Practice midterm 3 solutions

Complex analysis, lecture 2
(April 17, 2026)

As usual, be sure to include your method, and remember to write your name.

Problem 1. Find the Taylor expansion of f(z)=e2ze2 centered at z0=1. What is its radius of convergence?

The above problem is directed towards Objective 7 (power series).

Solution. The nth derivative of f for n1 is 2ne2z, so evaluating at 1 gives 2ne2; evaluating f at 1 gives 0, so the Taylor series is

f(z)=n=12ne2n!(z+1)n.

Alternatively we could observe that

e2z=e2e2(z+1)=n=02ne2n!(z+1)n,

and subtracting e2 kills off the constant term. Finally, the radius of convergence is infinite, either since that of ez is and we’re substituting 2z, so |2z|< holds for all z; or since f is entire, so the distance to the nearest singularity is infinite; or by the ratio test, the radius of convergence is

limn|2ne2/n!2n+1e2/(n+1)!|=limn|n+12|=.

Problem 2. Find the Taylor expansion of f(z)=z2z31 at infinity.

The above problem is directed towards Objective 7 (power series).

Solution. Let w=1z. Then f(z)=z2z31=w2w31=w1w3, which by the geometric series expansion is

f(z)=n=0w3n+1=n=0z3n1=z1+z4+z7+.

We could also write this as

n=0anzn

where an=1 if n=3m+1 for some integer n, and an=0 otherwise; either form is fine.

Problem 3. Compute the Laurent expansion of f(z)=1(z+1)(z+2) on the annulus {1<|z|<2}.

The above problem is directed towards Objective 8 (Laurent series).

Solution. Writing f(z)=1z+11z+2, we use partial fractions to write it as Az+1+Bz+2 for some constants A,B: we have Az+1+Bz+2=A(z+2)+B(z+1)(z+1)(z+2)=1(z+1)(z+2), so 2A+B=1 and A+B=0, hence A=1 and B=1; and we can check that indeed 1z+11z+2=1(z+1)(z+2). Since 1z+1 is analytic on |z|>1 (since its only singularity is at z=1) and 1z+2 is analytic on |z|<2 (since its only singularity is at z=2), we can take f0(z)=1z+2 and f1(z)=1z+1, and find their Taylor expansions at 0 and at infinity respectively. By the geometric series formula,

1z+2=1211+z/2=12n=0(z/2)n=n=0(1)n+12n1zn.

Meanwhile

1z+1=11/(1/z)+1=1/z1+1/z=n=1(1)n+1zn.

Combining the two, we get

f(z)=n=anzn

where an=(1)n+12n1 for n0 and an=(1)n+1 for n<0.

Note that the first power series converges for |z/2|<1, i.e. |z|<2, and the second converges for |1/z|<1, i.e. |z|>1, so the overall Laurent series converges on the desired annulus.

Problem 4. Find and classify the isolated singularities of f(z)=cos(1/z)z41. For any poles, find their order.

The above problem is directed towards Objective 8 (Laurent series).

Solution. The singularities are at z=0 for cos(1/z) and wherever z4=1, i.e. z=1,i,1,i. Factoring z41=(z21)(z2+1)=(z+1)(z1)(z+i)(zi), since cos(1/z) is nonzero at each of these points these are all simple poles.

At z=0, we have

cos(1/z)=n=0(1)n(2n)!z2n,

which has infinitely many negative terms; since z41 is analytic and nonzero at z=0, dividing by it does not change the type of singularity, so we have an essential singularity at z=0.

Since we’ve had less time with the residue theorem, I’ve included an extra practice problem on it below (so three instead of two). However, only two will appear on the actual exam.

Problem 5. Let D be the rectangle in with corners at i1, i1, 4+i, and 4i. Using the residue theorem, compute

Dzsinz𝑑z.

The above problem is directed towards Objective 9 (the residue theorem).

Solution. The singularities of zsinz are at zeros of sinz, i.e. at z=πn for n. Since sin(πn)=cos(πn)0, these are simple zeros, hence at worst simple poles of zsinz. The only zeros in D are at 0, which is removable since both numerator and denominator have a zero of order 1, and at π, where since it is a simple pole we get residue πcosπ=π. Therefore by the residue theorem the integral is π2πi=2π2i.

Problem 6. Using the residue theorem, compute

1x4+1𝑑x.

The above problem is directed towards Objective 9 (the residue theorem).

Solution. Let D be the upper half-disk of radius N, with boundary [N,N] together with the arc γ from N to N of radius N. The integral of 1z4+1 over γ will vanish in the limit by the ML bound: in absolute value the integrand is bounded by something of order 1N4 and the arc is of length πN, so the product goes to 0. Therefore it’s enough to evaluate the integral over D and take the limit. By the residue theorem, it suffices to find the residues at singularities in D, i.e. zeros of z4+1, which are at eπi/4,e3πi/4,e5πi/4,e7πi/4, of which the only two in D are eπi/4 and e3πi/4, at each of which 1z4+1 has a simple pole with residue 14z03, which at these points gives 14e3πi/4=14eπi/4=28(1+i) and 14e9πi/4=14eπi/4=28(1i). Adding these together gives 24i, so by the residue theorem the integral is 2πi24i=π22=π2.

Problem 7. Using the residue theorem, compute

ππ12+sinθ𝑑θ.

The above problem is directed towards Objective 9 (the residue theorem).

Solution. Writing z=eiθ, so dθ=dziz as in class and sinθ=12i(z1/z), this is

|z|=112+12i(z1/z)1iz𝑑z =2|z|=114iz+z21𝑑z.

To apply the residue theorem, we need to find the singularities of the integrand, for which we need the zeros of z2+4iz1. A nice way to do this is to write y=iz, so z2+4iz1=y2+4y1, so it suffices to find the zeros of y24y+1, which are at y=2±3 by the quadratic formula. Therefore the zeros of the original polynomial in z are at z=iy=i(2±3). Since i(2+3) is not in the unit disk but i(23) is, we only need the residue at the latter, which is 12z+4i|z=i(23)=12i3, and we have a leading factor of 2, so the overall integral is 22πi2i3=2π3.