Вы находитесь на странице: 1из 5

1.

Show that under stereographic projection the inversion f (z ) = 1/z on the complex plane corresponds to reection on the equatorial plane of the Riemann sphere. (Recall that w = (x1 , x2 , x3 ) on the sphere and z = x + iy on the plane are related by x = x1 /(1 x3 ), y = x2 /(1 x3 ) and dont forget that w is on the unit sphere of R3 .) 2. Use de Moivres formula to prove the well-known trigonometric identities: cos 3 = cos3 3 cos sin2 , sin 3 = 3 cos2 sin sin3 . De Moivres formula says cos n + i sin n = (cos + i sin )n . Setting n = 3 and expanding gives cos 3 + i sin 3 = cos3 + 3i cos2 sin + 3i2 cos sin2 + i3 sin3 = (cos3 3 cos sin2 ) + i(3 cos2 sin sin3 ). The result follows by equating real and imaginary parts. 3. Show that the function f (z ) = |z | is not analytic anywhere. If f (z ) were analytic in some domain D then certainly (by the chain rule) the function f (z )2 = |z |2 = x2 + iy 2 would be analytic in D as well. However, by the CauchyRiemann equations, the latter function only has a derivative at those points z = x + iy y2 ) x2 ) y2 ) x2 ) = ( and ( = ( . The rst equation gives x = y (the second gives where ( x y y x 0 = 0 which is no information), meaning that f (z )2 only has a complex derivative on the diagonal y = x. Since D is open, the derivative cannot exist at every point of D and f (z )2 is analytic nowhere, the same being therefore true of f (z ). z 2 + 2z + 1 4. Find every possible Laurent series for f (z ) = centered at z = 0. z3 + z 1 + z22 then, for 0 < |z | < 1 (so that |z 2 | < 1 as well), Since f (z ) = z +1 1 1 (z 2 )j = + 2 2z 2 + 2z 4 2z 6 + . . . . f (z ) = + 2 z z j =0 For |z | > 1, | z12 | < 1 so f (z ) = 1 2 + 2 z z 1 1 + z12 = 1 2 + 2 z z

j =0

1 z2

(1) (2)

1 2 2 2 2 = + 2 4 + 6 8 + .... z z z z z 1

5. Find the residues of f (z ) = z 2 at (a) z = 0, (b) z = +1, (c) z = 1. First of all we note that the problem is somewhat ill-dened because f (z ) could be taken to be any branch of the square-root function, which is multiple-valued. We will carefully keep this in mind in all that follows. (a) The residue at zero is not dened because any branch of z 2 has a non-isolated singularity at z = 0. To see this, assume that indeed it were possible to dene such a branch f (z ) of the square root function having 0 as an isolated singularity. Then we would have that f (z ) = exp 1 Log(z ) 2 (3)
1

for one of the two choices of sign (+ sign if, say, f (1) = +1, or sign if f (1) = 1), where Log stands for the standard branch of the logarithmic function, and the equality (3) would hold for all z not on the real semiaxis (, 0]. Observe that since we are assuming that 0 is an isolated singularity, the function f (z ) is analytic (hence continuous) everywhere else, so that the equality (3) would have to hold always with the + or always with the sign, at least for z not on the excluded half-line. Let us assume, for concreteness, that the sign is actually + so that f (z ) coincides with the standard branch of the square-root function on C (, 0]. However, f (z ) itself is assumed to be dened, even analytic and therefore continuous even at points on (, 0) (but of course not at 0 itself.) Therefore, as z approaches some point on this half-line, lets say 1, f (z ) should certainly have a limit. But this contradicts the fact that f (z ) is given by equation (3) (with the + sign) since Log(z ) approaches +i as z 1 from 1 above (giving lim f (z ) = exp( 2 i) = +i) and i as z 1 from below (giving 1 lim f (z ) = exp( 2 (i)) = i. Summing up, f (z ) cannot be analytic on a whole punctured neighborhood of 0, so 0 is not an isolated singularity and the residue is not dened. (b) Near z = +1, f (z ) must coincide with one of the two branches exp( 1 Log(z )) 2 of the square-root function, which are analytic at z = +1, so the residue is zero. (c) As above, near z = 1, f (z ) must coincide with one of the (analytic) branches exp( 1 Log0 (z )), so the residue is zero (note that here we used the branch Log0 2 which is analytic at z = 1 the standard branch Log fails to be analytic there.) 6. Find a harmonic function dened in the region contained between the circles x2 x + y 2 = 0 (boundary value 1) and x2 2x + y 2 = 0 (boundary value 10). Help: use the inversion w = 1/z . 2

The conformal inversion w = u + iv = 1/z maps the region into the vertical strip 1 < u < 2 with u = 1 corresponding to the second circle (boundary value 10) and u = 2 to the rst circle (boundary value 1). The function (w) = 19 9u meets these boundary conditions, so the function (z ) = 19 9 (1/z ) = 19 9x/(x2 + y 2 ) solves the problem. 7. Prove that if f (z ) is an analytic function in a domain D and such that its imaginary party f attains its maximum in D then f is constant. You may assume that D is simply connected. Since the imaginary part v (z ) = f (z ) is a harmonic function on D which by assumption attains it maximum in D, then the maximum/minimum principle for harmonic functions implies that v (z ) is constant in D (the textbook proves the maximum/minimum principle assuming D is simply connected but this assumption is not necessary.) Therefore f (z ) = v/y + iv/x is identically zero and f (z ) must be constant in D. 8. (a) Find the rst three terms of the Maclaurin series for tan z . 3 5 7 2 4 6 sin z tan z = cos with sin z = z z +z z + . . . and cos z = 1 z +z z + . . .. z 3! 5! 7! 2! 4! 6! Rather than evaluating derivatives of tan z , we use long division to nd tan z = z 1
z3 3! z2 2! 3

+ +

z5 5! z4 4!

z7 7! z6 6!

+ ... + ...

=z+

z 2 + z5 + . . . . 3 15

(b) What is the residue of the logarithmic derivative of cos( a ) at z = 0? (Here a is z any nonzero constant.) ) is za2 tan( a ), one may be tempted to just plug Since the logarithmic derivative of cos( a z z a in z z in the Maclaurin series above. However, this would be utterly wrong! To compute the residue at z = 0 we need a power series expansion valid for 0 < z < for some > 0 (perhaps small). The power series found in part (a) for tan z is valid |a | for |z | < . Plugging in a for z we get a series valid for |z | > 2 , in other words, not 2 z valid for |z | > 0 and small. Indeed, given that tan z has a pole at every z = k with k a integer, tan( a ) has poles at z = k for k = 0 integer and since these poles accumulate z ), so the residue at at z = 0 it follows that z = 0 is not an isolated singularity of tan( a z z = 0 is undened.

9. Evaluate
0

d 1 + sin2

Using sin2 = 1 (1 cos(2)) and letting = 2 we obtain the desired integral equals 2
2 0

d = 3 cos

|z |=1

dz iz z 1 z+2

2 i

|z |=1

dz 1 = 4 Res 2 2 6z z 1 z =3 5 6z z 1

= 4 10. Evaluate

1 6 2z

z =3 5

2 = . 5 dx . 2x + 4

x2

Hint: use calculus. Complex analysis wont help here! 11. Evaluate Pr. V.

dx , (x a)2 (x 1)

where a lies on the upper half-plane. We use an indented contour consisting of two segments [R, 1 ], [1 + , R] (oriented left-to-right) for R large and small together with the small half-circle |z 1| = , z 0 (avoiding the pole at +1 oriented clockwise) and the big half-circle |z | = R, z 0 (oriented counterclockwise). The integral of (za)1 2 (z 1) over equals 2i 2i Resz=a (za)1 2 (z 1) = (a1)2 . The integral over the large semicircle goes to zero as R since the integrand is a rational function whose denominator has degree at least two larger than the numerator. The integral over the small semicircle in the limit i 0 accounts for i Resz=1 (za)1 2 (z 1) = (1a)2 . The limit of the integral over the horizontal segments is the sought-after principal-value integral so putting everything i i 2i (1 = (a . together gives the answer (a 1)2 a )2 1)2 12. True or false (justify): (a) If f (z ) has a pole of order two at z = z0 then f (z )dz = 0 for any small circle centered at z0 . False, consider f (z ) = 1/z 2 + 1/z . Then the integral equals 2i.

(z ) (b) If f (z ) is meromorphic and Resz=0 f = 0 then f (z ) is analytic or has a removf (z ) able singularity at z = 0. True by the argument principle: the integral of f (z )/f (z ) over a small circle |z | = containing no other singularity than (possibly) z = 0 is equal to the multiplicity of the zero or the negative order of the pole at z = 0. Since this integral is, by the residue theorem and the fact Resz=z1 = 0, equal to zero, it follows that there is no pole there, and in fact f (z ) extends to an analytic function with f (0) = 0. (z ) (c) If f (z ) has an isolated singularity at z = 0 and if Resz=0 f = 0 then the f (z ) singularity is removable. True for meromorphic functions but false if the singularity is essential. E. g., let 1 f (z ) = exp( z ). Then f /f = 1/z 2 has residue zero at z = 0.

(d) A suitable branch of the function z 1/2 has a removable singularity at z = 0. False. The singularity at z = 0 can never be isolated for any choice of branch. (e) A harmonic function on a multiply connected domain does not have a harmonic conjugate False. There are many examples. An interesting one is given by the harmonic x function (z ) = (z 1 ) = x2 + on C {0}. Since is the real part of an analytic y2 function, it immediately has the harmonic conjugate = (z 1 ). I suggest you write down the formula for . (f) If f (z ) is complex-valued and continuous on a domain D and if any loop integral f (z )dz vanishes (i. e., equals zero) then f (z ) is analytic. True. The condition implies that f (z ) has an antiderivative F (z ). Since F (z ) has a complex derivative (namely f ), it is analytic, hence has complex derivatives of all orders so in particular f has a complex derivative, so it is analytic (this problem is actually just Moreras Theorem.)

Вам также может понравиться