ0 ) = − ๐ป(๐ cos ๐ , ๐ sin ๐ , ๐ฅ0 , ๐ฆ0 ) = 1 ln|(๐ cos ๐ − ๐ฅ0 ) + ๐(๐ sin ๐ − ๐ฆ0 )| 2๐ Potential Theory - Glossary is called a Green’s function. The essential feature of a Green’s function ๐บ is the singular behaviour specified by 1 ln|๐ง − ๐ง0 |. 2๐ The behaviour on ๐ถ is specified so that an application of Green’s theorem leads to the desired result. − Green's Theorem in the Plane Let ๐ be a region in the plane whose boundary ๐ถ consists of one, or more smooth, non-self-intersecting, closed curves that are positively oriented with respect to ๐ . If ๐(๐ฅ, ๐ฆ)i + ๐(๐ฅ, ๐ฆ)j is a smooth vector field on ๐ , then โฌ( ๐ ๐๐ ๐๐ − ) ๐๐ฅ ๐๐ฆ = โฎ๐(๐ฅ, ๐ฆ)๐๐ฅ + ๐(๐ฅ, ๐ฆ)๐๐ฆ. ๐๐ฅ ๐๐ฆ ๐ถ Harmonic conjugate Suppose that the real function ๐ข(๐ฅ, ๐ฆ) is harmonic on a domain ๐ท, and there exists a real function ๐ฃ(๐ฅ, ๐ฆ) such that the partial derivatives of ๐ข and ๐ฃ satisfy the Cauchy-Riemann equations throughout ๐ท. Then ๐ฃ is said to be a harmonic conjugate of ๐ข. Furthermore, it follows that ๐(๐ง) = ๐ข(๐ฅ, ๐ฆ) + ๐๐ฃ(๐ฅ, ๐ฆ) is analytic on ๐ท. Harmonic function A real valued function ๐(๐ฅ, ๐ฆ) that has continuous first and secondorder partial derivatives in a domain ๐ท, and satisfies Laplace’s equation in two dimensions ∇2 ๐ = ๐ 2๐ ๐ 2๐ + =0 ๐๐ฅ 2 ๐๐ฆ 2 is said to be harmonic in ๐ท. The function ๐ is called a harmonic (or potential) function. Laplace's equation (in two dimensions) Let Φ(๐ฅ, ๐ฆ) be a real valued function of two real variables ๐ฅ and ๐ฆ. The partial differential equation Φ๐ฅ๐ฅ (๐ฅ, ๐ฆ) + Φ๐ฆ๐ฆ (๐ฅ, ๐ฆ) = 0 or alternatively ๐ 2Φ ๐ 2Φ + =0 ๐๐ฅ 2 ๐๐ฆ 2 is known as Laplace's equation, or the potential equation. The polar form of Laplace's equation is given by ๐ 2 Φ๐๐ (๐, ๐) + ๐Φ๐ (๐, ๐) + Φ๐๐ (๐, ๐) = 0 or alternatively ๐ 2Φ ๐Φ ๐ 2 Φ + ๐ + = 0. ๐๐ 2 ๐๐ ๐๐ 2 Level curves Suppose the function ๐(๐ง) = ๐ข(๐ฅ, ๐ฆ) + ๐๐ฃ(๐ฅ, ๐ฆ) is analytic in a domain ๐ท. Then the real and imaginary parts of ๐ can be used to define two families of curves in ๐ท. The equations ๐2 ๐ข(๐ฅ, ๐ฆ) = ๐1 and ๐ฃ(๐ฅ, ๐ฆ) = ๐2 where ๐1 and ๐2 are arbitrary real constants, are called level curves of ๐ข and ๐ฃ, respectively. The level curves are orthogonal families. Neumann problem Let ๐ท be a simply connected domain in the plane bounded by a simple closed smooth contour ๐ถ, and let ๐ be a function defined on ๐ถ. The problem of finding a function ๐ข(๐ฅ, ๐ฆ), which satisfies Laplace's equation in ๐ท and such that ∇๐ข ⋅ N = ๐ where N is the unit outward normal on ๐ถ is called a Neumann problem. Potential Theory - Glossary Poisson integral formulae The Poisson integral formulae are general solutions to the Dirichlet problem in (i) the upper half plane ๐ฆ > 0, and (ii) the unit open disk |๐ง| < 1. (i) Poisson integral formula for the upper half plane. Let ๐(๐ฅ) be a piecewise continuous and bounded function on −∞ < ๐ฅ < ∞. Then the function defined by ๐(๐ฅ, ๐ฆ) = ๐ฆ ∞ ๐(๐ ) ∫ ๐๐ ๐ −∞ (๐ฅ − ๐ )2 + ๐ฆ 2 is a solution of the Dirichlet problem in the upper half plane ๐ฆ > 0 with boundary condition ๐(๐ฅ, 0) = ๐(๐ฅ) at all points of continuity of ๐. (ii) Poisson integral formula for the unit disk. Cartesian form. Let ๐(๐ง) be a complex function for which the values ๐(๐ ๐๐ ) on the unit circle ๐ง = ๐ ๐๐ give a piecewise continuous and bounded function for −๐ ≤ ๐ ≤ ๐. Then the function defined by ๐(๐ฅ, ๐ฆ) = 1 ๐ ๐(๐ ๐๐ )(1 − |๐ง|2 ) ∫ ๐๐ |๐ ๐๐ − ๐ง|2 2๐ −๐ is a solution of the Dirichlet problem in the open unit disk |๐ง| < 1 with boundary condition ๐(cos ๐ , sin ๐) = ๐(๐ ๐๐ ) at all points of continuity of ๐. Polar form. Let ๐(๐) be a piecewise continuous and bounded function on −๐ < ๐ < ๐. Then the function defined by ๐(๐, ๐) = 1 ๐ ๐(๐ )(1 − ๐ 2 ) ∫ ๐๐ 2๐ −๐ 1 − 2๐ cos(๐ − ๐ ) + ๐ 2 is a solution of the Dirichlet problem in the open unit disk |๐ง| < 1 with boundary condition ๐(๐ = 1, ๐) = ๐(๐) at all points of continuity of ๐. Poisson kernel The function ๐: โ → โ defined by ๐ 2 − ๐ 2 ๐(๐๐ ) = 2 ๐ − 2๐๐ cos ๐ + ๐ 2 is known as the Poisson kernel for the disk {๐ง: |๐ง| < ๐ }. It vanishes at every point of {๐ง: |๐ง| = ๐ } except at ๐ง = ๐ , where it has a singularity, and has the property of being positive for 0 < ๐ < ๐ . ๐๐ Polygonal region A polygonal region in the complex plane is a region that is bounded by a simple, connected, piecewise smooth curve consisting of a finite number of line segments. The boundary curve of a polygonal region is called a polygon and the endpoints of the line segments in the polygon are called vertices of the polygon. If a polygon is a closed curve, then the region enclosed by the polygon is called a bounded polygonal region, and a polygonal region that is not bounded is called an unbounded polygonal region. In the case of an unbounded polygonal region, the ideal point ∞ is also called a vertex of the polygon. Simple function A function ๐ค = ๐(๐ง) is said to be simple (or univalent) in a domain ๐ท if it is analytic and one-to-one in ๐ท. Sources and Sinks A source is a point at which fluid is produced (or introduced), whereas, a sink is a point at which fluid disappears (or is removed). Potential Theory - Glossary If the motion of an ideal fluid consists of an outward radial flow from a point and is symmetrical in all directions, then the point is called a simple source. The point is called a simple sink if the flow is inward rather than outward. Streamlining The process of constructing a flow of an ideal fluid that remains inside a given domain ๐ท is called streamlining. Potential Theory - Notes Harmonic Functions and Conjugates Theorem Suppose the complex function ๐(๐ง) = ๐ข(๐ฅ, ๐ฆ) + ๐๐ฃ(๐ฅ, ๐ฆ) is analytic in a domain ๐ท. Then the functions ๐ข(๐ฅ, ๐ฆ) and ๐ฃ(๐ฅ, ๐ฆ) are harmonic in ๐ท. Theorem Suppose that the real function ๐ข(๐ฅ, ๐ฆ) is harmonic in an ๐-neighbourhood of the point (๐ฅ0 , ๐ฆ0 ). Then there exists a conjugate harmonic function ๐ฃ(๐ฅ, ๐ฆ) = ∫ ๐ข๐ฅ (๐ฅ, ๐ฆ) ๐๐ฆ − ∫ ๐ข๐ฆ (๐ฅ, ๐ฆ) ๐๐ฅ − โฌ ๐ข๐ฅ๐ฅ (๐ฅ, ๐ฆ) ๐๐ฆ๐๐ฅ defined in this neighbourhood such that ๐(๐ง) = ๐ข(๐ฅ, ๐ฆ) + ๐๐ฃ(๐ฅ, ๐ฆ) is an analytic function. A harmonic conjugate can be constructed by using the following method. First confirm that ๐ข(๐ฅ, ๐ฆ) is harmonic by verifying that ๐ข๐ฅ๐ฅ (๐ฅ, ๐ฆ) + ๐ข๐ฆ๐ฆ (๐ฅ, ๐ฆ) = 0. Then, using the Cauchy-Riemann equation ๐ฃ๐ฆ (๐ฅ, ๐ฆ) = ๐ข๐ฅ (๐ฅ, ๐ฆ) gives ๐ฃ(๐ฅ, ๐ฆ) = ∫ ๐ฃ๐ฆ (๐ฅ, ๐ฆ) ๐๐ฆ + ๐ถ(๐ฅ) = ∫ ๐ข๐ฅ (๐ฅ, ๐ฆ) ๐๐ฆ + ๐ถ(๐ฅ) where ๐ถ(๐ฅ) is a function of ๐ฅ alone. Differentiating both sides with respect to ๐ฅ and using the Cauchy-Riemann equation ๐ฃ๐ฅ (๐ฅ, ๐ฆ) = −๐ข๐ฆ (๐ฅ, ๐ฆ) gives −๐ข๐ฆ (๐ฅ, ๐ฆ) = ๐ (∫ ๐ข๐ฅ (๐ฅ, ๐ฆ) ๐๐ฆ) + ๐ถ ′ (๐ฅ). ๐๐ฅ All terms except those involving ๐ฅ will cancel leaving a formula for ๐ถ ′ (๐ฅ) in terms of ๐ฅ alone, say ๐ถ ′ (๐ฅ) = ๐(๐ฅ). Integrating both sides with respect to ๐ฅ gives ๐ถ(๐ฅ) = ๐น(๐ฅ) + ๐ where ๐น is an antiderivative of ๐ and ๐ is an arbitrary constant. Finally, letting ๐ = 0 gives ๐ฃ(๐ฅ, ๐ฆ) = ∫ ๐ข๐ฅ (๐ฅ, ๐ฆ) ๐๐ฆ + ๐น(๐ฅ). A similar method can be used if we are given ๐ฃ(๐ฅ, ๐ฆ) and wish to find ๐ข(๐ฅ, ๐ฆ). Laplace’s Equation in Physics Electromagnetism For an electrostatic field, the electric field E = (๐ธ๐ฅ , ๐ธ๐ฆ , ๐ธ๐ง ), where ๐ธ๐ฅ , ๐ธ๐ฆ , ๐ธ๐ง are the Cartesian components of E, is given by E = grad ๐ = ๐๐ ๐๐ ๐๐ i+ j+ k, ๐๐ฅ ๐๐ฆ ๐๐ง where ๐ is an arbitrary function of position known as the electrostatic potential. The electrostatic potential function ๐ satisfies Laplace's equation and is therefore harmonic in some domain ๐ท. If we restrict our attention to two dimensions, it follows that there exists a Potential Theory - Notes harmonic conjugate function ๐(๐ฅ, ๐ฆ) defined in some domain ๐ธ in โ2 , so that the complex potential function Ω(๐ง) = ๐(๐ฅ, ๐ฆ) + ๐๐(๐ฅ, ๐ฆ) is analytic in ๐ธ. The electric field E (i.e. the electron flow) is given by ๐Ω = ๐ธ๐ฅ + ๐๐ธ๐ฆ . ๐๐ง The level curves ๐(๐ฅ, ๐ฆ) = ๐1 are called equipotential curves, that is, curves along which the electrostatic potential is constant. Whereas, the level curves ๐(๐ฅ, ๐ฆ) = ๐2 , curves that are orthogonal to the family ๐(๐ฅ, ๐ฆ) = ๐1, are called lines of force and are the paths along which a charged particle will move in the electrostatic field. Fluid Mechanics For an incompressible fluid, the velocity field v = (๐ฃ๐ฅ , ๐ฃ๐ฆ , ๐ฃ๐ง ), where ๐ฃ๐ฅ , ๐ฃ๐ฆ , ๐ฃ๐ง are the Cartesian components of v, is given by v = grad ๐ = ๐๐ ๐๐ ๐๐ i+ j+ k, ๐๐ฅ ๐๐ฆ ๐๐ง where ๐ is an arbitrary function of position known as the velocity potential. The velocity potential function ๐ satisfies Laplace's equation and is therefore harmonic in some domain ๐ท. If we restrict our attention to two dimensions, it follows that there exists a harmonic conjugate function ๐(๐ฅ, ๐ฆ) defined in some domain ๐ธ in โ2 , so that the complex potential function Ω(๐ง) = ๐(๐ฅ, ๐ฆ) + ๐๐(๐ฅ, ๐ฆ) is analytic in ๐ธ. In this case, the complex potential function Ω(๐ง) is called the complex velocity potential of the flow. The velocity field v (i.e. the fluid flow) is given by ๐Ω = ๐ฃ๐ฅ + ๐๐ฃ๐ฆ . ๐๐ง The level curves ๐(๐ฅ, ๐ฆ) = ๐1 are called equipotential curves, that is, curves along which the velocity potential is constant. Whereas, the level curves ๐(๐ฅ, ๐ฆ) = ๐2 , curves that are orthogonal to the family ๐(๐ฅ, ๐ฆ) = ๐1, are called streamlines and represent the actual paths along which particles in the fluid will move. The Dirichlet Problem Theorem 6.2.1 If the Dirichlet problem has a solution ๐ for a given domain ๐ท and set of functions ๐๐ , then the solution is unique. Theorem 6.2.2 The Dirichlet problem is solvable for the disk |๐ง| < ๐ . Corollary 6.2.1 Let ๐(๐) be a piecewise continuous function on 0 ≤ ๐ ≤ 2๐. Then the function defined by 1 2๐ ๐(๐ )(๐ 2 − ๐ 2 ) ๐(๐, ๐) = ∫ ๐๐ 2๐ 0 ๐ 2 − 2๐ ๐ cos(๐ − ๐ ) + ๐ 2 is harmonic in the open disk |๐ง| < ๐ , and lim ๐(๐, ๐) = ๐(๐) for all but a finite number of values of ๐. ๐→๐ Theorem 6.2.3 Let ๐ท be a simply connected domain with boundary ๐ถ. Let ๐ค = ๐๐ ๐๐ = ๐(๐ง) be analytic in ๐ท, and map ๐ท ∪ ๐ถ in a one to one fashion onto the closed disk |๐ค| ≤ 1 so that ๐ท maps onto the interior of the disk |๐ค| < 1 and ๐ถ maps onto its boundary |๐ค| = 1. Then Potential Theory - Notes ๐ ′ (๐ข, ๐ฃ) = 1 2๐ ๐(๐ )(1 − ๐2 ) ∫ ๐๐ 2๐ 0 1 − 2๐ cos(๐ − ๐ ) + ๐2 where ๐ข = Re(๐ค) = ๐ cos ๐ and ๐ฃ = Im(๐ค) = ๐ sin ๐, solves the Dirichlet problem in the sense that ๐(๐ฅ, ๐ฆ) = ๐ ′ (๐ข, ๐ฃ) at corresponding points under the mapping. Steps for Solving a Dirichlet Problem (i) Find an analytic function ๐(๐ง) = ๐ข(๐ฅ, ๐ฆ) + ๐๐ฃ(๐ฅ, ๐ฆ) that maps the domain ๐ท in the ๐งplane onto a simpler domain ๐ท′ in the ๐ค-plane, and that maps the boundary curves ๐ถ1 , ๐ถ2 , … , ๐ถ๐ onto the curves ๐ถ1′ , ๐ถ2′ , … , ๐ถ๐′ , respectively. (ii) Transform the boundary conditions on ๐ถ1 , ๐ถ2 , … , ๐ถ๐ to boundary conditions on ๐ถ1′ , ๐ถ2′ , … , ๐ถ๐′ . (iii) Solve this new (and easier ) Dirichlet problem in ๐ท′ to obtain a harmonic function Φ(๐ข, ๐ฃ). (iv) Substitute the real and imaginary parts ๐ข(๐ฅ, ๐ฆ) and ๐ฃ(๐ฅ, ๐ฆ) of ๐ for the variables ๐ข and ๐ฃ in Φ(๐ข, ๐ฃ). Then the function ๐(๐ฅ, ๐ฆ) = Φ(๐ข(๐ฅ, ๐ฆ), ๐ฃ(๐ฅ, ๐ฆ)) is a solution to the Dirichlet problem in ๐ท. Green’s Functions Theorem Let ๐ถ be a simple closed smooth contour with interior ๐ , then โฌ (๐ข∇2 ๐ฃ + ∇๐ข ⋅ ∇๐ฃ) ๐๐ฅ ๐๐ฆ = ∫๐ข∇๐ฃ ⋅ N ๐๐ ๐ ๐ถ and โฌ (๐ข∇2 ๐ฃ − ๐ฃ∇2 ๐ข) ๐๐ฅ ๐๐ฆ = ∫(๐ข∇๐ฃ ⋅ N − ๐ฃ∇๐ข ⋅ N) ๐๐ . ๐ ๐ถ where ๐ข and ๐ฃ are functions with continuous first and second order partial derivatives within, and on ๐ถ, N is the unit outward normal on ๐ถ, and ๐๐ is the element of arc length on ๐ถ. Green’s function for the Dirichlet problem Let ๐ท be a bounded simply connected domain bounded by a simple closed smooth contour ๐ถ, and let ๐ง0 = ๐ฅ0 + ๐๐ฆ0 be a point in ๐ท. If ๐บ(๐ฅ, ๐ฆ, ๐ฅ0 , ๐ฆ0 ) is a function satisfying the following properties: 1 (i) ๐บ(๐ฅ, ๐ฆ, ๐ฅ0 , ๐ฆ0 ) = − 2๐ ln|๐ง − ๐ง0 | + ๐ป(๐ฅ, ๐ฆ, ๐ฅ0 , ๐ฆ0 ) where ๐ป is harmonic inside ๐ถ; (ii) ๐บ is continuous in ๐ท ∪ ๐ถ, except at (๐ฅ0 , ๐ฆ0 ) in ๐ท; (iii) ๐บ(๐ฅ, ๐ฆ, ๐ฅ0 , ๐ฆ0 ) = 0 when (๐ฅ, ๐ฆ) is on ๐ถ. Then ๐บ is a Green’s function associated with the Dirichlet problem for the region ๐ = ๐ท ∪ ๐ถ. Theorem 6.3.1 If a Green’s function (for the Dirichlet problem) exists, then it is unique. Theorem 6.3.2 positive in ๐ท. The Green’s function (for the Dirichlet problem) for the region ๐ = ๐ท ∪ ๐ถ is Theorem 6.3.3 If ๐บ is the Green’s function (for the Dirichlet problem) for the region ๐ = ๐ท ∪ ๐ถ, then ๐บ(๐ฅ, ๐ฆ, ๐ฅ1 , ๐ฆ1 ) = ๐บ(๐ฅ, ๐ฆ, ๐ฅ2 , ๐ฆ2 ) for any pair of distinct points (๐ฅ1 , ๐ฆ1 ) and (๐ฅ2 , ๐ฆ2 ) in ๐ท. Theorem 6.3.4 If the Green’s function (for the Dirichlet problem) exists for the region ๐ = ๐ท ∪ ๐ถ, then for ๐(๐ ) continuous on ๐ถ Potential Theory - Notes ๐ข(๐ฅ0 , ๐ฆ0 ) = − ∫(๐(๐ )∇๐บ ⋅ N) ๐๐ ๐ถ where N is the unit outward normal on ๐ถ and ๐๐ is the element of arc length on ๐ถ, solves the Dirichlet problem ∇2 ๐ข = 0 in ๐ท with ๐ข = ๐ on ๐ถ. Theorem 6.3.5 Let ๐ค = ๐(๐ง) map the region ๐ onto the unit closed disk |๐ง| ≤ 1 so that ๐ท, the interior of ๐ , maps onto the open disk |๐ง| < 1 and ๐ถ, the boundary of ๐ , maps onto the unit circle |๐ง| = 1. Furthermore, let ๐ ′ (๐ง) exist and never vanish in ๐ท, and let ๐ง0 map into the 1 origin. Then ๐บ(๐ฅ, ๐ฆ, ๐ฅ1 , ๐ฆ1 ) = − 2๐ ln|๐(๐ง)| is the Green’s function (for the Dirichlet problem) for ๐ . The Neumann Problem An insulated boundary curve in a heat flow problem corresponds to a boundary condition of the form ∇๐ข ⋅ N = 0, and so is an example of a Neumann problem. The solution ๐ข of a Neumann problem is not unique since ๐ข + ๐ is also a solution for any constant ๐. Theorem 6.3.6 The solution ๐ข of the Neumann problem such that ∫๐ถ ๐ข ๐๐ = 0 is unique. Theorem 6.3.7 ∫๐ถ ๐(๐ ) ๐๐ = 0. A necessary condition for the Neumann problem to have a solution is that Green’s function for the Neumann problem Let ๐ท be a bounded simply connected domain bounded by a simple closed smooth contour ๐ถ, and let ๐ง0 = ๐ฅ0 + ๐๐ฆ0 be a point in ๐ท. If ๐บ(๐ฅ, ๐ฆ, ๐ฅ0 , ๐ฆ0 ) is a function satisfying the following properties: 1 (i) ๐บ(๐ฅ, ๐ฆ, ๐ฅ0 , ๐ฆ0 ) = − (ii) ๐บ and its first partial derivatives are continuous in ๐ท ∪ ๐ถ, except at (๐ฅ0 , ๐ฆ0 ) in ๐ท; 2๐ ln|๐ง − ๐ง0 | + ๐ป(๐ฅ, ๐ฆ, ๐ฅ0 , ๐ฆ0 ) where ๐ป is harmonic in ๐ท; (iii) ∇๐บ ⋅ N = − 1⁄๐ฟ where ๐ฟ is the length of ๐ถ, when (๐ฅ, ๐ฆ) is on ๐ถ; (iv) ∫๐ถ ๐บ ๐๐ = 0. Then ๐บ is a Green’s function associated with the Neumann problem for the region ๐ = ๐ท ∪ ๐ถ. Theorem If a Green’s function (for the Neumann problem) exists, then it is unique. Theorem 6.3.9 If the Green’s function (for the Neumann problem) exists for the region ๐ = ๐ท ∪ ๐ถ, then for ๐(๐ ) continuous on ๐ถ ๐ข(๐ฅ0 , ๐ฆ0 ) = ∫๐(๐ )๐บ ๐๐ ๐ถ where ๐ข satisfies Theorems 6.3.6 and 6.3.7 and ๐๐ is the element of arc length on ๐ถ, solves the Neumann problem ∇2 ๐ข = 0 in ๐ท with ∇๐ข ⋅ N = ๐ on ๐ถ. Conformal Mappings Theorem Let the complex function ๐(๐ง) be analytic in a domain ๐ท containing a point ๐ง0 . If ๐ ′ (๐ง0 ) ≠ 0, then ๐ค = ๐(๐ง) is conformal at ๐ง0 . Theorem Let the complex function ๐(๐ง) be analytic at a critical point ๐ง0 . If ๐ > 1 is an integer such that ๐ ′ (๐ง0 ) = ๐ ′′ (๐ง0 ) = โฏ = ๐ (๐−1) (๐ง0 ) = 0, and ๐ (๐) (๐ง0 ) ≠ 0, then the angle between any two smooth curves intersecting at ๐ง0 is increased by a factor ๐ by the complex mapping ๐ค = ๐(๐ง). In particular, ๐ค = ๐(๐ง) is not a conformal mapping at ๐ง0 . Potential Theory - Notes Theorem Laplace's equation is preserved under a mapping ๐ค = ๐(๐ง) provided that ๐ ′ (๐ง0 ) ≠ 0. Theorem 6.4.1 If ๐(๐ง) is simple in a domain ๐ท, then ๐ ′ (๐ง0 ) ≠ 0 in ๐ท. Corollary 6.4.1 If ๐(๐ง) is simple in a domain ๐ท, then it is conformal in ๐ท. Simple functions have the following desirable properties as mappings: (i) they are conformal, i.e. they preserve angles in both magnitude and direction; (ii) they preserve Laplace's equation; (iii) they have a unique inverse which is also a simple function and at every point is the same as the local inverse. Theorem 6.4.2 If ๐ค = ๐(๐ง) has range ๐ท′ and is simple in a domain ๐ท, and ๐น(๐ค) is simple in ๐ท′ , then the composite (๐น โ ๐)(๐ง) = ๐น[๐(๐ง)] is simple in ๐ท. Theorem 6.4.3 Let ๐ถ be a simple closed contour with interior ๐ท. Furthermore, let ๐(๐ง) be analytic within and on ๐ถ, and take no value more than once on ๐ถ. Then ๐(๐ง) is simple in ๐ท. Theorem 6.4.4 Let ๐๐ (๐ง) be simple in a domain ๐ท for all ๐ = 1,2,3, …. Furthermore, let ๐๐ (๐ง) converge uniformly in compact subsets of ๐ท. Then ๐๐ (๐ง) is either constant, or simple in ๐ท. A fundamental problem in the theory of conformal mappings is to find a simple function ๐ค = ๐(๐ง) which maps a given simply connected domain onto the open unit disk |๐ค| < 1. Theorem 6.4.7 (Riemann Mapping Theorem) Let ๐ท be a simply connected domain in the complex plane โ with at least two boundary points. Then there exists a simple function ๐ค = ๐(๐ง) which maps ๐ท onto the open unit disk |๐ค| < 1 in the ๐ค-plane. If we specify that a given point ๐ง0 in ๐ท maps into the origin ๐ค = 0, and a given direction at ๐ง0 is mapped into a given direction at the origin, then the mapping is unique. Let ๐ท and ๐ท′ be simply connected domains in โ. We can use the Riemann mapping theorem to establish the existence of a simple mapping from ๐ท onto ๐ท′ . According to the theorem there is a simple mapping ๐ from ๐ท onto the open unit disk |๐ค| < 1. Similarly, there is a simple mapping ๐ from ๐ท′ onto the open unit disk |๐ค| < 1. Since ๐ is simple, it has a well defined inverse function ๐−1 that maps the open unit disk |๐ค| < 1 onto ๐ท′ . The desired mapping from ๐ท onto ๐ท′ is then given by the composition ๐ค = (๐−1 โ ๐)(๐ง). The Riemann mapping theorem is inadequate in two ways: (i) it is strictly an existence theorem, which assert the existence of a mapping function but does not show how it can be constructed; (ii) it does not show how boundary points of ๐ท correspond to boundary points of ๐ท′ . Theorem 6.4.8 Let ๐ท and ๐ท′ be domains bounded by simple closed contours ๐ถ and ๐ถ ′ , respectively. Then the conformal map ๐: ๐ท → ๐ท′ is continuous in ๐ท ∪ ๐ถ and establishes a oneto-one correspondence between ๐ถ and ๐ถ ′ . The Schwarz-Christoffel Transformation The Schwarz-Christoffel transformation provides a recipe for a conformal mapping of the upper half of the ๐ง-plane to a polygonal region in the ๐ค-plane. Theorem (Schwarz-Christoffel formula) domain ๐ฆ > 0 and has the derivative Let ๐ be a complex function that is analytic in the ๐ ′ (๐ง) = ๐ด(๐ง − ๐ฅ1 )(๐ผ1 ⁄๐)−1 × (๐ง − ๐ฅ2 )(๐ผ2 ⁄๐)−1 × โฏ × (๐ง − ๐ฅ๐ )(๐ผ๐⁄๐)−1 Potential Theory - Notes where ๐ฅ1 < ๐ฅ2 < โฏ < ๐ฅ๐ are arbitrary points on the ๐ฅ-axis, 0 < ๐ผ๐ < 2๐ for 1 ≤ ๐ ≤ ๐, and ๐ด is a complex constant. Then the upper half plane ๐ฆ ≥ 0 is mapped by ๐ค = ๐(๐ง) onto an unbounded polygonal region with interior angles ๐ผ1 , ๐ผ2 , … , ๐ผ๐ . Since ๐ ′ (๐ฅ + ๐๐ฆ) ≠ 0 when ๐ฆ > 0, it follows that the function given by the Schwarz-Christoffel formula is a conformal mapping in the domain ๐ฆ > 0. Furthermore, although the mapping is defined on the upper half plane ๐ฆ ≥ 0, it is only conformal in the domain ๐ฆ > 0. The above theorem provides a formula for the derivative of ๐, which may in principle be integrated to obtain ๐ค as follows ๐ค = ๐(๐ง) = ๐ด ∫(๐ง − ๐ฅ1 )(๐ผ1 ⁄๐)−1 × (๐ง − ๐ฅ2 )(๐ผ2 ⁄๐)−1 × โฏ × (๐ง − ๐ฅ๐ )(๐ผ๐⁄๐)−1 ๐๐ง + ๐ต where ๐ด and ๐ต are complex constants. Thus ๐ is the composition of the function ๐(๐ง) = ∫(๐ง − ๐ฅ1 )(๐ผ1 ⁄๐)−1 × (๐ง − ๐ฅ2 )(๐ผ2 ⁄๐)−1 × โฏ × (๐ง − ๐ฅ๐ )(๐ผ๐⁄๐)−1 ๐๐ง and the linear mapping โ(๐ง) = ๐ด๐ง + ๐ต. The linear mapping โ allows us to rotate, scale, and translate the polygonal region produced by ๐. In practice we usually have a certain amount of freedom in the selection of the points ๐ฅ๐ on the ๐ฅ-axis. A judicious choice can simplify the computation of ๐(๐ง). The Schwarz-Christoffel formula may be used to construct a mapping of the upper half plane ๐ฆ ≥ 0 onto a bounded polygonal region. To do so, we apply the formula using only ๐ − 1 on the ๐ interior angles of the bounded polygonal region. Essentially, we are choosing the point ๐ฅ๐ on the ๐ฅ-axis to lie at infinity. Flows with Sources and Sinks Streamlining Suppose that the complex velocity potential Ω(๐ง) = ๐(๐ฅ, ๐ฆ) + ๐๐(๐ฅ, ๐ฆ) is analytic in a domain ๐ท and that ๐ is constant on the boundary of ๐ท. Then Ω′ (๐ง) is a complex representation of the velocity field of a flow of an ideal fluid in ๐ท. Moreover, if a particle is placed in ๐ท and allowed to flow with the fluid, then its path ๐ง(๐ก) remains in ๐ท. Solving a streamlining problem If ๐ค = ๐(๐ง) is a simple mapping of the domain ๐ท onto a domain ๐ท′ in the ๐ค-plane such that the image of the boundary ๐ถ of ๐ท is a horizontal line in the ๐ค-plane, then ๐ค = Ω(๐ง) is a complex velocity potential of the flow of an ideal fluid in ๐ท and Ω′ (๐ง) is a complex representation of its velocity field. An alternative method is to apply the mapping ๐ค = ๐(๐ง) directly to the complex velocity potential of a uniform flow in the ๐ค-plane Ω๐ท′ = ๐ฃ0 ๐ค ⇒ Ω๐ท = ๐ฃ0 ๐(๐ง). Sources, Sinks and Dipioles The complex velocity potential of a (simple) source of strength ๐ด at the point ๐ง0 is Ω(๐ง) = ๐ด log(๐ง − ๐ง0 ) where ๐ด is a positive real constant. The complex velocity potential of a (simple) sink of strength ๐ด at the point ๐ง0 is Ω(๐ง) = −๐ด log(๐ง − ๐ง0 ) Potential Theory - Notes where ๐ด is a positive real constant. A flow containing both sources and sinks can be described by adding together their respective complex velocity potentials. Suppose that there is a source of strength ๐ด at the point ๐ง1 = ๐๐ ๐๐ and a sink of the same strength at the point ๐ง0 . Then the complex velocity potential of the dipole in the direction ๐ at ๐ง0 is Ω(๐ง) = −๐ด๐ ๐ ๐๐ ๐ง − ๐ง0 where ๐ = ๐ด๐ is strength of the dipole. The velocity potential is ๐ [log|๐ง − ๐ง0 |] ๐๐ where ๐⁄๐๐ is the directional derivative in the direction ๐. ๐(๐ฅ, ๐ฆ) = −๐ Determining fluid flows produced by sources and sinks If ๐ค = ๐(๐ง) is a simple mapping of the domain ๐ท onto a domain ๐ท′ in the ๐ค-plane such that the image of boundary curves, sources, and sinks for the flow in ๐ทare mapped onto boundary curves, sources, and sinks in ๐ท′ where the complex potential is known to be Ω๐ท′ (๐ค). Then the complex potential in ๐ท is given by Ω๐ท (๐ง) = Ω๐ท′ (๐(๐ง)). Sources, sinks and dipoles, and their potentials, have their counterparts in electrostatics, magnetostatics, heat transfer, and so on. Potential Theory - SAQ’s SAQ 1 Which flow is given by the velocity potential ๐(๐ฅ, ๐ฆ) = ๐ฅ + ๐ฅ2 2๐ฅ ? + ๐ฆ2 For very large ๐ฅ 2 + ๐ฆ 2 , ๐ is approximately ๐ฅ, and ๐ฃ๐ฅ = ๐๐ =1 ๐๐ฅ and ๐ฃ๐ฆ = ๐๐ = 0. ๐๐ฆ Hence, at large distances from the origin, the flow is that of a uniform stream with velocity 1 to the right. Consider ๐ง+ 2 2 2(๐ฅ − ๐๐ฆ) 2(๐ฅ − ๐๐ฆ) = ๐ฅ + ๐๐ฆ + = ๐ฅ + ๐๐ฆ + = ๐ฅ + ๐๐ฆ + 2 (๐ฅ + ๐๐ฆ)(๐ฅ − ๐๐ฆ) ๐ง ๐ฅ + ๐๐ฆ ๐ฅ + ๐ฆ2 2๐ฅ 2๐ฆ =๐ฅ+ 2 + ๐ (๐ฆ − 2 ). 2 ๐ฅ +๐ฆ ๐ฅ + ๐ฆ2 Let ๐(๐ฅ, ๐ฆ) = ๐ฆ − ๐ฅ2 2๐ฆ . + ๐ฆ2 Now ๐๐ฅ = 1 + 2(๐ฅ 2 + ๐ฆ 2 ) − 4๐ฅ 2 2(๐ฅ 2 + ๐ฆ 2 ) − 4๐ฆ 2 = 1 − = ๐๐ฆ (๐ฅ 2 + ๐ฆ 2 )2 (๐ฅ 2 + ๐ฆ 2 )2 and ๐๐ฆ = −4๐ฅ๐ฆ = −๐๐ฅ . + ๐ฆ 2 )2 (๐ฅ 2 Hence, the Cauchy-Riemann equations are satisfied, and ๐ is a harmonic conjugate of ๐. It follows that Ω(๐ง) = ๐ง + 2 2๐ฅ 2๐ฆ =๐ฅ+ 2 + ๐ (๐ฆ − ) ๐ง ๐ฅ + ๐ฆ2 ๐ฅ2 + ๐ฆ2 is the complex potential corresponding to ๐, and the streamlines of the flow are given by the level curves of ๐. Now ๐ฆ− ๐ฅ2 2๐ฆ = 0 ⇒ ๐ฆ 3 + ๐ฅ 2 ๐ฆ − 2๐ฆ = 0 ⇒ ๐ฅ 2 + ๐ฆ 2 = 2, + ๐ฆ2 which is the equation of a circle of radius √2 centred at the origin. Since the flow is tangential to this circle it can be considered as a rigid obstacle past which the stream is flowing. Therefore the flow given by the velocity potential ๐ is that of a uniform stream past a circular cylinder of radius √2. SAQ 2 Express ∇2 ๐ข = 0 in two dimensions in polar coordinates by considering ๐ข as the real part of an analytic function and using the Cauchy-Riemann equations in polar form. Potential Theory - SAQ’s Let ๐(๐ง) = ๐ข(๐, ๐) + ๐๐ฃ(๐, ๐) be an analytic function, then the polar form of the CauchyRiemann equations give ๐๐ข ๐๐ฃ ๐๐ฃ ๐๐ข = and ๐ =− . ๐๐ ๐๐ ๐๐ ๐๐ Since ๐ข and ๐ฃ are analytic, ๐ ๐ ๐๐ข ๐ ๐๐ฃ ๐ ๐๐ฃ ๐ 1 ๐๐ข (๐ ) = ( ) = ( )=− ( ). ๐๐ ๐๐ ๐๐ ๐๐ ๐๐ ๐๐ ๐๐ ๐ ๐๐ Hence ๐ ๐๐ข ๐ 1 ๐๐ข (๐ ) + ( ) = 0. ๐๐ ๐๐ ๐๐ ๐ ๐๐ Using the chain rule gives ๐ ๐๐ข ๐ 1 ๐๐ข ๐ 2 ๐ข ๐๐ข 1 ๐ 2 ๐ข (๐ ) + ( )=๐ 2+ + = 0. ๐๐ ๐๐ ๐๐ ๐ ๐๐ ๐๐ ๐๐ ๐ ๐๐ 2 Which gives ๐2 ๐ 2๐ข ๐๐ข ๐ 2 ๐ข + ๐ + = 0. ๐๐ 2 ๐๐ ๐๐ 2 SAQ 3 Find the potential inside a capacitor consisting of an infinitely long circular cylinder of radius ๐ charged to a potential of 10 volts, and an infinitely long wire of radius ๐ along the axis of the cylinder at a potential of zero. Hint: by symmetry, the solution must be independent of the polar angle ๐. The electrostatic potential ๐(๐) inside the capacitor is independent of the polar angle ๐ and so satisfies the polar form of Laplace's equation ๐2 ๐ ๐๐ +๐ = 0, ๐ < ๐ < ๐ 2 ๐๐ ๐๐ with boundary conditions ๐2 ๐(๐) = 0, ๐(๐) = 10. The differential equation above is a Cauchy-Euler equation. Indicial equation: ๐2 = 0 ⇒ ๐ = 0. General solution: ๐(๐) = ๐ด + ๐ต ln ๐. Applying the boundary conditions gives ๐(๐) = ๐ด + ๐ต ln ๐ = 0 ⇒ ๐ด = −๐ต ln ๐ and ๐(๐) = ๐ด + ๐ต ln ๐ = 10 ⇒ −๐ต ln ๐ + ๐ต ln ๐ = 10 ⇒ ๐ต = Hence, the required potential equation is 10 . ln ๐ − ln ๐ Potential Theory - SAQ’s ๐(๐) = −๐ต ln ๐ + 10 10(ln ๐ − ln ๐) ln ๐ = . ln ๐ − ln ๐ ln ๐ − ln ๐ SAQ 4 Find the potential in an annulus bounded by circles of radii ๐ < ๐ if the inner circle has potential ๐ผ and the outer circle has potential ๐ฝ. The electrostatic potential ๐(๐) inside the capacitor is independent of the polar angle ๐ and so satisfies the polar form of Laplace's equation ๐2 ๐ ๐๐ +๐ = 0, ๐ < ๐ < ๐ 2 ๐๐ ๐๐ with boundary conditions ๐2 ๐(๐) = ๐ผ, ๐(๐) = ๐ฝ. The differential equation above is a Cauchy-Euler equation. Indicial equation: ๐2 = 0 ⇒ ๐ = 0. General solution: ๐(๐) = ๐ด + ๐ต ln ๐. Applying the boundary conditions gives ๐(๐) = ๐ด + ๐ต ln ๐ = ๐ผ ⇒ ๐ด = ๐ผ − ๐ต ln ๐ and ๐(๐) = ๐ด + ๐ต ln ๐ = ๐ฝ ⇒ ๐ผ − ๐ต ln ๐ + ๐ต ln ๐ = ๐ฝ ⇒ ๐ต = ๐ฝ−๐ผ . ln ๐ − ln ๐ Hence, the required potential equation is ๐(๐) = ๐ผ − ๐ต ln ๐ + SAQ 5 ๐ฝ−๐ผ ๐ผ ln ๐ − ๐ฝ ln ๐ + (๐ฝ − ๐ผ) ln ๐ ln ๐ = . ln ๐ − ln ๐ ln ๐ − ln ๐ What conditions on the real numbers ๐, ๐, ๐, ๐ ensure that ๐๐ฅ 3 + ๐๐ฅ 2 ๐ฆ + ๐๐ฅ๐ฆ 2 + ๐๐ฆ 3 is harmonic. Let ๐(๐ฅ, ๐ฆ) = ๐๐ฅ 3 + ๐๐ฅ 2 ๐ฆ + ๐๐ฅ๐ฆ 2 + ๐๐ฆ 3 . Then ๐ is harmonic if it has continuous first and second-order partial derivatives in a domain ๐ท, and satisfies Laplace’s equation ∇2 ๐ = Now ๐ 2๐ ๐ 2๐ + = 0. ๐๐ฅ 2 ๐๐ฆ 2 Potential Theory - SAQ’s ๐๐ = 3๐๐ฅ 2 + 2๐๐ฅ๐ฆ + ๐๐ฆ 2 ๐๐ฅ and ๐ 2๐ = 6๐๐ฅ + 2๐๐ฆ ๐๐ฅ 2 ๐๐ = 2๐๐ฅ + 2๐๐ฅ๐ฆ + 3๐๐ฆ 2 ๐๐ฆ and ๐ 2๐ = 2๐๐ฅ + 6๐๐ฆ. ๐๐ฆ 2 and So ๐ has continuous first and second-order partial derivatives in the complex plane. Substituting the second-order partial derivatives into Laplace's equation gives 6๐๐ฅ + 2๐๐ฆ + 2๐๐ฅ + 6๐๐ฆ = 2(3๐ + ๐)๐ฅ + 2(๐ + 3๐)๐ฆ = 0 which is satisfied when ๐ = −3๐ and ๐ = −3๐ . Hence ๐(๐ฅ, ๐ฆ) = ๐๐ฅ 3 − 3๐๐ฅ 2 ๐ฆ − 3๐๐ฅ๐ฆ 2 + ๐๐ฆ 3 = ๐(๐ฅ 3 − 3๐ฅ๐ฆ 2 ) + ๐(๐ฆ 3 − 3๐ฅ 2 ๐ฆ). Now ๐ง 3 = (๐ฅ + ๐๐ฆ)3 = ๐ฅ 3 − 3๐ฅ๐ฆ 2 + ๐(3๐ฅ 2 ๐ฆ − ๐ฆ 3 ). Which gives ๐(๐ฅ, ๐ฆ) = ๐ Re(๐ง 3 ) − ๐ Im(๐ง 3 ) for all complex ๐ง. SAQ 6 ๐ง+1 By considering arg (๐ง−1), find a function ๐ harmonic in the unit disk such that ๐(๐ ๐๐ ) = { 1 0<๐<๐ −1 −๐ < ๐ < 0. Let ๐ง = ๐ ๐๐ , then |๐ง| < 1 and (cos ๐ + 1) + ๐ sin ๐ ๐ ๐๐ + 1 arg ( ๐๐ ) = arg ( ) (cos ๐ − 1) + ๐ sin ๐ ๐ −1 [(cos ๐ + 1) + ๐ sin ๐][(cos ๐ − 1) − ๐ sin ๐] = arg ( ) (cos ๐ − 1)2 + sin2 ๐ − sin ๐ = arg ( ๐) 1 − cos ๐ ๐ − 0<๐<๐ 2 ={ ๐ −๐ < ๐ < 0. 2 So, we define 2 ๐ง+1 ๐(๐ง) = − arg ( ). ๐ ๐ง−1 Let ๐ง+1 ๐ง+1 ๐(๐ง) = log ( )=| | + ๐๐(๐ง). ๐ง−1 ๐ง−1 Potential Theory - SAQ’s Since ๐ is analytic in the unit open disk |๐ง| < 1, it follows that ๐(๐ง) = Im(๐(๐ง)) is harmonic there. SAQ 7 By considering the definition of a Green's function on page 245 and Theorem 6.3.4 prove the result given in Exercises 6.3 #2. We must show that 1 1 ln √(๐ฅ − ๐ฅ0 )2 + (๐ฆ − ๐ฆ0 )2 + ln √(๐ฅ − ๐ฅ0 )2 + (๐ฆ + ๐ฆ0 )2 2๐ 2๐ is a Green's function for the upper half-plane problem in the sense that the solution, if it exists, can be expressed as ๐บ(๐ฅ, ๐ฆ, ๐ฅ0 , ๐ฆ0 ) = − ∞ ๐ฆ0 ∞ ๐(๐ฅ) ๐ข(๐ฅ0 , ๐ฆ0 ) = ∫ ๐(๐ฅ)๐บ๐ฆ (๐ฅ, 0, ๐ฅ0 , ๐ฆ0 ) ๐๐ฅ = ∫ ๐๐ฅ. ๐ −∞ (๐ฅ − ๐ฅ0 )2 + ๐ฆ02 −∞ Let let ๐ง0 = ๐ฅ0 + ๐๐ฆ0 be a point in the open upper half-plane๐ท, and let ๐ = ๐ท ∪ ๐ถ where ๐ถ is the ๐ฅ-axis. Then 1 1 ln|๐ง − ๐ง0 | + ln|๐ง − ๐ง0 | 2๐ 2๐ 1 1 =− Re(log(๐ง − ๐ง0 )) + Re(log(๐ง − ๐ง0 )). 2๐ 2๐ ๐บ(๐ฅ, ๐ฆ, ๐ฅ0 , ๐ฆ0 ) = − Hence ๐บ is analytic in ๐ท. It follows that 1 1 Re(log(๐ง − ๐ง0 )) = ln √(๐ฅ − ๐ฅ0 )2 + (๐ฆ + ๐ฆ0 )2 2๐ 2๐ is harmonic in ๐ท. Also, ๐บ is continuous in ๐ท ∪ ๐ถ, except at (๐ฅ0 , ๐ฆ0 ) in ๐ท. Finally, since ๐บ(๐ฅ, 0, ๐ฅ0 , ๐ฆ0 ) = 0, we have ๐บ(๐ฅ, ๐ฆ, ๐ฅ0 , ๐ฆ0 ) = 0 when (๐ฅ, ๐ฆ) is on ๐ถ. Hence, ๐บ is a Green’s function associated with the Dirichlet problem for the closed upper halfplane. By Theorem 6.3.4, the solution to the Dirichlet problem is given by ๐ข(๐ฅ0 , ๐ฆ0 ) = − ∫(๐(๐ )∇๐บ ⋅ N) ๐๐ ๐ถ where N is the unit outward normal on ๐ถ and ๐๐ is the element of arc length on ๐ถ, solves the Dirichlet problem ∇2 ๐ข = 0 in ๐ท with ๐ข = ๐ on ๐ถ. In this case N = −j and ๐๐ = ๐๐ฅ. Hence, integrating along the ๐ฅ-axis gives ∞ ∞ ๐ข(๐ฅ0 , ๐ฆ0 ) = ∫ (๐(๐ฅ)∇๐บ ⋅ j) ๐๐ฅ = ∫ (๐(๐ฅ)๐บ๐ฆ (๐ฅ, ๐ฆ, ๐ฅ0 , ๐ฆ0 )|๐ฆ=0 ) ๐๐ฅ −∞ −∞ Potential Theory - SAQ’s ∞ = ∫ (๐(๐ฅ) −∞ + 1 ln √(๐ฅ − ๐ฅ0 )2 + (๐ฆ + ๐ฆ0 )2 ] ) ๐๐ฅ 2๐ ๐ฆ=0 ∞ = ∫ (๐(๐ฅ) −∞ + ๐ 1 [− ln √(๐ฅ − ๐ฅ0 )2 + (๐ฆ − ๐ฆ0 )2 ๐๐ฆ 2๐ ๐ 1 [− ln((๐ฅ − ๐ฅ0 )2 + (๐ฆ − ๐ฆ0 )2 ) ๐๐ฆ 4๐ 1 ln((๐ฅ − ๐ฅ0 )2 + (๐ฆ + ๐ฆ0 )2 )] ) ๐๐ฅ 4๐ ๐ฆ=0 =− 1 ∞ 2(๐ฆ − ๐ฆ0 ) 2(๐ฆ + ๐ฆ0 ) ∫ (๐(๐ฅ) [ − ] ) ๐๐ฅ 2 2 (๐ฅ − ๐ฅ0 ) + (๐ฆ − ๐ฆ0 ) (๐ฅ − ๐ฅ0 )2 + (๐ฆ + ๐ฆ0 )2 4๐ −∞ ๐ฆ=0 =− 1 ∞ −2๐ฆ0 2๐ฆ0 ∫ (๐(๐ฅ) [ − ]) ๐๐ฅ 2 2 (๐ฅ − ๐ฅ0 ) + ๐ฆ0 (๐ฅ − ๐ฅ0 )2 + ๐ฆ02 4๐ −∞ ๐ฆ0 ∞ ๐(๐ฅ) = ∫ ๐๐ฅ. ๐ −∞ (๐ฅ − ๐ฅ0 )2 + ๐ฆ02 SAQ 8 Prove that the function ๐(๐ง) = ๐ง + ๐2 ๐ง 2 + ๐3 ๐ง 3 + โฏ is simple for |๐ง| < 1 if ∑∞ ๐=2 ๐|๐๐ | ≤ 1. Since ∞ ∞ ๐ด ∑ ๐ ≤ ∑ ๐|๐๐ | ≤ 1 ๐=2 ๐=2 where ๐ด = min{|๐2 |, |๐3 |, … }, we have ๐ด≤ 1 ∑∞ ๐=2 ๐ which implies that |๐๐ | → 0 as ๐ → ∞. ๐ Hence, by the root test, ๐(๐ง) = ∑∞ ๐=2 ๐๐ ๐ง defines an analytic function on |๐ง| < ∞. Suppose that ๐ง1 ≠ ๐ง2 , then ๐(๐ง1 ) − ๐(๐ง2 ) = (๐ง1 + ๐2 ๐ง12 + ๐3 ๐ง13 + โฏ ) − (๐ง2 + ๐2 ๐ง22 + ๐3 ๐ง23 + โฏ ) = (๐ง1 − ๐ง2 + ๐2 (๐ง12 − ๐ง22 ) + ๐3 (๐ง13 − ๐ง23 ) + โฏ ) = (๐ง1 − ๐ง2 )(1 + ๐2 (๐ง1 + ๐ง2 ) + ๐3 (๐ง12 + ๐ง1 ๐ง2 + ๐ง22 ) + โฏ ). Hence, by the reverse triangle inequality |๐(๐ง1 ) − ๐(๐ง2 )| ≥ ||๐ง1 + ๐2 ๐ง12 + ๐3 ๐ง13 + โฏ | − |๐ง2 + ๐2 ๐ง22 + ๐3 ๐ง23 + โฏ || = |๐ง1 − ๐ง2 |(1 + |๐2 |(|๐ง1 | + |๐ง2 |) + |๐3 |(|๐ง12 | + |๐ง1 ๐ง2 | + |๐ง22 |) + โฏ ) ≥ |๐ง1 − ๐ง2 |(1 + 2|๐2 | + 3|๐3 | + โฏ ) > 0. Potential Theory - SAQ’s So, ๐ง1 ≠ ๐ง2 implies that ๐(๐ง1 ) ≠ ๐(๐ง2 ), which is the contrapositive of ๐(๐ง1 ) = ๐(๐ง2 ) ⇒ ๐ง1 = ๐ง2 ; the necessary condition for a one-to-one function. It follows that ๐(๐ง) is analytic and one-to-one on |๐ง| < 1, and so is simple there. SAQ 9 Let ๐(๐ง) be simple within and on a simple closed contour ๐ถ. Let ๐ท the interior of ๐ถ, map onto ๐ท′ , and ๐ถ map onto ๐ถ ′ , the boundary of ๐ท′ . (i) Prove that the length of ๐ถ ′ is ∫๐ถ|๐ ′ (๐ง)| |๐๐ง|; (ii) Prove that the area of ๐ท′ is โฌ๐ท|๐ ′ (๐ง)|2 ๐๐ฅ๐๐ฆ; (iii) If ๐ถ is a circle of radius ๐ centred at the origin, prove that the area of ๐ท ′ is greater than, or equal to ๐|๐ ′ (0)|2 ๐ 2. (i) Since ๐ is simple, the curve ๐ถ is mapped one-to-one onto the curve ๐ถ ′ . Hence, ๐ถ ′ is smooth and has a length ๐ฟ given by ๐ฟ = ∫ |๐๐ค|. ๐ถ′ Now ๐๐ค = ๐ ′ (๐ง)๐๐ง, which gives ๐ฟ = ∫ |๐ ′ (๐ง)๐๐ง| = ∫|๐ ′ (๐ง)||๐๐ง|. ๐ถ′ ๐ถ (ii) Let ๐ง = ๐ฅ + ๐๐ฆ, then the area element ๐๐ด at ๐ง is ๐๐ฅ๐๐ฆ. The mapping ๐ค = ๐(๐ง) changes small distances near ๐ง0 by the scale factor |๐ ′ (๐ง0 )|. Therefore, ๐๐ด is transformed by ๐ into an element of area of ๐ท′ given by |๐ ′ (๐ง)|2 ๐๐ฅ๐๐ฆ at ๐(๐ง) ∈ ๐ท′ . Hence, ๐ท′ has an area ๐ด given by ๐ด = โฌ |๐ ′ (๐ง)|2 ๐๐ฅ๐๐ฆ. ๐ท (iii) Too difficult to waste time on. SAQ 10 Prove that ๐ค = 3๐ง + ๐ง 2 is simple on the disk |๐ง| < 1. Let ๐ค = ๐(๐ง), then ๐(๐ง) is analytic on |๐ง| < 1. Suppose that ๐ง1 ≠ ๐ง2 , then ๐(๐ง1 ) − ๐(๐ง2 ) = (3๐ง1 + ๐ง12 ) − (3๐ง2 + ๐ง22 ) = (3(๐ง1 − ๐ง2 ) + (๐ง12 − ๐ง22 )) = (๐ง1 − ๐ง2 )(3 + ๐ง1 + ๐ง2 ). Hence, by the reverse triangle inequality |๐(๐ง1 ) − ๐(๐ง2 )| ≥ ||3๐ง1 + ๐ง12 | − |3๐ง2 + ๐ง22 || = |๐ง1 − ๐ง2 |(3 + |๐ง1 | + |๐ง2 |) ≥ 5|๐ง1 − ๐ง2 | > 0. So, ๐ง1 ≠ ๐ง2 implies that ๐(๐ง1 ) ≠ ๐(๐ง2 ), which is the contrapositive of ๐(๐ง1 ) = ๐(๐ง2 ) ⇒ ๐ง1 = ๐ง2 ; the necessary condition for a one-to-one function. It follows that ๐(๐ง) is analytic and one-to-one on |๐ง| < 1, and so is simple there. Potential Theory - Problems P1 Consider a 2-dimensional flow outside a simple curve ๐ถ in the (๐ฅ, ๐ฆ)-plane. If the velocity v = (๐ฃ๐ฅ , ๐ฃ๐ฆ ) is such that ๐ฃ๐ฅ − ๐๐ฃ๐ฆ is analytic outside ๐ถ, i.e. ๐ฃ๐ฅ − ๐๐ฃ๐ฆ = ๐(๐ง) with ๐ง = ๐ฅ + ๐๐ฆ and ๐(๐ง) analytic, show that the circulation is the same round any simple curve surrounding ๐ถ. Cannot answer this problem. There is nothing in the text book or course notes that suggests a way to answer this question. P2 Consider the hypothesis of the preceding problem with ๐ถ the unit circle {|๐ง| = 1}. Prove ๐ that if the circulation is non-zero, then ๐(๐ง) is of the form ∑∞ −∞ ๐๐ ๐ง with Im(๐−1 ) = 0. Since ๐ is analytic on the open annulus ๐ด = {1 < |๐ง| < ∞}, it has the Laurent series representation ∞ ๐(๐ง) = ∑ ๐๐ ๐ง ๐ ๐=−∞ for all ๐ง ∈ ๐ด. The velocity field v is given by ๐Ω = ๐ฃ๐ฅ + ๐๐ฃ๐ฆ . ๐๐ง Hence ๐Ω = ๐ฃ๐ฅ − ๐๐ฃ๐ฆ = ๐(๐ง) ⇒ ๐๐ง Ω(๐ง) = ∫๐(๐ง) ๐๐ง. ๐ถ Which gives, for the circle |๐ง| = 1 ∞ ๐(๐ฅ, ๐ฆ) = Re (∫ ∑ ๐๐ ๐ง ๐ ๐๐ง). ๐(๐ง) ๐๐ง) = Re (∫ |๐ง|=1 |๐ง|=1 ๐=−∞ On the unit circle ๐ง = ๐ , which gives ๐๐ ∞ 2๐ ๐(๐ฅ, ๐ฆ) = Re (∫ (๐๐ ๐๐ 0 ∑ ๐๐ ๐ ๐๐๐ ) ๐๐) ๐=−∞ ∞ 2๐ = Re (๐ ∑ ๐๐ ∫ ๐ ๐(๐+1)๐ ๐๐) 0 ๐=−∞ 2๐ = Re (๐๐−1 ∫ ๐๐) = Re(๐2๐๐−1 ) 0 since 2๐ ∫ ๐ ๐๐๐ ๐๐ = 0 0 for any non-zero integer ๐. Consequently, ๐(๐ฅ, ๐ฆ) is non-zero if Im(๐−1 ) = 0. Potential Theory - Problems P3 Find a steady state temperature for a thin metal infinite strip {(๐ฅ, ๐ฆ): −∞ < ๐ฅ < ∞, 0 ≤ ๐ฆ ≤ 1} when the upper edge is perfectly insulated and ๐ข(−1,0) = 3 and ๐ข(1,0) = 2. How many solutions are there for ๐ข(๐ฅ, ๐ฆ)? The steady-state temperature ๐ข is a solution of Laplace's equation in the domain ๐ท defined by −∞ < ๐ฅ < ∞, 0 ≤ ๐ฆ ≤ 1, which satisfies the boundary conditions ๐ข(−1,0) = 3; ๐ข(1,0) = 2; ∇๐ข ⋅ N = 0 for −∞ < ๐ฅ < ∞, ๐ฆ = 1. In this case, the unit outward facing normal is N = j, hence (∇๐ข ⋅ N)|๐ฆ=1 = ๐๐ข | = 0. ๐๐ฆ ๐ฆ=1 We seek a harmonic function ๐ข, which behaves as required on the boundary, and in particular satisfies the condition (∇๐ข ⋅ N)|๐ฆ=1 = 0. One possibility is that ๐ข(๐ฅ, ๐ฆ) is a function of ๐ฅ alone, i.e. ๐ข(๐ฅ, ๐ฆ) = ๐(๐ฅ) for some function ๐ of ๐ฅ. In order to satisfy Laplace's equation we must have ๐2 ๐ข ๐2๐ข ๐2๐ + = = 0. ๐๐ฅ 2 ๐๐ฆ 2 ๐๐ฅ 2 Integrating twice gives ๐(๐ฅ) = ๐ด๐ฅ + ๐ต for arbitrary constants ๐ด, ๐ต. Applying the remaining boundary conditions gives ๐ข(−1,0) = ๐(−1) = −๐ด + ๐ต = 3 and ๐ข(1,0) = ๐(1) = ๐ด + ๐ต = 2. Hence 1 5 and ๐ต = 2 2 which gives a solution ๐ด=− 5 ๐ฅ − . 2 2 Another possibility is ๐ข(๐ฅ, ๐ฆ) = ๐(๐ฅ)โ(๐ฆ) where โ′ (1) = 0. ๐ข(๐ฅ, ๐ฆ) = Consider the function ๐(๐ง) = ๐ ๐๐ง−๐๐ for real ๐. Since ๐ is analytic in ๐ท, it follows that Re(๐ ๐๐ง−๐๐ ) = Re(๐ ๐๐ฅ ๐ ๐(๐ฆ−1)๐ ) = ๐ ๐๐ฅ cos(๐(๐ฆ − 1)) Potential Theory - Problems is harmonic in ๐ท. Now ๐ ๐๐ฅ [๐ cos(๐(๐ฆ − 1))]| = −๐ ๐๐ฅ ๐ sin(0) = 0. ๐๐ฆ ๐ฆ=1 Choosing ๐ = ๐⁄2 gives ๐ ๐ ๐๐ฅ⁄2 cos ( (๐ฆ − 1)) 2 which is equal to zero when ๐ฆ = 0. Therefore ๐ข(๐ฅ, ๐ฆ) = 5 ๐ฅ ๐ − + ๐ถ๐ ๐๐ฅ⁄2 cos ( (๐ฆ − 1)) 2 2 2 solves the problem for any constant ๐ถ, which means there are infinitely many solutions. P4 Prove the assertion on p.235 that ๐พ๐ 2 ∫ 3 (๐ฅ๐๐ฅ + ๐ฆ๐๐ฆ + ๐ง๐๐ง) ๐ถ ๐ is independent of the path assuming that it is smooth. Note: the quantity represented by ๐ 3 is the distance from the origin. Let the path ๐ถ be parameterised by (๐ฅ(๐ก), ๐ฆ(๐ก), ๐ง(๐ก)) for ๐≤๐ก≤๐ where ๐ and ๐ are the left and right endpoints of ๐ถ, respectively. Since ๐ถ is smooth, the derivatives ๐ฅ ′ (๐ก), ๐ฆ ′ (๐ก), ๐ง ′ (๐ก) exist. Which gives ๐ฅ๐๐ฅ + ๐ฆ๐๐ฆ + ๐ง๐๐ง = ๐ฅ(๐ก) ๐๐ฅ ๐๐ฆ ๐๐ง ๐๐ก + ๐ฆ(๐ก) ๐๐ก + ๐ง(๐ก) ๐๐ก ๐๐ก ๐๐ก ๐๐ก on ๐ถ. Hence ๐ ๐พ๐ 2 ๐ฅ(๐ก)๐ฅ ′ (๐ก) + ๐ฆ(๐ก)๐ฆ ′ (๐ก) + ๐ง(๐ก)๐ง ′ (๐ก) 2 (๐ฅ๐๐ฅ + ๐ฆ๐๐ฆ + ๐ง๐๐ง) = ๐พ๐ ∫ ๐๐ก 3 3⁄2 ๐ถ ๐ ๐ (๐ฅ 2 (๐ก) + ๐ฆ 2 (๐ก) + ๐ง 2 (๐ก)) ∫ = −๐พ๐ 2 ๐ 1 (๐ฅ 2 (๐ก) + ๐ฆ 2 (๐ก) + ๐ง 2 (๐ก)) 1⁄2 | ๐ since ๐ 2 2 2 ๐ 1 1 ๐๐ก [๐ฅ (๐ก) + ๐ฆ (๐ก) + ๐ง (๐ก)] [ ]=− ๐๐ก (๐ฅ 2 (๐ก) + ๐ฆ 2 (๐ก) + ๐ง 2 (๐ก))1⁄2 2 (๐ฅ 2 (๐ก) + ๐ฆ 2 (๐ก) + ๐ง 2 (๐ก))3⁄2 Potential Theory - Problems =− ๐ฅ(๐ก)๐ฅ ′ (๐ก) + ๐ฆ(๐ก)๐ฆ ′ (๐ก) + ๐ง(๐ก)๐ง ′ (๐ก) (๐ฅ 2 (๐ก) + ๐ฆ 2 (๐ก) + ๐ง 2 (๐ก)) 3⁄2 . So the line integral depends only on the endpoints of ๐ถ, which are independent of its actual path. P5 If ๐ข(๐ง) is harmonic and positive in the unit disk and continuous in the closed unit disk show that 1−๐ 1+๐ ≤ ๐ข(๐๐ ๐๐ ) ≤ ๐ข(0) . 1+๐ 1−๐ Hint: first find upper and lower bounds for the Poisson kernel of the unit disk. ๐ข(0) As shown in the proof of Theorem 6.2.2, for the unit disk {๐ง: |๐ง| < ๐ = 1} (1 − ๐ 2 )๐ข(๐ ๐๐ ) 1 2๐ ๐ข(๐๐ ) = ∫ ๐๐ 2๐ 0 1 − 2๐ cos(๐ − ๐) + ๐ 2 ๐๐ 1 2๐ 1 − ๐2 = ∫ ( ) ๐ข(๐ ๐๐ ) ๐๐ 2 2๐ 0 1 − 2๐ cos(๐ก) + ๐ with ๐ก = ๐ − ๐. Hence ๐ข(0) = 1 2๐ ∫ ๐ข(๐ ๐๐ ) ๐๐. 2๐ 0 Since both the Poisson kernel and ๐ข(๐ ๐๐ ) are positive 1 2๐ 1 − ๐2 ∫ min ๐ข(๐ ๐๐ ) ๐๐ ≤ ๐ข(๐๐ ๐๐ ) 2 0≤๐ก≤2๐ 2๐ 0 1 − 2๐ cos(๐ก) + ๐ ≤ 1 2๐ 1 − ๐2 ∫ max ๐ข(๐ ๐๐ ) ๐๐. 2๐ 0 0≤๐ก≤2๐ 1 − 2๐ cos(๐ก) + ๐ 2 Since 1 − ๐2 1 − ๐2 1 − ๐2 = = 0≤๐ก≤2๐ 1 − 2๐ cos(๐ก) + ๐ 2 1 + 2๐ + ๐ 2 (1 + ๐)2 min and 1 − ๐2 1 − ๐2 1 − ๐2 = = . 0≤๐ก≤2๐ 1 − 2๐ cos(๐ก) + ๐ 2 1 − 2๐ + ๐ 2 (1 − ๐)2 max We have 1 − ๐ 2 1 2๐ 1 − ๐ 2 1 2๐ ๐๐ ๐๐ ∫ ๐ข(๐ ) ๐๐ ≤ ๐ข(๐๐ ) ≤ ∫ ๐ข(๐ ๐๐ ) ๐๐. (1 + ๐)2 2๐ 0 (1 − ๐)2 2๐ 0 Which gives, since ๐ < 1 1−๐ 1+๐ ๐ข(0) ≤ ๐ข(๐๐ ๐๐ ) ≤ ๐ข(0). 1+๐ 1−๐ Potential Theory - Problems P6 Let ๐ท be a domain in the plane bounded by a simple closed curve ๐ถ. Let ๐ be a realvalued smooth function defined on ๐ถ, and let ๐น be the family of real-valued functions with smooth derivatives on ๐ท and ๐ถ and boundary values ๐. If inf {โฌ (๐ข๐ฅ2 + ๐ข๐ฆ2 ) ๐๐ด: ๐ข ∈ ๐น} ๐ท is attained by ๐ ∈ ๐น, show that ๐ is the solution of the Dirichlet problem for ๐ท with boundary values ๐. A long and complicated proof is the required answer. Not worth wasting time on. P7 Find a univalent mapping of the half disk {๐ง: |๐ง| < 1, Im(๐ง) > 0} onto the unit disk. Let ๐ท = {๐ง: |๐ง| < 1, Im(๐ง) > 0}. According to the Riemann mapping theorem, there exists a simple function ๐ค = ๐(๐ง) which maps ๐ท onto the open unit disk |๐ค| < 1 in the ๐ค-plane. The boundary of the domain ๐ท consists of two circular arcs (i) the upper semicircle {๐ง: |๐ง| = 1, Im(๐ง) > 0}; (ii) the ๐ฅ-axis as a circle through ๐ง = 0 and ๐ง = ∞. Consider the linear fractional transformation 1+๐ง . 1−๐ง The images of the points −1, ๐, and 1 under this transformation are ๐ค= ๐ค(−1) = 0, ๐ค(๐) = ๐, and ๐ค(1) = ∞. Therefore, the upper semicircle is mapped to the positive imaginary axis. The images of the points −1,0, and 1 under this transformation are ๐ค(−1) = 0, ๐ค(0) = 1, and ๐ค(1) = ∞. Therefore, the ๐ฅ-axis is mapped to the positive real axis. The test point ๐ง = 1⁄2 is mapped to ๐ค(1 + ๐) = 1 + ๐ ⁄2 3 4 = + ๐. 1 − ๐ ⁄2 5 5 Hence, ๐ค maps ๐ท onto the first quadrant in the ๐ค-plane. The mapping ๐ = ๐ค 2 maps the first quadrant in the ๐ค-plane onto the upper half ๐-plane. The mapping ๐ − ๐ −1(๐ − ๐) = ๐+๐ ๐+๐ maps the the upper half ๐-plane onto the unit disk in the ๐-plane. ๐= Forming the composition of these mappings gives Potential Theory - Problems 2 ๐= ๐−๐ ๐−๐ค = = ๐ + ๐ ๐ + ๐ค2 1+๐ง 2 ๐ − (1 − ๐ง) 1+๐ง 2 ๐ + (1 − ๐ง) = ๐(1 − ๐ง)2 − (1 + ๐ง)2 1 − 2๐๐ง + ๐ง 2 = ๐ . ๐(1 − ๐ง)2 + (1 + ๐ง)2 1 + 2๐๐ง + ๐ง 2 P8 Let ๐(๐ง) be univalent in {|๐ง| < 1} with ๐(0) = 0. Show that one can find ๐(๐ง) univalent in the unit disk and satisfying 2 (๐(๐ง)) = ๐(๐ง 2 ) for |๐ง| < 1. The solution is too complex to be worth bothering with. P9 Let ๐(๐ง) be univalent and map the annulus {1 < |๐ง| < ๐ 1 } univalently onto the annulus {1 < |๐ค| < ๐ 2 } so that the inner and outer boundaries correspond. Show that ๐(๐ง) = ๐ด๐ง for some constant ๐ด with |๐ด| = 1, and so ๐ 1 = ๐ 2 . What happens when {|๐ง| = ๐ 1 } maps to {|๐ค| = 1} and {|๐ง| = 1}? The solution is too complex to be worth bothering with. P10 Find a function which maps the strip ๐ = {๐ง: 0 < Im(๐ง) < ๐} into the interior of the unit circle. Write down a solution of the Dirichlet problem for this strip. The mapping ๐ = ๐๐ง maps ๐ onto the upper half ๐-plane, and the mapping ๐−๐ ๐+๐ maps the upper half ๐-plane onto the unit disk in the ๐ค-plane. ๐ค= Therefore, the mapping ๐ − ๐ ๐ − ๐๐ง ๐ค= = = ๐(๐ง) ๐ + ๐ ๐ + ๐๐ง maps ๐ onto the interior of the unit circle. The Poisson integral formula gives a solution for the unit disk, from which a solution in ๐ can be obtained. P11 Find a function which maps the strip ๐ = {๐ง: 0 ≤ Re(๐ง) ≤ ๐, Im(๐ง) ≥ 0} onto the upper half plane so that the origin maps into the origin, and ๐ maps into 1. Potential Theory - Problems The strip ๐ is a polygonal region with interior angles ๐ผ1 = ๐ผ2 = ๐⁄2 and vertices ๐ง1 = 0 and ๐ง2 = ๐. Applying the Schwarz-Christoffel transformation gives ๐๐ง ๐ด = ๐ด(๐ค − 0)(๐⁄2⁄๐)−1 × (๐ค − 1)(๐⁄2⁄๐)−1 = ๐ด๐ค −1⁄2 (๐ค − 1)−1⁄2 = . 1⁄2 ๐๐ค (๐ค(๐ค − 1)) Hence 1 ๐ง = ๐ด∫ (๐ค(๐ค − 1)) = ๐ด∫ 1⁄2 ๐๐ค = ๐ด ∫ 1 1 2 1 2 ((๐ค − 2) − (2) ) = ๐ด log (๐ค − 1⁄2 (๐ค 2 1 ๐๐ค − ๐ค)1⁄2 ๐๐ค 1 + (๐ค 2 − ๐ค)1⁄2 ). 2 This isn't very helpful. Further manipulation gives ๐๐ง ๐ด ๐ด 1 = = . ⁄ 1 2 ๐๐ค (−1๐ค(1 − ๐ค)) ๐ (๐ค(1 − ๐ค))1⁄2 Let ๐ต = ๐ด⁄๐ , then ๐ง = ๐ต∫ 1 (๐ค(1 − ๐ค)) = ๐ต∫ 1⁄2 ๐๐ค = ๐ต ∫ 1 1 2 1 2 ((2) − (๐ค − 2) ) 1⁄2 1 ๐๐ค (๐ค − ๐ค 2 )1⁄2 ๐๐ค 1 ๐ค−2 = ๐ต sin−1 ( ) + ๐ถ. 1 2 Now ๐ง1 = 0 when ๐ค1 = 0, hence ๐ 0 = ๐ต sin−1(−1) + ๐ถ = − ๐ต + ๐ถ. 2 And ๐ง2 = ๐ when ๐ค2 = 1, hence ๐ ๐ = ๐ต sin−1(1) + ๐ถ = ๐ต + ๐ถ. 2 Subtracting gives ๐ต=1 ๐. ๐ = ๐๐ต ⇒ { ๐ถ= 2 Hence Potential Theory - Problems ๐ง = sin −1 ( 1 2 ) + ๐. 1 2 2 ๐ค− Rearranging gives 1 ๐ 1 1 1 1 ๐ง ๐ค = sin (๐ง − ) + = − cos ๐ง + = (1 − cos ๐ง) = sin2 ( ). 2 2 2 2 2 2 2