SMA 2371-PARTIAL DIFFRENTIAL EQUATIONS January 4, 1980 Course Outline 1. Surfaces and Curves in three dimensions 2. -Simultaneous dierential equations of the rst order. 3. -Methods of solution of dx P = dy Q = dz R 4. -Orthogonal trajectories of systems cuves on a surface 5. -Linear partial dierential equationd of the rst order. 6. Partial dierential equations of the second order: Laplace, Poisson, heat and wave equations. 7. -Methods of solution by separation of variables foe Cartesian, Spherical, polar and cylindrical polar coordinates and by Laplace and Fourier Transforms 8. Applications to Engineering Reference Books 1. Elements of Partial dierential equations by I,N, Sneddon, Mc Graw Hill(1957) 1 2. p.d.e-Schaum Outline Series 3. p.d.e-Garabedian (1964) 2 Contents TABLE OF CONTENTS 3 1 INTRODUCTION 5 2 Partial Dierential Equation 6 2.1 Denition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Formation of Partial Dierential Equations. . . . . . . . . . 2.2.1 Formation of P.D.E by eliminating arbitrary constant 2.2.2 Formation of P.D.E by eliminating arbitrary function 2.3 Derivation of P.D.E equations by eliminating arbitrary functions φ (u, v) = 0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 6 7 9 . 10 3 LANGRAGE METHOD OF SOLVING THE QUASI LINEAR PARTIAL DIFFERENTIAL EQUATIONS OF ORDER 1 NAMELY; P p + Qq = R 14 3.1 Integral surfaces passing through a given curve . . . . . . . . . 16 3.2 Surfaces Orthogonal to given system of surfaces . . . . . . . . 20 4 SPECIAL TYPES OF FIRST ORDER PARTIAL DIFFERENTAL EQUATIONS 23 5 SURFACES AND CURVES IN THREE DIMENSIONS: 30 6 METHODS OF SOLVING 33 dx p = dy q = dz R =λ 7 ORTHOGONAL TRAJECTORIES OF CURVES: 41 8 METHOD OF SEPARATION OF VARIABLES: 49 9 DERIVATION OF WAVE EQUATIONS 59 9.1 Forces on the bit of the string . . . . . . . . . . . . . . . . . . 60 3 10 HEAT EQUATION 62 11 DERIVATION OF THE WAVE EQUATION 72 12 DERIVATION OF HEAT EQUATION 74 13 OTHER BOUNDARY VALUE PROBLEMS 76 14 METHOD OF LAPLACE TRANSFORMS AND FOURIER TRANSFORMS: 77 14.1 SINE AND COSINE FOURIER TRANSFORMS . . . . . . . 78 15 FURTHER EXAMPLES ON METHOD OF SEPARATION VAIABLES 79 4 1 INTRODUCTION Partial dierential equations are used to formulate problems involving functions of several variables and are either solved by hand, or used to create a relevant computer model. Denition A dierential equation is an equation that relates the derivatives at a scalar function depending on one or more variables. Example: 2 δ4 u + δδxu2 +u2 = cos(x)is a dierential equation of the function u(x)depending δx4 2 2 on x, while δu = δδxu2 + δδyu2 − uis a dierential equation involving a function δt u(x, y, t) depending on x, y, t. This equation can be written in subscript notation as ut = uxx + uyy − u. Partial dierential operators (i) δxδ = δx 2 (ii) δxδ 2 = δx2 or δxx 3 (iii) δxδ2 δy = δxxy or δx2 δy Surfaces and Curves in three dimensions (i) 3-D coordinate system in often denoted by R3 (ii) 2-D coordinate system in often denoted by R2 (iii) 1-D coordinate system in often denoted by R (iv) n-D coordinate system in often denoted by Rn q Distance between any two points is given by: d (p1 , p2 ) = (x2 − x1 )2 + (y2 − y1 )2 + (z2 − z1 )2 and the general equation for a sphere with center (h, k, l)and radius r is given by (x − h)2 + (y − k)2 + (z − l)2 = r2 Examples 1. Graph x = 3in R,R2 and R3 Solution In Rwe have a single coordinated system and so x = 3is a point in a 1-D coordinate system 5 In R2 the equation x = 3tells us to graph all the points that are in the form (3, y).This is a vertical line in a 2-D coordinate system In R3 the equation x = 3tells us to graph all the points onf the form (3, y, z).This is similar to the coordinates for the yz plane except this time we have x = 3instead of x = 0. So in a 3-D system this is a plane that will be parallel to the yz plane and pass through the x-axis at x = 3. 2 Partial Dierential Equation 2.1 Denition This is an equation involving partial dierential coecients of functions of two or more variables is known as partial dierential equation. If a p.d.e contains nth and lower derivatives of nth order the degree of such equation is the greatest exponent of the highest order. We adopt the following notion through out the study of p.d.e's δz δ 2 z δ2 z δ2 z δz ,q = δy , δx2 = r, δxδy = sand δy p = δx 2 = twhere z is a dependent variable. In case there are n independent variables we take them to be x1 , x2 , x3 , ..., xn and z is regarded as dependent variable i.e. p1 = δxδz1 ,p2 = δxδz2 ,....pn = δxδzn , Sometimes the partial derivatives are as denoted by making use of suxes thus 2 δ2 u ux = δu ,u = δu ,u = δδxu2 ,ut = δu ,uxy = δxδy . δx y δy xx δt 2.2 Formation of Partial Dierential Equations. There are two main methods of forming a p.d.e. (i) Method of elimination of arbitrary constants (ii)Method of elimination of arbitrary functions 6 2.2.1 Formation of P.D.E by eliminating arbitrary constant Example (i) z = ax + by + a2 + b2 .............(i) Solution δz = a.................................(ii) δx δz = b.................................(iii) δy Replacinng (ii) and (iii) in equation (i), wehave: z= δz δx x+ δz δy y+ δz δx 2 + δz δy 2 z = px + qy + p2 + q 2 (ii) z = (x − a)2 + (y − b)2 ..............(i) Solution δz = 2 (x − a) δx δz = 2 (y − b) δy 2 2 1 δz 1 δz z= + . 4 δx 4 δy 1 1 z = p2 + q 2 4 4 4z = p2 + q 2 Exercise: Form a P.D.E. a) z = Aept sin(px) Solution δz = Apept cos(px) δx δ2z = −Ap2 ept sin(px)......................(i) δx2 7 δz = Apept sin(px) δt δ2z = Ap2 ept sin(px)......................(ii) δt2 Ading, (i) and (ii), we have: δ2z δ2z + 2 = −Ap2 ept sin(px) + Ap2 ept sin(px) = 0 2 δx δt b) xa2 + yb2 + zc2 = 1..................(i) Solution Dierentiting (i) with respect to x, 2 2 2 2x 2z δz δz + 2 = 0 =⇒ c2 x + a2 z = 0..............(ii) 2 a c δx δx Dierentiting (i) with respect to y, δz 2y 2z δz + 2 = 0 =⇒ c2 y + b2 z = 0..............(iii) 2 b c δy δy Dierentiting (ii) with respect to x, 2 c +a δz δx 2 2 δ2z + a z 2 = 0..................(iv) δx 2 Dierentiting (iii) with respect to y, 2 c +b From (ii) 2 2 δz δy + b2 z a2 z c =− x 2 Putting (vi) into(iv) a2 z − x δz δx +a 2 δz δx δz δx 2 δ2z = 0..................(v) δy 2 .....................(vi) δ2z + a z 2 = 0..................(vii) δx 2 Dividing (vii) by a2 andmultiplying by x δ2z zx 2 + x δx δz δx 2 −z δz δx Similarly from (iii), 8 = 0....................(viii) b2 z c =− y 2 Putting (ix) into (v) 2 bz − y δz δy +b 2 δz δy δz δy 2 δ2z + b z 2 = 0..................(x) δy .....................(ix) 2 Dividing (x) by a2 and multiplying by y δ2z zy 2 + y δy δz δy 2 −z δz δy = 0....................(xi) 2.2.2 Formation of P.D.E by eliminating arbitrary function This is done by eliminating the arbitrary function; f or F. Example:4 a) y = f (x − at) + F (x + at) .............(i) Solution δy = 1 · f 0 (x − at) + 1 · F 0 (x + at) δx δy = −a · f 0 (x − at) + a · F 0 (x + at) δt δ2y = f 00 (x − at) + F 00 (x + at) 2 δx δ2y = a2 [f 00 (x − at) + F 00 (x + at)] δt2 2 δ2y 2δ y = a δt2 δx2 b) z = f (x2 − y) + g (x2 + y) ......(i) Solution δz = 2xf 0 x2 − y + 2xg 0 x2 + y ........(ii) δx δ2z = 2x · 2xf 00 x2 − y + 2f 0 x2 + y + 2x · 2xg 0 x2 + y + 2g 0 x2 + y 2 δx 00 2 0 2 δ2z 2 00 2 0 2 = 4x f x − y + g x + y +2 g x + y + f x + y ........(iii) δx2 9 δz = −f 0 x2 − y + g 0 x2 + y .......(iv) δy δ2z 00 2 00 2 = f x − y + g x + y .............(v) δy 2 1 δz = f 0 x2 − y + g 0 x2 + y ............(vi) 2x δx Substituting (vi) and (v) into (iii), 2 δ2z 1 δz 2 δ z +2 ........(vii) = 4x δx2 δy 2 2x δx 2 δ2z δz 3 δ z x 2 = 4x + ........(viii) 2 δx δy δx 2.3 Derivation of P.D.E equations by eliminating arbitrary functions φ (u, v) = 0 Consider φ (u, v) = 0...........................(i), where u and v are functions of x,y and z. This expression represents a solution that has variables u and v. We take z to be dependent variable while x and y are the independent variables δy δz δz so that δx = p, δy = q, δx = 0, δx =0 δy Dierentiating(i) partially to xand y in turn we get; with respect δu δv δu δφ δv +p +p + .........................(ii) δx δz δv δx δz δφ δu δu δv δφ δv +q + +q .........................(iii) δu δy δz δv δy δz δφ δu From (ii), δφ δu − δφ = δv Similarly from (iii) δφ δu − δφ = δv δv δx δu δx + p δv δz ...........................(iv) δu + p δz δv δy δu δy + q δv δz + q δu δz .............................(v) Equating (iv) and (v) and simplifying yields, 10 δv δv +p δx δz δu δu +q δy δz = δv δv +q δy δz δu δu +p δx δz δv δu δv δu δv δu δv δu δv δu δv δu δv δu δv δu · +q · +p · +pq = · +p · +q · +pq · δx δy δx δz δz δy δz δz δy δx δy δz δz δx δz δz δv δu δv δu δv δu δv δu δv δu δv δu δv δu δv δu · − · +q · − · = − · −pq + · +pq · p δz δy δy δz δx δz δz δx δx δy δz δz δy δx δz δz δv δu δv δu δv δu δv δu δv δu δv δu p · − · +q · − · = · − · δz δy δy δz δx δz δz δx δy δx δx δy P p + Qq = R This is a partial dierential equation and can be classied as follows: (i) If it is a linear p.d.e in p and q(i.e) P (x, y, z) p + Q (x, y, z) q = R (x, y, z) (ii) Semi-Linear p.d.e: if it is linear in p and q and the coecient of p and q are functions of x and y only i.e. P (x, y) p + Q (x, y) q = R (x, y, z) Example a) Form a partial dierential equation whose solution is φ (x2 ez , yez ) = 0 Solution φ x2 ez , yez = 0.................................(i) Here u = x2 ez and v = yez φ (u, v) = 0 δu δx δv = 2xez , δu = x2 ez , δu = 0, δx = 0, δv = yez δz δy δz Substituting this in the equation, δv δx δu δx + p δv δz = + p δu δz δv δy δu δy + q δv δz + q δu δz 0 + pyez ez + qyez = 2xez + px2 ez 0 + qx2 ez 11 pyez ez + qyez = 2xez + px2 ez qx2 ez py 1 + qy = 2 2x + px qx2 pyqx2 = 2x + px2 (1 + qy) pyqx2 = 2x + px2 + 2xqy + pqx2 y + x2 p + 2xyq x2 p + 2xyq = −2x xp + 2yq = −2 From this, p = x, q = 2y , R = −2 b) Form a p.d.e equation whose solution is φ (x + y + z, x2 + y 2 − z 2 ) = 0 Solution Here u = x + y + z and v = x2 + y 2 − z 2 φ (u, v) = 0 δu δx δv δv = 1, δu = 1, δu = 1, δx = 2x, δv = −2z, δy = 2y δz δy δz Substituting this in the equation, δv δx δu δx + p δv δz = + p δu δz δv δy δu δy + q δv δz + q δu δz 2x − 2pz 2y − 2qz = 1+p 1+q (2x − 2pz) (1 + q) = (1 + p) (2y − 2qz) 2x + 2xq − 2pz − 2pqz = 2y − 2qz + 2py − 2pqz x + xq − zp = y − zq + py (x + z) q − (z + y) p = y − x wherep = −z − y , q = x + z , R = y − x (b) Form a partial dierential equation whose solution is φ (x2 + y 2 + z 2 , z 2 − 2xy) = 0 12 Solution φ x2 ez , yez = 0.................................(i) Here u = x2 + y 2 + z 2 and v = z 2 − 2xy φ (u, v) = 0 δu δx δv = 2x, δu = 2z, δu = 2y, δx = −2y, δv = 2z δz δy δz Substituting this in the equation, δv δx δu δx + p δv δz = δu + p δz δv δy δu δy + q δv δz + q δu δz −2y + 2pz −2x + 2qz = 2x + 2pz 2y + 2qz −x + qz −y + pz = x + pz y + qz (−y + pz) (y + qz) = (x + pz) (−x + qz) −y 2 − qyz + pyz + pqz 2 = −x2 − pxz + qxz + pqz 2 (yz + xz) p − (yz + xz) q = y 2 − x2 z (y + x) p − z (y + x) q = (y + x) (y − x) zp − zq = (y − x) wherep = z , q = −z , R = y − x 13 3 LANGRAGE METHOD OF SOLVING THE QUASI LINEAR PARTIAL DIFFERENTIAL EQUATIONS OF ORDER 1 NAMELY; P p+ Qq = R The general solution of the linear partial dierential equation are P p + Qq = R where p,q, R are functions of x,y,z is φ (u, v) = 0where φis an arbitrary function and u (x, y, z) = c1 )andv (x, y, z) = c2 ) form a solution of equations dy dz dx = = p q R Example: 2 1. Solve yxz p + x2 q = y 2 Solution This dierential equation is of the form P p + Qq = R 2 wherep = yxz , q = xz , R = y 2 Using langrage auxillary equation, dx y2 z x Taking the rst two fractions, = dz dy = 2 xz y dy xdx = 2 y z xz x2 dx − y 2 dy = 0 x3 y 3 c1 − = 3 3 3 or x3 − y 3 = c1 =⇒ u Taking rst and third fraction xdx dz = 2 2 y z y xdx − zdz = 0 14 x2 z 2 c2 − = 2 2 2 x2 − z 2 = c2 =⇒ v But φ (u, v) = 0, therefore φ x3 − y 3 , x2 − z 2 = 0 where φis an arbitrary function. Exercise: a) p + 3q = 5z + tan (y − 3x) This dierential equation is of the form P p + Qq = R wherep = 1, q = 3, R = 5z + tan (y − 3x) Using langrage auxillary equation, dx dy dz = = 1 3 5z + tan (y − 3x) Using the rst two fractions dx dy = 1 3 3dx = dy 3x − y = c y − 3x = c1 Using the rst and the third fraction, dx dz = 1 5z + tan (y − 3x) dx dz = 1 5z + tan (c1 ) 1 x − ln |5z + tan (c1 )| = c2 5 5x − ln |5z + tan (c1 )| = c2 But φ (u, v) = 0, therefore φ (y − 3x, 5x − ln |5z + tan (y − 3x)|) = 0 15 thus, v (x, y, z) = φ(u) where φis an arbitrary function. Note: The possible solutions of p.d.e is φ (u, v) = 0,u = φ(v)or v = φ(u) 3.1 Integral surfaces passing through a given curve 1.Determine the integral surface of the p.d.e. (x − y) p + (y − x − z) q = z through the circle z = 1, x + y = 1 Solution The equation (x − y) p + (y − x − z) q = z is of the form P p + Qq = R wherep = (x − y), q = (y − x − z), R = z Using langrage auxillary equation, 2 2 dy dz dx = = ......(i) (x − y) (y − x − z) z Using all the fractions dx + dy + dz dx + dy + dz = x−y+y−x−z+z 0 dx + dy + dz = 0 x + y + z = c1 ...................(ii) Using the second and the third fraction, dy dz = . (y − (x + z)) z From (ii) (x + z) = (c1 − y) dy dz = . (y − (c1 − y)) z dy dz = . 2y − c1 z 1 ln |2y − c1 | = ln |z| + ln |c| 2 16 (2y − c1 ) = c2 z 2 2y − x − y − z = c2 z2 y−x−z = c2 .......................(iii) z2 Put z =1 in equation (ii) and (iii) x + y = c1 − 1...................(iv) y − x = c2 + 1...................(v) But, 2 x2 + y 2 = (x + y)2 + (y − x)2 ........................(vi) But x2 + y 2 = 1 2 = (c1 − 1)2 + (c2 + 1)2 ........................(vii) 2 = c21 − 2c1 + 1 + c22 + 2c2 + 1 c21 − 2c1 + c22 + 2c2 = 0 2 2 (y − x − z) (y − x − z) 2 =0 + (x + y + z) − 2 (x + y + z) + 2 z z2 (x + y + z)2 − 2 (x + y + z) + (y − x − z)2 2 (y − x − z) + =0 z4 z2 z 4 (x + y + z)2 + (y − x − z)2 − 2z 4 (x + y + z) + 2z 2 (y − x − z) = 0 2. Find the equation of the surface satisfying the equation 4yzp + q + 2y = 0 and passing through y 2 + x2 = 1, x + z = 2 Solution The equation (x − y) p + (y − x − z) q = z is of the form P p + Qq = R wherep = 4yz , q = 1, R = −2y Using langrage auxillary equation, dx dy dz = = ......(i) 4yz 1 −2y Taking rst and third fractions, 17 dx dz dx dz 1 = =⇒ = − =⇒ dx = −zdz 4yz −2y 2z 1 2 1 z2 x+ = c1 2 2 x + z 2 = c1 ..................(ii) Taking second and third fractions, dy dz = =⇒ dz + 2ydy = 0 1 −2y z + y 2 = c2 ..............(iii) Equations (iii) and (iv) are the integrable surfaces x + z 2 + z + y 2 = c1 + c2 (x + z) + z 2 + y 2 = c1 + c2 2 + 1 = c1 + c2 3 = c1 + c2 (x + z) + z 2 + y 2 = 3 is the given integral surface. 3) Find the integral surface of x2 p + y 2 q + z 2 = 0 which passes through the hyperbolar xy = x + y , z = 1 Solution The dierential equation x2 p + y 2 q + z 2 = 0can be written in the form P p + Qq = R wherep = x2 , q = y 2 , R = −z 2 Using langrage auxillary equation, dy dz dx = 2 = ......(i) 2 x y −z 2 Taking rst and third fractions dx dz 1 1 = =⇒ − − = c 2 2 x −z x z 18 1 1 + = c1 x z Taking second and third fractions dy dz 1 1 − =c = =⇒ − y2 −z 2 y z 1 1 + = c2 y z Adding (i) and (ii) 1 1 2 + + = c1 + c2 x y z y+x 2 + = c1 + c2 yx z 2 + 1 = c1 + c2 1 1 2 + + =3 x y z yz + xz + 2xy = 3xyz ASSIGNMENT Find the integral surface of equation 4yzp − q + 2y = 0which passes through the hyperbola y 2 = 9 + z and z = 5 − x. Solution The dierential equation 4yzp − q + 2y = 0can be written in the form P p + Qq = R wherep = 4yz , q = −1, R = −2y Using langrage auxillary equation, dy dz dx = = ......(i) 4yz −1 −2y Taking rst and third fractions dx dz dx dz 1 zdz 1 z2 = =⇒ = =⇒ dx = − =⇒ x + = c1 4yz −2y 2z −1 2 1 2 2 x + z 2 = c1 ...........................(ii) Taking second and third fractions 19 dy dz dz y2 1 = =⇒ ydy = =⇒ − z = c2 −1 −2y 2 2 2 y 2 − z = c2 ..................................(iii) Adding (ii) and (iii) x + 2z + y 2 − z = c1 + c2 However there are no combination suiting our given equations, thus; But y 2 = 9 + z 2 and z = 5 − x. Rewriting, 9 + z2 = y2 9 + z 2 = z + c2 But z = 5 − xand z 2 = c1 − x 9 + c1 − x = 5 − x + c2 4 = c2 − c1 Replacing c1 and c2 from equations(ii) and (iii), 4 = y2 − z − x + z2 4 = y2 − z − x − z2 3.2 Surfaces Orthogonal to given system of surfaces ..............(i) are The curves whose equations are the solution of dxp = dyq = dz R orthogonal to the system of the surfaces whose equations are pdx + qdy + Rdz = 0..............(ii). 1. Find the family of surfaces orthogonal to the family of surfaces given by dierential equations (y + z) p + (z + x) q = x + y Solution p = y + z , q = z + x, R = x + y Then the given dierential equation can be written in the form P p+Qq = R. 20 The dierential equation of the family of the surface is orthogonal to the given family is given by pdx + qdy + Rdz = 0where (y + z) dx + (z + x) dy + (x + y) dz = 0 Rearranging, (y + z) dx + (z + x) dy + (x + y) dz = 0 (ydx + zdx) + (zdy + xdy) + (xdz + ydz) = 0 Rearranging, (ydx + xdy) + (zdy + ydz) + (xdz + zdx) = 0 d (xy) + d (yz) + d (xz) = 0 Integrate, xy + yz + xz = c where c is a constant of integration. Exercise: 1. Find the family of curves orthogonal to φ z (x + y)2 , x2 − y 2 = 0 Solution Solution φ z (x + y)2 , x2 − y 2 = 0.................................(i) Here u = z (x + y)2 and v = x2 − y 2 φ (u, v) = 0 δv δv = 2z (x + y) , δu = (x + y)2 , δu = 2z (x + y) , δx = 2x, δv = 0, δy = δz δy δz δv −2y, δz = 0 δu δx Substituting this in the equation, δv δx δu δx + p δv δz = + p δu δz δv δy δu δy + q δv δz + q δu δz 2x −2y 2 = (x + y) + p (x + y) 2z (x + y) + q (x + y)2 2 x [2z + q (x + y)] = −y [2z + p (x + y)] 21 2xz + qx2 + qxy = −2zy − xyp − y 2 p xy + y 2 p + x2 + xy q = −2z (y + x) py (x + y) + qx (x + y) = −2z (y + x) py + qx = −2z From this, p = y , q = x, R = −2z Hence the equation is of the form P p + Qq = R and is orthogonal to pdx + qdy + Rdz = 0. Thus, ydx + xdy − 2zdz = 0 But ydx + xdy = d (xy) Hence, d (xy) − 2zdz = 0 xy − z 2 = c 22 4 SPECIAL TYPES OF FIRST ORDER PARTIAL DIFFERENTAL EQUATIONS We discuss special types of rst order partial dierential equations whose solution can be determined by the Charpit's method. Case 1: A rst order partial dierential equation is said to be of the form Charpit's form, if it can be written in the form: z = px + qy + f (p, q) ................(i) Solution Let F (x, y, z, p, q) = z − px − q(y) − f (p, q).................(ii) Therefore the Charpit's auxillary equation is: δF δx dp = + p δF δZ δF δy dq dz dx dy = = δF = δF δF δF δF + q δZ −p δp − q δq − δp − δq Using the rst and the second fractions, dq dp = −p + p −q + q This implies that dp = 0 =⇒ p = aand dq = 0 =⇒ q = b.............................(iii) Substituting equation(iii) into equation (i) z = ax + by + f (a, b) ................(iv) Equation (iv) gives the complete integral of equation (i) Example: a) Find the complete integral(C.I) and the Singular Integral (S.I) of (i) z = px + qy + log(pq)..................(i) Solution Comparing this with the given equation of Charpit form, the complete integral is given by z = ax + by + log(ab) z = ax + by + log(a) + log(b)......................(ii) 23 To obtain the singular integral we dierentiate (ii) partially with respect to a and b to obtain (iii) and (iv), 1 1 =⇒ a = − ..................(iii) a x 1 1 0 = y + =⇒ b = − ..................(iv) b y 0=x+ We then substitute equations (iii) and (iv) in (ii) to obtain the singular integral. z= − 1 x x+ − 1 y y + log − 1 x + log − 1 y 1 1 z = −1 − 1 + log − + log − x y 1 z = −2 + log xy √ (ii) z = px + qy − 2 pq..................(i) Solution Comparing this with the given equation of Chaurat's form, the complete integral is given by √ z = ax + by − 2 ab..................(ii) To obtain the singular integral we dierentiate (ii) partially with respect to a and b to obtain (iii) and (iv), r 2b b b 0 = x − √ = x − √ =⇒ x = √ = 2 ab ab ab b .........................(iii) a r 2a a a a 0 = y − √ = y − √ =⇒ y = √ = .........................(iv) b 2 ab ab ab r r r r √ b b a b x−z = −a −b + 2 ab = ...............................(v) a a b a r r r r √ a b a a y−z = −a −b + 2 ab = ...............................(vi) b a b b Multiplying (v) and (vi) 24 r (x − z) (y − z) = b × a r a =1 b (x − z) (y − z) = 1 pq (iii) Prove that z = px + qy + pq−p−q represents all planes such that the algebraic sum of the intercepts on three coordinate axes is unity. Solution Comparing this with the given equation of Charpit's form, the complete integral is given by z = ax + by + ab .................(ii) ab − a − b Rearranging equation (ii) in order that outlines the intercepts, ax + by − z = ax −ab ab−a−b x −b ab−a−b + y −a ab−a−b + + by −ab ab−a−b z ab ab−a−b −ab ab − a − b − z −ab ab−a−b =1 = 1.....................(iii) −b −a From equation (iii), the x intercept is given by: ab−a−b , y- intercept by ab−a−b and ab z-intercept ab−a−b . The sum of the intercepts is given by z−a ab −b − a + ab −b + + = =1 ab − a − b ab − a − b ab − a − b ab − a − b (iv) Find the C.F and S.I. of 4xyz = pq + 2px2 y + 2qxy 2 ......................(i) Solution This equation can only be compared to Charpit's form if it is transformed rst, Tranforming equation (i) to Charpit's form δz δz Recall p = δx , q = δy But 4xy = (2x) (2y) ............................(ii) Divide (i) by (ii), 25 1 δz 1 δz 2x2 y δz 2xy 2 δz z= · · + + 2x δx 2y δy 4xy δx 4xy δy 1 δz 1 δz 1 δz 1 δz 2 2 · · +x · +y · z= 2x δx 2y δy 2x δx 2y δy Let 2xδx = δX and 2yδy = δY let x2 = X and y 2 = Y z= p= δz ,q δX = δz δX δz δY +X δz δX +Y δz δY δz δY Z = pq + pX + qY which is of Charpit's form Comparing this with the given equation of Charpit's form, the complete integral is given by z = aX + bY + ab.................(iii) But x2 = X and y 2 = Y Replacing yields; z = ax2 + by 2 + ab.................(iv) For the singular integral,we dierentiate (iv) partially with respect to a and b to obtain equations (v) and (vi) 0 = x2 + b =⇒ b = −x2 ........................(v) 0 = y 2 + a =⇒ a = −y 2 ...........................(vi) z = −y 2 x2 − x2 y 2 + x2 y 2 .................(vii) z = −y 2 x2 ........(viii) 1 (v) Show that the complete integral of the equation z = px+qy+(p2 + q 2 + 1) 2 ..........(i)represents all planes at a unit distance from the origin. Solution This equation is of Chaurit's form, thus 26 z = ax + by + a2 + b2 + 1 12 ............................(ii) Rearranging equation (ii) in order that outlines the intercepts, ax + by − z = − a2 + b2 + 1 21 ............(iii) 1 Dividing through by − (a2 + b2 + 1) 2 , we get ax − (a2 + b2 + 1) x 1 2 + a 1 −(a2 +b2 +1) 2 by − z 1 = 1............(iv) − (a2 + b2 + 1) − (a2 + b2 + 1) 2 y z + + = 1............(v) 1 b 1 2 1 1 (a2 +b2 +1) 2 −(a2 +b2 +1) 2 Squaring the intercepts and adding them to get the distance, 2 b2 1 a2 + b 2 + 1 a + + = =1 a2 + b 2 + 1 a2 + b 2 + 1 a2 + b 2 + 1 a2 + b 2 + 1 Case 11: A dierentail equation with only p,q,z in the form: f (p, q, z) = 0..................................(i) Solution Apply Charpit's auxillary equation dq dz dx dy dp = = 0 = = 0 0 0 0 0 + pf (p, q, z) 0 + qf (p, q, z) pf (p, q, z) −f (p, q, z) −qf (p, q, z) pf 0 dp dq dz dx dy = 0 = 0 = = 0 0 (p, q, z) qf (p, q, z) pf (p, q, z) −f (p, q, z) −qf (p, q, z) dp dq = δf δf p δz q δz dp dq = p q lnp = lnq + lna p = qa or q = pa.................................(ii) dz = pdx + qdy dz = p (dx + ady) = pd (x + ay) ...............................(iii) Letu = x + ay 27 dz = pdu p= dz ............................(iv) du Thus by equation (ii), q=a dz ............................(v) du Replacing back in equation (i) f dz dz , a , 0 = 0......................(vi) du du Equation (vi) is a p.d.e of rst order. Solving (vi), we get z as a function of U. The Complete Integral is then obtained by replacing u with x+ay. Example: a) Find the Complete Integral and the Singular Integral of q 2 = z 2 p2 (1 − p2 ) .....................(i) Solution Step 1: dz dz We take p = du ,q = a du and u = x + ay and substitute in ! (i) dz a du 2 2 2 dz dz =z 1− du du " 2 # dz a2 = z 2 1 − du 2 2 2 a =z −z 2 dz du 2 2 dz z 2 − a2 = du z2 r √ dz z 2 − a2 z 2 − a2 = = du z2 z ˆ ˆ z √ du = dz 2 z − a2 Integrating by substitution yields, u+b= √ 28 z 2 − a2 Squaring both sides, (u + b)2 = z 2 − a2 Replacing u = x + ay (x + ay + b)2 = z 2 − a2 ..............................(ii) Dierentiatiate (ii) with respect to a 2y (x + ay + b) = −2a..................................(iii) Dierentiatiate (ii) with respect to b 2 (x + ay + b) = 0..................................(iv) From (iii) and (iv) we get, z 2 = 0 =⇒ z = 0.............................(v) Equation (v) is an xy plane. (Incomplete) 29 5 SURFACES AND CURVES IN THREE DIMENSIONS: In rectangular co-ordinates, the equation of a surface is expressed in the form f (x, y, z) = 0..........................(i). For example, the equation x2 + y 2 + z 2 = a2 is the equation of a sphere centered at the origin (0, 0, 0)radius r. From dx + δf dy + δf dz = 0showing that equation (i) the total dierential of F is δf δx δy δz δf δf δf the triad δx , δy , δz are the directional rations normal to the surface at a given point while (dx, dy, dz)are the direction ratios tangent to the surface of the same point. Consider a curve which is the intersection of the surfaces f (x, y, z)and g (x, y, z) = 0. With the two equations of surfaces satisfying their total dierentials i.e. df = 0and dg = 0or δf δf δf dx + dy + dz = 0 δx δy δz δg δg δg dx + dy + dz = 0 δx δy δz which on solving for dx, we have; δf δf δf dx + dy = − dz δx δy δz δg δg δg dx + dy = − dz δx δy δz 4= δf δx δg δx δf δy δg δy δf δg δf δg = − =J δx δy δy δx δf δy δg δy − δf dz δz dx = − δg dz δz δf δx δg δx δf δy δg δy δz = J 30 δf δy δg δy δf δz δg δz f,g x,y f, g x, y or dx = f,g x,y δzJ dx dz or = .....................(A) f,g f,g J x,y J f,g J x,y y,z Next solving for dy, we have; δf δf δf dx + dy = − dz δx δy δz δg δg δg dx + dy = − dz δx δy δz δf δx δg δx 4= δf δy δg δy δf δx δg δx dy = or δf δg δf δg = − =J δx δy δy δx − δf dz δz − δg dz δz δf δx δg δx δz = δf δy δg δy J δf δz δg δz δf δx δg δx f,g x,y f, g x, y f,g δzJ x,y dy dz or = .........................(B) dx = f,g f,g f,g J x,y J z,x J x,y From A and B, J dx = f,g y,z which is a system of the form, where p=J f, g y, z J dy = f,g z,x J dz f,g x,y dy dz dx = = p q R or q = J 31 f, g z, x or R = J dz f,g x,y p= δf δy δg δy δf δz δg δz or q = δf δz δg δz δf δx δg δx or R = δf δx δg δx δf δy δg δy The reverse problem is that of nding the curves of intersection for these surface i.e. We show that the surface f (x, y, z) = c1 will involve only one arbitrary constant c1 and f (x, y, z) = c2 will involve only one arbitrary constant c2 . Therefore, the curve of intersection will involve two parameters and we call them a two parameters and we call them a two parameter family of curves. 32 6 METHODS OF SOLVING dx p = dy q = dz R =λ (i) The Spot Method ..........................(i) Suppose that U (x, y, z)is a solution of dxp = dyq = dz R δu δu δu δu δu dx + dy + dz = 0 hence λ dx + dy + dz = 0where λ 6= then δu δx δy δz δx δy δz 0hence p du + q du + R du = 0. We then try to spot functions p0 , q 0 and R0 such dx dy dz that pp0 + qq 0 + RR0 = 0implying that p0 dx + q 0 dy + R0 dz = 0will be an exact dierential equation with the solution U (x, y, z) = c1 . Repeats the same process with the function p00 , q 00 and R00 to get V (x, y, z) = c2 which is the required two parameter family of curves c1 and c2 . Example: dy dz dx = x(x+y)−az = z(x+y) 1. Find the integral curve of the equations y(x+y)+az Solution p = y (x + y) + az , q = x (x + y) − az and R = z (x + y) By intution take p0 = x, q 0 = y thus pp0 + qq 0 + RR0 will become exact; xy (x + y) + xaz + yx (x + y) − yaz + R0 z (x + y) = 0 x2 y + xy 2 + xaz + yx2 + y 2 x − yaz + R0 z (x + y) = 0 az (x + y) + R0 z (x + y) = 0 =⇒ az (x + y) = −R0 z (x + y) = 0 Thus R0 = −a Then pp0 +qq 0 +RR0 = 0is now exact implying that p0 dx+q 0 dy +R0 dz = 0will be an exact dierential equation with the solution U (x, y, z) = c1 xdx − ydy − adz = 0 On integration, x2 y 2 − − az = c or x2 − y 2 − 2az = c1 2 2 where c1 = 2c Again by intuition, take p00 = 1, q 00 = i thus pp00 + qq 00 + RR00 = 0 33 y (x + y) + az + x (x + y) − az + R00 z (x + y) = 0 (x + y)2 + R00 z (x + y) = 0 R00 = − (x + y) z Thus p00 dx + q 00 dy + R00 dz = 0will be an exact dierential equation with the solution V (x, y, z) = c2 (x + y) dz = 0 z dx + dy 1 − dz = 0 x+y z dx + dy − d (x + y) 1 − dz = 0 x+y z ln |x + y| − ln |z| = lnc2 x+y = c2 z Thus the integral curves are given by U (x2 − y 2 − 2az) = c1 and V x+y = c2 z Note: This method can be simplied and shortened by the following properties: 0 dx+q 0 dy+R0 dz dx = dy = dz = p pp = dx−dy−2dz so that if pp0 + qq 0 + RR0 = 0 then 0 +qq 0 +RR0 p q R p−q−2R p0 dx + q 0 dy + R0 dz = 0is exact. (ii) Method 2 Given that dx+3dy−dz pp0 +3qq 0 −RR0 dx p = e.t.c dy q = dz R , It should be noted that dx p = dy q Example: dy dx dz Find the integral curves of y(x+y)+az = x(x+y)−az = z(x+y) Solution Adding the rst two and equating to the third equation, dz dx + dy 2 = z (x + y) (x + y) dx + dy dz = x+y z 34 = dz R = p0 dx+q 0 dy pp0 +qq 0 = ln |x + y| = ln |z| + lnc1 x+y = c1 z Also, xdx ydy dz = = xy (x + y) + azx xy (x + y) − azy z (x + y) x2 y Therefore, + xy 2 xdx − ydy xdx − ydy = 2 2 + azx − xy − x y + azy az (x + y) xdx − ydy dz = az (x + y) z (x + y) xdx − ydy = adz x2 y 2 − = az + c or x2 − y 2 − 2az = c2 2 2 where c2 = 2c (iii) Method 3: It is possible to use the rst solution to obtain the second solution by eliminating one of the variables in one of the fractions: dy dx dz Example: Obtain integral curve of y(x+y)+az = x(x+y)−az = z(x+y) ...............(i) Solution Adding the rst two and equating to the third equation, dx + dy dz 2 = z (x + y) (x + y) dx + dy dz = x+y z ln |x + y| = ln |z| + lnc1 x+y = c1 z Eliminating z from the rst and second fraction using the rst solution, we have; z= x+y c1 35 Replacing in equation (i), dx y (x + y) + a x+y c1 = dy x (x + y) − a x+y c1 = x+y c1 dz (x + y) Using the rst two fractions, dx dy a = y + c1 x − ca1 a a x− dx = y + dy c1 c1 x2 azx y2 azy − = + +c 2 x+y 2 x+y x2 y 2 az (x + y) − − = c or x2 − y 2 − 2az = c2 2 2 (x + y) where c2 = 2c (iv) Method 4 If one of the variables is absent in one of the equations, one of the solutions can be denied easily. dz dx = dy = z+y Example: Find the integral curves of the equations x+z 2 ...............(i) y Solution dz x is absent from the equation dyy = z+y 2 and this equation is linear with z and ´ 1 the dependent variable with I.F = e − y dy = e−lny = y1 . Therefore dz z + y2 dz z = or − =y dy y dy y 1 dz z − 2 =1 y dy y z d =1 y z = y + c1 or z = c1 y + y 2 y 36 Comparing the rst and the second fractions, dx dy = 2 x + (c1 y + y ) y This is a linear equation with x as the dependent variable dx x + c1 y + y 2 = dy y dx x = + c1 + y dy y dx x − = c1 + y dy y ´ I.F = e − y1 dy = e−lny = y1 . Therefore 1 x = (y + c1 ) d y y c1 x d =1+ y y x = y + c1 lny + c2 y Replacing c1 = z−y 2 y x z − y2 =y+ lny + c2 y y Example: Find the integral curves of the equation dx dz a) xy = dy = zxy−2x 2 y2 The rst two fractions do not have the variable z dy dx = 2 xy y dx dy = x y lnx = lny + lnc =⇒ Also, 37 x = c1 y dy dz = 2 y zxy − 2x2 But x = c1 y dy dz = 2 2 y zc1 y − 2c21 y 2 dy = b) dx xz(z 2 +xy) = dy −yz(z 2 +xy) dz zc1 − 2c21 y= 1 ln zc1 − 2c21 + c2 c1 y= y xz 2x2 ln − 2 + c2 x y y = dz z4 −dy dx = x y lnx = −lny + lnc =⇒ xy = c1 Also, dx dz = 4 2 xz (z + xy) z z (z 2 + xy) dx = dz x z4 dx 1 c1 = + dz x z z3 c1 lnx = lnz − 2 + c2 2z xy lnx − lnz + 2 = c2 2z x xy + 2 = c2 z 2z dy dx dz c) x(y−z) = y(z−x) = z(x−y) Solution Adding the rst two fractions and equating with the third, 38 dx + dy dz = −z (x + y) z (x − y) dx + dy + dz = 0 Integrating, x + y + z = c1 Also, ydx + xdy dz = yx (y − z) + xy (z − x) z (x − y) dz ydx + xdy = yx (x − y) z (x − y) dx dy dz + + =0 x y z lnx + lny + lnz = lnc2 xyz = c2 d) x2 (ydx 3 −z 3 ) = Solution dy y 2 (z 3 −x3 ) x3 = (y 3 dz z 2 (x3 −y 3 ) zdz xdx + ydy = 3 3 3 3 3 3 − z ) + y (z − x ) z (x − y 3 ) Considering the denominator, x3 y 3 − x3 z 3 + y 3 z 3 − x3 y 3 = z 3 x3 − z 3 y 3 z 3 y 3 − x3 = −z 3 x3 − y 3 Therefore, zdz xdx + ydy = 3 3 3 3 3 −z (x − y ) z (x − y 3 ) xdx + ydy + zdz = 0 x2 y 2 z 2 + + = c or x2 + y 2 + z 2 = c1 2 2 2 y3 − dx x2 x3 + + dy y2 z3 − x3 39 = dz z2 x3 − y 3 dx x2 y3 + + dy y2 z3 = dz z2 x3 − y 3 dx dy dz + 2 + 2 =0 x2 y z 1 1 1 − − − =c x y z 1 1 1 + + = c2 x y z 40 7 ORTHOGONAL TRAJECTORIES OF CURVES: p= δf δy δg δy δf δz δg δz ,q = δf δz δg δz δf δx δg δx ,R = δf δx δg δx δf δy δg δy In three dimensions, given a surface f (x, y, z) = 0..............................(i)and a system of curves, we have to nd a system of curves each of which lies on the surface (i) and cuts every curve of the given system at right angles. The new system of curve is called the system of orthogonal trajectories on the surface of the given system of curves. The original system of curves may be thought of as the intersection of the surface (i) with the parameter family of curves g (x, y, z) = c1 ..............................(ii) 41 In general, the tangential direction (dx, dy, dz)to the given curve through the points x,y,z on the surface (i) satises the equations δf dx + δf dy + δf dz = 0 δx δy δz δg δg and δx dx+ δy dy+ δg dz = 0. Which on solving gives dx = dy = dz .............(iii) δz p q R where δf δf δf δf fy fz fz fx fx fy δx δz δx = ,q = δg δg = ,R = δg δy = p= δg gy gz gz gx gx gy δz δx δx δy The curve through point (x, y, z)of the orthogonal system has tangential direction (dx0 , dy 0 , dz 0 )which lies on the surface (i) such that δf dx0 + δf dy 0 + δx δy δf dz 0 = 0...................(iv) δz δf δy δg δy δf δz δg δz 42 This implies that dxdx0 + dydy 0 + dzdz 0 = 0 Hence from (iii), we have pdx0 + qdy 0 + Rdz 0 = 0...........................(v) Solving (iv) and (v) we have, 0 0 dx0 = dy = dz .........................(vi)where p0 = Rfy − qfz , q 0 = pfz − Rfx and p0 q0 R0 R0 = qfx − pfy .........................(vii) Note that: δf dx0 δx 0 + δf dy 0 = − δf dz 0 δy δz pdx + qdy 0 = −Rdz 0 dx0 = fz δz 0 fy −Rδz 0 q = fx fy p q dx0 fz fy −R q δz 0 fx fy p q = fx fy p q dz 0 fx fy p q 43 fx fz δz 0 p −Rδz 0 dy 0 = δz 0 = fx fy p q dy 0 fx fz p −R fx fy p q dz 0 = fx fy p q fx fy p q Example: Find the orthogonal trajectories of the sphere x2 + y 2 + z 2 = a2 at its intersection with a parabola xy = cz where c is a parameter. Solution Let f (x, y, z)be identiacal to x2 + y 2 + z 2 = a2 Let g (x, y, z) = xyz = c Then (dx, dy, dz)satises 2y x z ,q = and R= dx p 2z 2xy 2 = − − 2x = z2 − xy 2 z δf δz δg δz δf δx δg δx δf δx δg δx δf δy δg δy = = = dy q = dz R where p = δf δy δg δy δf δz δg δz = fy fz = gy gz −2xy 2 −2xz 2 z2 fz fx 2z 2x = = y gz gx − xy z2 z 2yz z + 2x2 y z2 = 2yz 2 +2x2 y z2 = 2y (z 2 +x2 ) z2 fx fy 2x 2y 2 2 2(x2 −y 2 ) = y x = 2x −2y = z z gx gy z z dx dy dz 2 2 = 2y(z2 +x2 ) = 2(x2 −y2 ) −2x y z+z 2 z2 z dx dy dz = = 2 2 2 2 2 −x (y + z ) y (z + x ) z (x − y 2 ) Then the triad (dx0 , dy 0 , dz 0 )satises p0 = dx0 p0 = dy 0 q0 = dz 0 R0 where fy fz 2y 2z = = 2yz (x2 − y 2 ) − 2zy (x2 + z 2 ) = 2 2 2 q R y (x + z ) z (x − y 2 ) 44 −2yz (y 2 + z 2 ) fz fx 2z 2x q0 = = = −2zx (y 2 + z 2 )−2xz (x2 − y 2 ) = 2 2 2 2 R p z (x − y ) −x (y + z ) 2 2 −2xz (x + z ) fx fy 2x 2y R0 = = = 2xy (x2 + z 2 )+2xy (y 2 + z 2 ) = 2 2 2 p q −x (y + z ) y (x + z 2 ) 2xy (x2 + 2z 2 + y 2 ) Hence, dx0 dy 0 dz 0 = = −2yz (y 2 + z 2 ) −2xz (x2 + z 2 ) 2xy (x2 + 2z 2 + y 2 ) dx0 dy 0 dz 0 = = −yz (y 2 + z 2 ) −xz (x2 + z 2 ) xy (x2 + 2z 2 + y 2 ) For the rst two fractions, dy 0 dx0 = −yz (y 2 + z 2 ) −xz (x2 + z 2 ) dx0 dy 0 = y (y 2 + z 2 ) x (x2 + z 2 ) ydy 0 xdx0 = 2 (y 2 + z 2 ) (x + z 2 ) From;x2 + y 2 + z 2 = a2 ,x2 + z 2 = a2 − y 2 and y 2 + z 2 = a2 − x2 xdx0 ydy 0 = (a2 − x2 ) (a2 − y 2 ) 1 1 − ln a2 − x2 = − ln a2 − y 2 + lnk 2 2 a2 − y 2 = k −2 = c a2 − x 2 −y 2 2 2 = a2 are the Thus aa2 −x 2 = cand the given surface equation is x + y + z systems of orthogonal trajectories. b) Find the orthogonal trajectories on the cone x2 + y 2 = z 2 tan2 αat its intersection with the family of planes z = c Solution 2 2 45 The triad (dx, dy, dz)satises the equation p= ,q = R= δf δy δg δy δf δz δg δz δf δx δg δx δf δz δg δz δf δx δg δx δf δy δg δy dx p = = dz R = fy fz 2y −2ztan2 α = = 2y gy gz 0 1 = fz fx −2ztan2 α 2x = = −2x and gz gx 1 0 = fx fy 2x 2y = =0 gx gy 0 0 dx dy dz dx dy dz = = =⇒ = = 2y −2x 0 y −x 0 Then the triad (dx0 , dy 0 , dz 0 )satises dx0 p0 = dy 0 q0 = p0 = fy fz 2y −2ztan2 α = = −4xztan2 α q R −2x 0 q0 = fz fx −2ztan2 α 2x = = −4yztan2 α R p 0 2y R0 = dy q dz 0 R0 where fx fy 2x 2y = = −4 (x2 + y 2 ) p q 2y −2x dy 0 dz 0 dx0 = = −4xztan2 α −4yztan2 α −4 (x2 + y 2 ) dy 0 dz 0 dx0 = = xztan2 α yztan2 α (x2 + y 2 ) Comparing the rst two fractions, dx0 dy 0 = x y lnx = lny + lnc x =c y Therefore the required system of orthogonal trajectories are given by x2 +y 2 = z 2 tan2 αand xy = c. c) Find the orthogonal trajectories on the conicoid (x + y) z = 1of the conics in which its cut by the system of planes x − y + z = k where k is a parameter. 46 Solution The triad (dx, dy, dz)satises the equation dx p = dy q = dz R f = xz + yz − 1 = 0 g =x−y+z−k δf δf fy fz z x+y δy δz p = δg δg = = =z+x+y gy gz −1 1 δy δz ,q = R= δf δz δg δz δf δx δg δx δf δx δg δx δf δy δg δy = fz fx x+y z = = x + y − z and gz gx 1 1 fx fy z z = = −z − z = −2z gx gy 1 −1 dy dz dx dy dz dx = = =⇒ = = x+y+z x+y−z −2z y −x 0 = Then the triad (dx0 , dy 0 , dz 0 )satises dx0 p0 = dy 0 q0 = dz 0 R0 where p0 = fy fz z y+x = = −2z 2 − (x + y) (x + y − z) q R x + y − z −2z q0 = fz fx x+y z = = (x + y) (x + y + z) + 2z 2 R p −2z x + y + z R0 = −2z 2 fx fy z z = = z (x + y − z) − z (x + y + z) = p q x+y+z x+y−z dy 0 dz 0 dx0 = = −2z 2 − (x + y) (x + y − z) (x + y) (x + y + z) + 2z 2 −2z 2 Adding the rst two fractions, dx0 + dy 0 dz 0 = −2z 2 − (x + y) (x + y − z) + (x + y) (x + y + z) + 2z 2 −2z 2 dx0 + dy 0 dz 0 = 2z (x + y) −2z 2 dz 0 dx0 + dy 0 = (x + y) −z ln |x + y| = −ln |z| + lnc 47 (x + y) z = c If c = 1, (x + y) z = 1 Using x + y = z1 in the rst and the third fractions, dx0 dz 0 = −2z 2 − (x + y) (x + y − z) −2z 2 dx0 −2z 2 − z1 1 z −z = dz 0 −2z 2 dz 0 dx0 = −2z 2 −2z 2 − z12 + 1 1 dz 0 1 1 0 0 2 dx = −2z − 2 + 1 = dz 1 + 4 − 2 −2z 2 z 2z 2z On integrating, x=z− 1 1 + +c 3 6z 2z d) Find the orthogonal trajectories on the surface f = x2 +y 2 +2p (yz)+d = 0 of its intersections with the one family of planes parallel toz = c whre c is a parameter. 48 8 METHOD OF SEPARATION OF VARIABLES: Example 1: Displacement of vibrating spring or wave equation. The partial dierential equation governing the motion of the vibrating string 2 2 ension is given as δδt2u = c2 δδxu2 where c2 = TM ass Therefore our task is to determine u (x, t)under the boundary conditions u (0, t) = 0,u (1, t) = 0 and the initial conditions u (x, 0) = φ(x)and ut (x, 0) = Ï•(x). Here u (x, t)represents displacement of the string at any position and at any time t. u(t)represents the velocity of the wave produced by the string at time t. By the method of separation of variables, the dependent variable u (x, t)may be expressed as u (x, t) = X(x)T (t)where X(x)is a function of x alone and T (t)is a function of t alone. Then it follows that uxx = X 00 (x)T (t)andutt = X 00 (x)T (t) Given, 2 δ2u 2δ u = c ....................(i) δt2 δx2 49 Let U = XT........................(ii) δX δT δX δu = T +X = T = X 0 T.........................(iii) δx δx δx δx δ2u δX 0 δT δ2X δ 0 (X T ) = T + X = = T = X 00 T.............................(iv) 2 2 δx δx δx δx δx Similarly, δ2u = T 00 X.........................(v) δt2 Substituting (v) and (iv) into (i) XT 00 = c2 X 00 T c2 X 00 T 00 = =k T X These two functions could only be equal to a constant, say k as shown above, T 00 = k =⇒ T 00 − kT = 0 T and k c2 X 00 = k =⇒ X 00 − 2 X = 0 X c Assuming that we need to consider only the physical solution, there are three cases to be be investigated. Case 1: k=0 If k=0, then T 00 = 0, T 0 = A,T = At + B and X 00 = 0, X 0 = C ,X = Cx + D Therefore, U (x, t) = X(x)T (t) =⇒ U (x, t) = (Cx + D) (At + B) These solutions cannot describe the vibtrating of a system because it is not periodic. Applying U (0, t) = 0,U (1, t) = 0 D (At + B) = 0 Dt = 0, t 6= 0 hence D = 0 U (x, t) = Cx (At + B) 0 = Ct 50 hence C = 0 This shows that U (x, t) = 0has trivial solutions. Case 2:k = λ2 Then the two dierential equations are T 00 = λ2 =⇒ T 00 − λ2 T = 0 T and c2 X 00 λ2 = λ2 =⇒ X 00 − 2 X = 0 X c T = Aeλt + Be−λt λ λ X = Ce c x + De− c x Thus U (x, t) = Ce λ x c + De − λc x Aeλt + Be−λt This solution cannot describe the oscillating motion of a system because it is not periodic. THerefore we reject the solution; U (0, t) = (C C = −D, D = −C λ +λD) t =⇒ λC + D = 0 =⇒ λ λ x −cx x −cx x − λc x c c c t=C e −e t =⇒ t 6= 0, e − e 6= U (x, 0) = Ce − Ce 0, C = 0hence the solution is trivial Case 3:k = −λ2 Then the two dierential equations are T 00 = λ2 =⇒ T 00 + λ2 T = 0 T and c2 X 00 λ2 = λ2 =⇒ X 00 + 2 X = 0 X c T = Acosλt + Bsinλt λ λ X = Ccos x + Dsin x c c λ λ U (x, t) = Ccos x + Dsin x [Acosλt + Bsinλt] c c 51 . However, we need to determine This solution is periodic with a period of 2π λ λ, A, B, C, andD from the rst boundary condition. U (0, t) = 0 =⇒ 0 = C (Acosλt + Bsinλt) =⇒ 0 = CT, T 6= 0 =⇒ C = 0 Hence U (x, t) = Dsin λc x [Acosλt + Bsinλt] The second boundary condition gives U (l, t) = 0 =⇒ 0 = Dsin λc l [Acosλt + Bsinλt] or 0 = Dsin λc l · T l 6= 0and we cannot choose D = 0because it gives a trivial solution. Solution sin λlc = 0, only possible solution is when λlc = nπ =⇒ λ = nπ c or λn = l nπ c, n = 1, 2, 3, ... l λn is called the eigen solution. Then, we therefore have innite number of λn for which the product solution of the wave equation exists./ Therefore we write the solution as Un (x, t) = sin λn x (An cosλnt + Bn sinλnt) l or nπ nπc nπc x An cos t + Bn sin t l l l where An and Bn are denes as fourier coecients. To obtain An , we use the Un (x, t) = sin initial displacement conditions U (x, 0) = φ(x) = ∞ X An sin n=1 nπ x l This series can be organized as the half range since the expansion of a periodic function φ(x)is denes in the range (0, l). Now An can be obtained by multiplying the series by sin nπl xand integrating with respect to x from 0 to ´ l which gives An = π2 0l φ(x)sin nπl xdx ∞ a0 X F (x) = + (an cos(nx) + bn sin(nx)) 2 n=1 To determine Bn a, we use the fourth condition U (x, 0) = ψ(x, t) = ∞ n X nπ nπc nπc nπc nπc o sin x − An sin t+ Bn cos t l l l l l n=1 52 ψ(x, t) = ∞ X nπc l n=1 Bn sin nπc x l Then multiplying both sides of this equation by sin nπl xand integrating with respect to x, we get ˆ Bn = 2 nπc l ψ(x)sin 0 nπ xdx l Therefore the equation : nπ nπc nπc x An cos t + Bn sin t l l l is the required solution of the wave equation with An and Bn as shown above. Un (x, t) = sin Example: a) A string is stretched and fastened to two points. The motion is started by displacing the string in the form y (x, t) = asin πxl from which it is released at time t = 0. Show that the displacement of any point at y (x, t) = asin πx l Solution cos πct l The equation of a vibration of a string is given by: 53 2 δ2u 2δ u = c ....................(i) δt2 δx2 Boundary conditions are y (0, t) = 0and y (l, t) = 0........................(ii) The string is in the shape of y (x, t) = asin πxl .........................(iii)at rest i.e whent = 0. Also dy =0 dt The objective isto solve equation (i) subject to conditions given by equations (ii) and (ii). As discussed in the previous section the solution to(i) is given by Y = XT Y (x, t) = (a1 coskx + a2 sinkx) (a3 coskct + a4 sinkct) .........................(iv) Using the rst boundary condition in equation (ii) 0 = (a1 cos0 + a2 sin0) (a3 coskct + a4 sinkct) 0 = a1 (a3 coskct + a4 sinkct) Therefore a1 = 0since the solution must be true for all values of t, hence equation (iv) reduces to Y (x, t) = a2 sinkx (a3 coskct + a4 sinkct) Y (x, t) = sinkx (a2 a3 coskct + a2 a4 sinkct) Let a2 a3 = b1 and a2 a4 = b2 Y (x, t) = sinkx (b1 coskct + b2 sinkct) .....................(v) Using equation (ii) in (v),y (l, t) = 0 0 = sinkl (b1 coskct + b2 sinkct) sinkl = 0 k= Using (vi) in (v), Y (x, t) = sin Using nπ .........................(vi) l nπ nπ nπ x b1 cos ct + b2 sin ct .....................(vii) l l l = 0 at t= 0 nπc nπ nπ nπc nπ 0 = sin x −b1 sin ct + b2 cos ct .....................(viii) l l l l l dy dt 54 nπc nπ x 0 + b2 ...................(ix) l l Since the answer must be true for all values of x, b2 = 0 and equation (vii) 0 = sin becomes; nπ nπ x b1 cos ct l l nπ nπ Y (x, t) = b1 sin xcos ct....................(x) l l Applying the initial conditiont = 0 in (x), nπ nπ Y (x, 0) = b1 sin xcos(0) = b1 sin x.......................(xi) l l Comparing equation (xi) with te givenequation a = b1 , n = 1 πx πct cos Y (x, t) = asin ....................(xii) l l Y (x, t) = sin b) A tightly stretched spring with xed ends at x = 0and x = l is initially in equilibrium position. It is vibrating by giving each of the poins an initial 3 πx = bsin . Find the dispacement Y (x, t). velocity, Y (x, 0) = 0, ∂y ∂t t=0 l Solution 55 Equation of a vibrating string is 2 δ2u 2δ u = c ....................(i) δt2 δx2 We solve equation (i) subject to the bondary conditionsy (0, t) = 0................(ii)and y (l, t) = 0........................(iii) Initial conditions are y (x, 0) = 0................(iv)and The solution to equation (i) is of the form ∂y ∂t t=0 = bsin3 πx l ....................(v). Y (x, t) = (a1 coskx + a2 sinkx) (a3 coskct + a4 sinkct) .........................(vi) Using condition (ii), 0 = (a1 cos0 + a2 sin0) (a3 coskct + a4 sinkct) 0 = a1 (a3 coskct + a4 sinkct) ..................(vii) Therefore a1 = 0since the solution must be true for all values of t, hence equation (vii) reduces to Y (x, t) = a2 sinkx (a3 coskct + a4 sinkct) Y (x, t) = sinkx (a2 a3 coskct + a2 a4 sinkct) Let a2 a3 = b1 and a2 a4 = b2 Y (x, t) = sinkx (b1 coskct + b2 sinkct) .....................(viii) Using equation (ii) in (v),y (l, t) = 0 0 = sinkl (b1 coskct + b2 sinkct) sinkl = 0 k= nπ .........................(ix) l Using (ix) in (viii), nπ nπ nπ x b1 cos ct + b2 sin ct .....................(x) l l l Using initial condition, y (x, 0) = 0 nπ 0 = sin x (b1 cos0 + b2 sin0) .....................(xi) l Y (x, t) = sin 56 nπ x (b1 ) ...................(xii) l Since the answer must be true for all values of x, b1 = 0 and equation (x) 0 = sin becomes; nπ nπ Y (x, t) = sin x b2 sin ct l l nπ nπ Y (x, t) = b2 sin xsin ct....................(xiii) l l where n=1,2,3,... Therefore the sum of this solution is also a solution by superposition principle of linear solutions ∞ Y (x, t) = X An sin n=0 nπ nπ xsin ct....................(xiv) l l Dierentiating (xiv) with repect to t ∞ δy X nπc nπ nπc = An sin xcos t δt l l l n=0 3 πx = bsin Applying the initial condition in ∂y , ∂t t=0 l ∞ X nπc nπ nπc δy = An sin xcos (0) δt t=0 n=0 l l l bsin 3 πx l = ∞ X n=0 An nπc nπ sin x..............(xv) l l 3 From trigonometric identities sin3θ = 3sinθ − 4sin θ πx πx 3πx = 3sin − 4sin3 l l l 3 πx 1 3πx 3 πx = sin − sin sin l 4 l 4 l X ∞ πx 1 3 3πx nπ nπc − bsin bsin = An sin x 4 l 4 l l l n=0 πx 1 3πx πc π 2πc 2π 3πc 3π 4πc 4π 3 bsin − bsin = 0+A1 sin x+A2 sin x+A3 sin x+A4 sin x+..... 4 l 4 l l l l l l l l l sin 57 Equating coecient of sin πx l 3 πc 3bl b = A1 =⇒ A1 = 4 l 4πc Equating coecient of sin 2πl x Equating coecient of sin 2πc =⇒ A2 = 0 0 = A2 l 3πx l 1 3πc −bl − b = A3 =⇒ A3 = 4 l 12πc Equating coecient of sin 4πl x 0 = A4 4πc =⇒ A4 = 0 l Substitututing A1 , A2 , A3 , A4 into equation (xvii) Y (x, t) = 3bl πx πct bl 3πx 3πct sin sin − sin sin 4πc l l 12πc l l Equation (vii) represents displacement of the wave. 58 9 DERIVATION OF WAVE EQUATIONS The wave equation is derived with respect to travelling transverse waves on a taut string. The following assumptions are made: (i) The amplitude of the wave is always small. (ii) the string has a constant linear mass density (iii) The string is under a constant tension of magnitude T. (iv) The string is not made of atoms or molecules. It is compresses of continuous non-quantized matter. (v) There is no damping(no loss of energy due to friction) (vi) Gravity is ignored in the treatment. INTRODUCTION Through out the derivation, we will focus on a tiny bity of strings, so small that we can pretend that it is almost straight line. The bit of a string which will have a tension pulling at each end(with magnitude T ). Length of the bit of string will be 4s. Assuming that the amptitude of the wave is extremely small, then the ap59 proximation of the string length is 4x. Dening a linear mass density as the mass of the string in kilograms per meter of string to be µ, then the mass of bit of string is given by: m = µ4x.........................................(i). 9.1 Forces on the bit of the string Since we are looking at transverse travelling waves, we worry about the net force in the vertical direction. X Fnet = Fy Note that: Deing vertical coordinates above, onm the string to be positive. The net force on the bit of the string is: Fnet = T sin (θ2 ) − T sin (θ1 ) Converting sine functons into tangent functions, 60 Since we are dealing only with small amplitudes, we use this approximation 4x = 4s. Therefore, sinθ = tanθ Then the new net force is expressed as Fnet = T tan (θ2 ) − T tan (θ1 ) 4y looking at small and smaller Using tan is more convenient because tanθ = 4x δy bits of string letting 4xapproach zero then tanθ = δx (tangent to the string at a point) Letting the string bit smaller smaller, the net force is and Fnet = T δy δx − R δy δx Employing Newton's Second Law ...........................(ii) Looking at the actual motion of the bit of string. Newton's second law tells us that Fnet = ma...............(iii) where Fnet is the net force, m is the mass while a is the acceleration. Puting vertical acceleration as the second time 61 derivative of vertical displacements Fnet = m δ2y δt2 ............................(iv) Substituting values of mass from (i)in equation (iv) Fnet = µ4x Equating equation (v) to (ii) T " T δy δx 2 − 4x δ2y δt2 2 δy δ y = µ4x δx 1 δt2 2 δ y =µ ............................(vi) δt2 δy − δx 2 # δy δx 1 ............................(v) From denition of a derivative f (x + 4x) − f (x) δf = lim δx 4x→0 4x If 4xgets smaller and smaller, equation (vi) becomes: T δ2y δx2 =µ δ2y δt2 ............................(vii) δ y δ y where T (Tension)and µ (mass)are constants. δx 2 and δt2 are second derivatives of vertical displacement. Writing both constants as a single constant, 2 δ2y δx2 c2 = δ2y δt2 2 ........................(viii) where c2 = Tµ . Equation (viii) is a p.d.e since vertical diplacement depends on both position (x) and time (t) 10 HEAT EQUATION Consider the heat conduction equation in a thin metal bar of legth l with insulated sides. Let us assume that the ends are x = 0and x = l and 62 are hels at temperature 00 cfor all timet > 0. In addition, let us assume that the temperature distribution at t=0 is U (x, 0) = f (x). Determine the temperature distribution in a bar of some subsequent time t > 0. Solution The heat equation is given by Ut = αUxx i.e. δu δ2u =α 2 δt δx where U (x, t)is the heat distribution subject to boundary conditions U (0, t) = 0and U (1, t) = 0and the initial conditions U (x, 0) = f (x). By method of separation of variables, Let U = XT........................(ii) δX δT δX δu = T +X = T = X 0 T.........................(iii) δx δx δx δx δ2u δ δX 0 δT δ2X 0 = (X T ) = T + X = T = X 00 T.............................(iv) δx2 δx δx δx δx2 Similarly, δu = T 0 X.........................(v) δt Substituting (v) and (iv) into (i) T 0 X = αX 00 T T0 X 00 = =k αT X These two functions could only be equal to a constant, say k as shown above, T0 = k =⇒ T 0 − αkT = 0 αT and X 00 = k =⇒ X 00 − kX = 0 X Assuming that we need to consider only the physical solution, there are three cases to be be investigated. Case 1: k=0 If k=0, then T 0 = 0, T = Aand X 00 = 0, X 0 = C ,X = Cx + D 63 Therefore, U (x, t) = X(x)T (t) =⇒ U (x, t) = (Cx + D) (A) = CAx + DA = Ex + F These solutions cannot describe the vibtrating of a system because it is not periodic. Applying U (0, t) = 0,U (1, t) = 0 F = 0 T 6= 0 hence F = 0 U (x, t) = Ex X 6= 0 hence E = 0 This shows that U (x, t) = 0has trivial solutions. Case 2:k = λ2 Then the two dierential equations are T 0 X = αX 00 T X 00 T0 = = λ2 αT X These two functions could only be equal to a constant, say k as shown above, T0 = λ2 =⇒ T 0 − αλ2 T = 0 αT δT δT 2 = αλ2 T =⇒ = αλ2 δt =⇒ lnT = αλ2 t + A ⇒ T = Aeαλ t δt T and X 00 = k =⇒ X 00 − λ2 X = 0 X T = Aeαλ 2t X = Ceλx + De−λx Thus 2 U (x, t) = Ceλx + De−λx Aeαλ t 2 U (x, t) = Beλx + F e−λx eαλ t but when t increases the temperature in this case increases too much instead 64 of decreasing i.e. physical condition does not allow. Therefore we reject the solution; Case 3 k = −λ2 Then the two dierential equations are T 0 X = αX 00 T T0 X 00 = = −λ2 αT X These two functions could only be equal to a constant, say k as shown above, T0 = λ2 =⇒ T 0 + αλ2 T = 0 αT δT δT 2 = −αλ2 T =⇒ = −αλ2 δt =⇒ lnT = −αλ2 t + A ⇒ T = Ae−αλ t δt T and X 00 = λ2 =⇒ X 00 + λ2 X = 0 X T = Ae−αλ 2t X = Ccosλx + Dsinλx Thus U (x, t) = (Ccosλx + Dsinλx) Ae−αλ U (x, t) = (Bcosλx + F sinλx) e−αλ 2t 2t This is quite possible, applying the boundary conditions a) U (0, t) = 0 2t 0 = (B + 0) e−αλ 2t 0 = Be−αλ B=0 U (x, t) = (F sinλx) e−αλ b) U (l, t) = 0 65 2t 0 = (F sinλl) e−αλ 2t Here, F 6= 0, e−αλ t = 6 0. Thus, 2 sinλl = 0 λl = nπ =⇒ λ = Then nπ l nπ −α( nπl )2 t U (x, t) = F sin x e l λn are dened to be the eigen values and the function Un (x, t)corresponds to any λn is called the eigen function. Thus the general solution becomes nπ 2 nπ Un (x, t) = Fn sin x e−α( l ) t l From the superposition principle all the linear solutions, we get the general solutions as ∞ ∞ U (x, t) = X Un (x, t) = X n=1 n=1 nπ −α( nπl )2 t Fn sin x e l This can be recognized as the fourier half range sine series and Fn is the fourier coecient which can be determined from the initial condition as follows: ∞ U (x, t) = X Fn sin n=1 nπ x l Multiplying both sides by sin nπl xand integrating, we obtain: 2 Fn = l f (x) = ˆ l f (x)sin 0 nπ xdx l ∞ X a0 nπx nπx + an cos + bn sin 2 l l n=1 Assignment 1. Use the method of separation of variables to solve the heat equation. 66 2. By stating the necessary assumption show that the one dimensional wave 2 δ2 y µ T ension 2 equation is given by δδt2y = c2 δx . 2 where c = m = M ass 3. Derive the heat equation. Example 2: A rod whose surface is measured has a length of 3 units. The end of the rod is kept at 00 cand its initial temperature at any point x, 0 < x < 3is given by U (x, 0) = 5sin4πx − 3sin8πx + 2sin10πx. Find the temperature at any given time t. Solution Let U (x, t)be temperature at any given time t. The boundary conditions are x = 3and x = 0 Initial conditions are U (x, 0) = 5sin4πx − 3sin8πx + 2sin10πx. The heat equation is one dimension δ2u δu =α 2 δt δx Using case 3 wherek = −λ2 since case 1 and case 2 give trivial solutions; Then the two dierential equations are T 0 X = αX 00 T X 00 T0 = = −λ2 αT X These two functions could only be equal to a constant, say k as shown above, T0 = λ2 =⇒ T 0 + αλ2 T = 0 αT δT δT 2 = −αλ2 T =⇒ = −αλ2 δt =⇒ lnT = −αλ2 t + A ⇒ T = Ae−αλ t δt T and X 00 = λ2 =⇒ X 00 + λ2 X = 0 X T = Ae−αλ 2t X = Ccosλx + Dsinλx 67 Thus U (x, t) = (Ccosλx + Dsinλx) Ae−αλ U (x, t) = (Bcosλx + F sinλx) e−αλ 2t 2t This is quite possible, applying the boundary conditions a) U (0, t) = 0 2t 0 = (B + 0) e−αλ 2t 0 = Be−αλ B=0 U (x, t) = (F sinλx) e−αλ 2t b) U (3, t) = 0 0 = (F sinλ3) e−αλ 2t Here, F 6= 0, e−αλ t = 6 0. Thus, 2 sinλ3 = 0 λl = nπ =⇒ λ = Then nπ 3 2 nπ −α( nπ t 3 ) U (x, t) = F sin x e 3 λn are dened to be the eigen values and the function Un (x, t)corresponds to any λn is called the eigen function. Thus the general solution becomes nπ 2 nπ Un (x, t) = Fn sin x e−α( 3 ) t 3 From the superposition principle all the linear solutions, we get the general solutions as ∞ ∞ U (x, t) = X n=1 Un (x, t) = X n=1 68 2 nπ −α( nπ t 3 ) Fn sin x e 3 U (x, 0) = ∞ X Un (x, 0) = n=1 ∞ X n=1 Fn sin nπ π 2π 3π 4π 5 x = F1 sin x+F2 sin x+F3 sin x+F4 sin x+F5 sin 3 3 3 3 3 3 Using the initial conditions and comparing, (a) U (x, 0) = 5sin4πx − 3sin8πx + 2sin10πx. π 2π 3π 4π 5π 5sin4πx−3sin8πx+2sin10πx = F1 sin x+F2 sin x+F3 sin x+F4 sin x+F5 sin x+... 3 3 3 3 3 F1 = 5, n3 = 4 =⇒ n = 12 = 8 =⇒ n = 24 F2 = −3, 2n 3 F3 = 2, 3n = 10 =⇒ n = 10 3 F4 = 0 F5 = 0 Therefore, 2 2 U (x, t) = 5sin4πxe−α(4π) t − 3sin8πxe−α(8π) t + 2sin10πxe−α(10π) 2 t Example: A rod of length l and with insulated side is initially at a uniform temperature U0 . It ends are suddenly cooled at 00 cand kept at that temperature. Find the temperature distribution function U (x, t). Ans: U (x, t) = U0 sin nπ l xe− c 2 n2 π 2 t l Example: A bar whose surface is insulated and has a length of 3 units and diusivity 2 units. If the ends are kept at temperature 0 units and initial temperature is U (x, 0) = 5sin4πx − 3sin8πx + 2sin10πx. Find the sistribution of temperature at any point x at any time t. Solution 69 Let the bar be placed on the x-ais such that one end is at the origin and the other end is at x = 3. The distribution of heat is generally given by Let U (x, t)be temperature at any given time t. The boundary conditions are x = 3and x = 0 Initial conditions are U (x, 0) = 5sin4πx − 3sin8πx + 2sin10πx. The heat equation is one dimension δ2u δu =2 2 δt δx Using case 3 wherek = −λ2 since case 1 and case 2 give trivial solutions; Then the two dierential equations are T 0 X = 2X 00 T T0 X 00 = = −λ2 2T X These two functions could only be equal to a constant, say k as shown above, T0 = λ2 =⇒ T 0 + 2λ2 T = 0 αT δT δT 2 = −2λ2 T =⇒ = −2λ2 δt =⇒ lnT = −2λ2 t + A ⇒ T = Ae−2λ t δt T 70 and X 00 = λ2 =⇒ X 00 + λ2 X = 0 X 2t T = Ae−2λ X = Ccosλx + Dsinλx Thus U (x, t) = (Ccosλx + Dsinλx) Ae−2λ 2t 2t U (x, t) = (Bcosλx + F sinλx) e−2λ This is quite possible, applying the boundary conditions a) U (0, t) = 0 0 = (B + 0) e−2λ 0 = Be−2λ 2t 2t B=0 U (x, t) = (F sinλx) e−2λ 2t b) U (3, t) = 0 2t 0 = (F sinλ3) e−2λ Here, F 6= 0, e−αλ t = 6 0. Thus, 2 sinλ3 = 0 λl = nπ =⇒ λ = Then nπ 3 nπ 2 nπ U (x, t) = F sin x e−2( 3 ) t 3 λn are dened to be the eigen values and the function Un (x, t)corresponds to 71 any λn is called the eigen function. Thus the general solution becomes nπ 2 nπ Un (x, t) = Fn sin x e−2( 3 ) t 3 From the superposition principle all the linear solutions, we get the general solutions as ∞ ∞ U (x, t) = X Un (x, t) = n=1 U (x, 0) = ∞ X Un (x, 0) = n=1 Fn sin n=1 ∞ X n=1 X Fn sin 2 nπ −2( nπ t 3 ) x e 3 nπ π 2π 3π 4π 5 x = F1 sin x+F2 sin x+F3 sin x+F4 sin x+F5 sin 3 3 3 3 3 3 Using the initial conditions and comparing, (a) U (x, 0) = 5sin4πx − 3sin8πx + 2sin10πx. 5sin4πx−3sin8πx+2sin10πx = F1 sin F1 F2 F3 F4 F5 n1 π n2 2π n3 3π n4 4π n5 5π x+F2 sin x+F3 sin x+F4 sin x+F5 sin 3 3 3 3 3 = 5, n31 = 4 =⇒ n1 = 12 = −3, 2n3 2 = 8 =⇒ n2 = 24 = 2, 3n3 3 = 10 =⇒ n3 = 10 =0 =0 Therefore, U (x, t) = 5sin4πxe− 288 2 π t 9 − 3sin8πxe− 2576 2 π t 9 + 2sin10πxe− 1800 (10π)2 t 9 11 DERIVATION OF THE WAVE EQUATION Let the string be elastic with linear density ρstretched to length l with tension T and xed at 0 and A then allowed to vibrate: 72 Let the tension at P and θbe T1 and T2 respectively. Horizontal T along PA is −T1 cosθ1 , Vertical T along PA is −T1 sinθ1 . Horizontal T along θB is −T2 sinθ2 , Vertical T along PA is −T2 cosθ2 . PQ does not move horizontally. This implies that T1 cosθ1 = T2 cosθ2 = T.......................(i) Now, mass of PQ is given by linear denity multiplied by length. 4s · ρ and accelLet the displacement of PQ be u1 . This implies that velocity = δu δt 2 eration is δδt2u . Then from Newton's second law of motion F = ma. 4s · ρ δ2u = −T1 sinθ1 + T2 sinθ2 δt2 4s · ρ δ2u = T2 sinθ2 − T1 sinθ1 δt2 Dividing left hand side and right hand side by T 2 4s · ρ δδt2u T2 sinθ2 T1 sinθ1 = − T T T 73 2 4s · ρ δδt2u T2 sinθ2 T1 sinθ1 = − T T2 cosθ2 T1 cosθ1 2 4s · ρ δδt2u = T anθ2 − T anθ1 T Now T anθ2 = δu |at Q (slope or gradient) δx x+4x Similarly T anθ1 = δu |at P (slope or gradient) δx x 4s · ρ δ 2 u δu δu |x+4x − |x = 2 T δt δx δx δ2u T = 2 δt ρ δu | δx x+4x − δu | δx x 4x δ2u T δ2u = δt2 ρ δx2 where 4x = 4s 12 DERIVATION OF HEAT EQUATION The diusivity of material is given by This can be expressed as α = sρk 74 T hermal conductivity Specif ic heat capacity×density of the material . Let U (x, l)be the temperature at any point x at any time t. Let one end be the origin and the bar be kept on x axis so that the other end is kept at x = l. At A consider the cross-sectional area a at planepand another plane p0 at x + 4x Let pp0 = 4x → 0. Let temperature atpq beu and atp0 q 0 be u + 4u. The 4t |at x . a is the area while k is heat ow from pq to p0 q 0 is given by −ka 4u 4x the constant of proportionality(thermal conductivity). By fouries law, the heat ux is directly proportional to the temperature gradient. 4t |at x+4x . The heat fow from p0 q 0 to the next string is given by −ka 4u 4x Therefore the heat retained is given by: ka 4u 4u 4t |x+4x −ka 4t |x 4x 4x The amount of heat retained by the small element Also, heat gained or lost frompqp0 q 0 is given by M sθwhere M is the mass (volume multiplied by density), s is the specic heat capacity while θis the temperature change. = (a4x × ρ) × s4u 75 ρas4u4x = ka 4u 4u 4t |x+4x −ka 4t |x 4x 4x With the limit 4x → 0, 4u −→ 0 in time 4t Dividing both sides by 4tthen by 4x | −ka 4u | ka 4u 4u 4x x+4x 4x x ρas = 4t 4x ka δu δu δu = |x+4x − |x δt ρas δx δx δu k δ2u = δt ρs δx2 By denition of a p.d.e zero. k ρs = c2 . c2 is a positive quantity and is not equal to δu δ2u = c2 2 δt δx 13 OTHER BOUNDARY VALUE PROBLEMS Solve the boundary value problem δu = 4 δu , where U (0, y) = 5e−7y δx δy Solution By method of separation of variables U (x, y) = X(x)Y (y) δu = X 0Y δx δu = Y 0X δy X 0 Y = 4Y 0 X X0 Y0 = =c 4X Y Taking X0 4X = cand Y0 Y =c X 0 − 4cX = 0 Y 0 − cY = 0 76 Integrating factor is given by ´ I.F = e −4cdx ´ I.F = e −cdy = e−4cx = e−cy X = Ae4cx Y = Becy Since the solution is a product of X and Y i.e.U (x, y) = X(x)Y (y) = Ae4cx (Becy ) = Dec(4x+y) If compared to the boundary conditionsU (0, y) = 5e−7y , 5e−7y = Decy c = −7, D = 5 Thus, U (x, y) = 5e−7(4x+y) Exercise: Solve the following boundary value problems: (i) δu = 2 δu + u ; U (x, 0) = 6e−3x δx δy (ii) 3 δu + 2 δu = 0 ; U (x, 0) = 4ex δx δy Answers (i) U (x, t) = 6e−3x−2t 3 (ii)U (x, t) = 4e−x− 2 y 14 METHOD OF LAPLACE TRANSFORMS AND FOURIER TRANSFORMS: − The Laplace transorm of the function f(t) is Lf (t) = f (s) = f (t) = 77 ´∞ 0 e−st f (t)dt ˆ 0 ˆ ∞ L [Ux ] = ˆ ∞ U (x, t) dt L [U (x, t)] = 0 δu −st d e dt = δx dx ˆ ∞ e−st U (x, t) dt 0 ˆ ∞ δ u −st d e−st U (x, t) dt L [Uxx ] = e dt = 2 2 δx dx 0 0 ˆ ∞ δu −st δu L [Ut ] = L = e dt δt δt 0 ∞ 2 2 u = e−st =⇒ δu = −se−st δt δv δu = =⇒ v = u δt δt ˆ ∞ δu −st ∞ L [Ut ] = L = U e |0 +s U e−st dt δt 0 − = 0 − u(0) + su − = u(0) + su 14.1 SINE AND COSINE FOURIER TRANSFORMS The sine and cosine fourier transforms can be employed when the range of integration varies from 0 to ∞. The choice of sine and cosine transform is decided by the four of the boundary conditions. Now if a sine transform is being used, we multiply the dierential equation by sinγxand integrate with respect to x from 0 to ∞.Thus the sine transform of the term Uxx can be written as follows: ˆ ∞ ˆ ∞ sinγxUxx dx = Ux sinγx |∞ 0 −γ 0 cosγxUx dx 0 The rst term of the right hand side vanishes at the lower limit through the sine term. It also vanishes at the upper limit if U (x, t)is such that Ux → ∞as x → ∞. Thus 78 ˆ ∞ sinγxUxx dx = γ [u(0, t)] − γ 2 Fs {u} ´∞ 0 where Fs {u} = 0 U (x, t)sinγxdxand U (0, t)is the value of U at x = 0. Example If U (x, t)is the temperature at time t and γ the thermal diusivity of the metal bar nd U (x, t) from the partial dierential equation. Ut = αUxx , x > 0 and t > 0with the boundary conditions U (0, t) = U0 , t > 0 and initial condition U (x, 0) = 0, x > 0 . Solution Here we use the sine transform. Thus, taking the sine transform of Ut = αUxx , x > 0 and t > 0, weˆhave: ˆ ∞ ∞ Uxx sinγxdx Ut sinγxdx = α 0 0 or − du 2 = α γU (0, t) − γ U dt − − or ddtu + αγ 2 u = αγU0 using equation (ii) which is an ordinary dierential − − equation. The soluion is given by U = Ae−αγ t + γ1 U0 at t = 0, U = 0. − Therefore U = U0 γ 2 i h 2 1 − e−αγ t . The inversion is 2U0 U (x, t) = π ˆ ∞ 1 − e−αγ 0 2t sinγx γ dx 15 FURTHER EXAMPLES ON METHOD OF SEPARATION VAIABLES 1. Consider Laplace's equation Uxx +Uyy = 0..........................(i)where U (x, y)epresents the velocity potential of a uid particle in a certain domain. For example we need to determine U (x, y)inside a unit circle x2 + y 2 < 1, when its values on the circumference x2 + y 2 = 1are described. 79 Solution This boundary value problem for Laplace's equation is dened as the dirichilet problem. In obtaining the solution of (i), we make the following transformation. x = Rcosθand y = Rsinθ. Then the laplacee's equation can be reduced to: URR + where U = U (R, θ). R2 = x2 + y 2 , θ = tan−1 xy − y2 dR x δθ x = = = cosθ , = y2 dx R δx 1+ dR dy = y R δθ = sinθ, δy = We know that x2 1 x 2 1+ y 2 x = 1 1 UR + 2 Uθθ = 0 R R −Rsinθ R2 x R2 = = − R1 sinθ Rcosθ R2 = cosθ R and δ δ δR δθ δ δ = δR + δθδ δx = cosθ δR − R1 sinθ δθδ = δx δx δy δ 1 δ sinθ δR + R cosθ δθ δ δu δ δ δ δ δ2 u = δx = δx u = cosθ δR − R1 sinθ δθδ cosθ δR − δx2 δx δx δu 1 δu δu 1 δu cosθ δR− R sinθ δθ cosθ δR − R sinθ δθ δ δR δR δy δθ + δθδ δy = 1 sinθ δθδ R u = 1 δ δu 1 δu = cosθ − sinθ cosθ − sinθ R δθ δR R δθ 1 δu δ2u 1 δ2u 1 δu δ2u 1 δu 1 = cosθ cosθ 2 + 2 sinθ − sinθ − sinθ −sinθ + cosθ − cosθ − s δR R δθ R δuδθ R δR δRδθ R δR R δ δR = cos2 θ δu 1 δu cosθ − sinθ δR R δθ δ2u 1 δu 1 δ2u 1 1 δ2u 1 δu 1 2 + cosθsinθ − sinθcosθ + sin θ− sinθcosθ + 2 sinθcosθ + 2 sin 2 2 δR R δθ R δuδθ R R δRδθ R δR R 80 2 δ2u 2 δ2u 1 1 δu 2 2 δu 2 δ u − sinθcosθ + + + cosθsinθ sin θ sin θ .........................(a) δR2 R2 δθ R δRδθ R2 δR R2 δθ2 δu δ δ δ δ δ2 u δ = u = sinθ δR = + R1 cosθ δθδ sinθ δR + R1 cosθ δθδ u = δy 2 δy δy δy δy δu δu 1 δu 1 δu sinθ δR sinθ + cosθ + cosθ δR R δθ R δθ δ δu 1 δu 1 δ δu 1 δu = sinθ sinθ + cosθ + cosθ sinθ + cosθ δR δR R δθ R δθ δR R δθ = cos2 θ = sin2 θ 2 1 δ2u 2 δu 2 δ2u 1 2 δu 2 δ u − + cos θ + + sinθcosθ sinθcosθ cos θ ..........................(b) δR2 R2 δθ R2 δRδθ R δR R2 δθ2 Adding (a) and (b), δ2u 1 δu 1 δ2u δ2u δ2u + = + + . δx2 δy 2 δR2 R δR R2 δθ2 This can be written as: URR + 1 1 UR + 2 Uθθ = 0 R R where U = U (R, θ). The boundary condition for equation (i) is given by U (1, θ) = f (θ). By using method of separation of variables, we have U (R, θ) = U (R)θ(θ), R2 R00 + RR0 − U 2 = 0.............................(iii) θ4 + n2 θ = 0 ; n = 0, 1, 2, .........................(iv) Since R00 θ + 1 0 1 R θ + 2 θ00 R = 0 R R 1 0 1 R + R θ + 2 θ00 R = 0 R R 00 1 0 1 R + R θ = − 2 θ00 R R R 00 81 or R00 + R1 R0 1 θ00 =− 2 = n2 R R θ Equation (iii) is a cauchy euler type of equation and its solution can be obtained by changing the independet variables. R = ez or logR = z Now dR dz dR = · dR dz dR = 1 dz · R dR = where D = 1 DR R d dt d d2 R = 2 dR dR 1 DR R 1 dR 1 d =− 2 + R dz x dz dR dz 1 dR 1 dz d =− 2 + · · R dz R dR dz =− =− dR dz 1 dy 1 d2 R + · R2 dt R2 dz 2 1 1 1 2 DR + 2 DR = 2 D2 − D R 2 R R R = 1 D (D − 1) R R2 82 Thus, R R2 dR dR = DR = dR dz d2 R d2 R dR = D (D − 1) = − R dR2 dz 2 dz On substituting these values in equation (iii), d2 R dR dR + − n2 R = 0 − dz 2 dz dz The solution is: R = Aenz + Be−nz = ARn + BR−n The solution of equation (iv) is given by θ = Ccosnθ+Dsinnθ........................(vi) and hence U (R, θ) = (ARn + BR−n ) (Ccosnθ + Dsinnθ) .............................(vii) From physical ground, we ae looking for a bounded solution, equation (vii) is bounded provided Un = (An cosnθ + Bn sinnθ) Rn or n = 0, 1, 2, 3...where ´ 2π ´ 2π An = π1 0 f (θ)cos(θ)dθand Bn = π1 0 f (θ)sin(θ)dθ Example 2: Consider the heat conduction in a thin metal bar of length l with insulated sides. Let us assume that the end x = 0 is held at U0 degree celcius and the end x = lis held at U1 degree celcius for all time t > 0. Let us suppose that the temperature distribution at t = 0is (x, 0) = f (x) 0 ≤ x ≤ l. Determine the temperature distribution in the bar at any position at any time t > 0.(10 marks) Solution The mathematical problem is the following: The heat conduction equaiton Ut = αUxx ............................(i)is to be solved subject to the boundary conditions x = 0 ,U (0, t) = U0 and x = l, U (l, t) = Ul and the initial conditions U (x, 0) = f (x). Use the method of separation 83 of variables for the solution of equation (i). Example 3: A transmission line of negligible resistance and capacitance has its sending end at x = 0and receiving end at x = l. A constant voltage E0 is applied to the sending end while an open circuit is maintained at the receiving end so that the current there is zero. Assuming that the initial voltage and current are zero, determine the voltage at any position x at any time t > 0 Solution The telephone equation can be written as d2 E dE d2 E = LC + (RC + LG) + RGE 2 2 dx dt dT It is given that R = 0, C = 0, hence the equation reduces to d2 E dE = LG ...................................(i) 2 dx dT with boundary conditions given as: dE x = 0, E (0, t) = E0 and x = L, dX (L, t) = 0........................................(ii) The initial conditions are given as t = 0, U (x, 0) = 0 Taking Laplace transforms of equation (i), we have − d2 E L 2 = LG S E − E (0, 0) dx d2 E − LGSE = 0 dx2 √ E = Ae LGSx + Be− √ LGSx A and B are constants to be determined. To determine these constants, we rst need to transform the boundary conditions as follows: 84 x = 0, L E (0, t) = E (0, t) = Es0 x = L, L E (L, t) = E (0) = 0 The application of the above boundary conditions yields x = 0, A + B = Es0 √ √ x = l, Ae LGSl + Be− LGSl = 0 Solving the two equations, we have; √ E0 E0 e2L Las √ √ and B = A= s 1 + e2L Las s 1 + e2L Las Then substituting A and B we have; √ √ √ E0 e2L Las E0 LGSx √ √ e e− LGSx + E= s 1 + e2L Las s 1 + e2L Las √ E0 cosh Las (L − x) √ E= scosh Las · L Using the inverse laplace transforms, we have; " E (x, L) = E0 ∞ 4 X (−1)n (2n−1)2 π2 (2n − 1) (L − x) π 1+ y e 4LGR cos π n=1 2n − 1 2L # Example 2: Unsteady ow of a at plate Lets have a uid of uid density ρconned over half plane y = 0. Let at t = 0, the plate y = 0(which is initially at rest) start moving with velocity u along the x-axis, then the uid above it will start moving because of velocity. Since the plate is taken to be innite, we shall assume that the variables are 2 functions of y and t only. The equation of motion is δu = γ δδyu2 .......(i). It δt resembles the heat equation. Its boundary conditions are a) t < 0, U (y, t) = 0 b) t > 0, U (0, t) = U 85 c) y −→ ∞, U (∞, t) = 0 Its initial conditions are a) t = 0, U (y, 0) = 0 The equation of motion U given by (i) which on taking laplace transforms d2 u s d2 u s u or = − u=0 dy 2 γ dy 2 γ u = Ae √s γ y + Be − √s γ y where A and B are constants of integration. The transformed boundary conditions are : a) u (0, s) = us b) u (∞, s) = 0 Using the boundary conditions, we have A = 0and B = us Thus √ u= u − e s s y γ The inverse transform is 1 u= 2πi ˆ k+i∞ st u (y, s) e dt = u 1 − ekf k−i∞ y √ 2 st The Fourier transform of U (x, t), Ux (x, t), Uxx (x, t)with respect to t are given below: ˆ +∞ U (x, t) e−ist dt F {Uxx } = −∞ ˆ +∞ F {Ux } = −∞ ˆ +∞ F {Uxx } = −∞ δ d U (x, t) e−ist dt = F [U (x, t)] δx dx δ2 d2 −ist U (x, t) e dt = F [U (x, t)] δx2 dx2 86 and ˆ +∞ F {Ut } = −∞ ˆ = e−iγt |+∞ −∞ δ −iγt e dt δt +∞ U e−iγt dt +ir −∞ If U (±∞) = 0 F {Ut } = irF (u) 87