Hardy’s second proof of the fundamental theorem of algebra is the one most textbooks give, and it is a model of a very common kind of argument: a minimum exists, and a minimum cannot be at a bad point, so it is at a good point. It needs two ingredients. One is the two-dimensional version of the theorem of §103 that a continuous function on a closed bounded region attains its bounds. The other is an algebraic observation, often called d’Alembert’s lemma: at any point where f(z0)=0, there are nearby points where ∣f∣ is smaller.
(The appendix has no section numbers. Hardy’s proof extends the results of §§103–105 to functions of two variables.)
The idea
Think of ∣f(x+iy)∣ as the height of a landscape over the plane. Far out, ∣f(z)∣≈∣a0∣∣z∣n is huge, so the landscape is a giant bowl. A bowl that is continuous and closed off at the rim has a lowest point. The theorem says that the lowest point is at height 0, which is the same as saying that f has a root there.
Why can the lowest point not be at a positive height? Near z0, a polynomial looks like its Taylor expansion:
f(z0+ζ)=f(z0)+Akζk+(higher powers of ζ),
where Ak is the first nonzero coefficient after the constant. For small ζ, the term Akζk dominates the rest, and as ζ goes round a small circle, Akζk goes k times round a small circle. So the values f(z0+ζ) circle round f(z0), and some of them must lie on the side of f(z0) facing the origin. In the real landscape there are no “pits with a flat positive bottom”: every point not on a root has a downhill direction.
ExploreWalking downhill on ∣f(z)∣
Try this.Drag the purple point z0 anywhere in the left panel. The small circle round it is green where ∣f∣<∣f(z0)∣ and orange where it is not, and the purple arrow points in Hardy's direction. Press “Step downhill” repeatedly. Then press “Go to a point where f′=0” and look at both panels again.
f(z) = z3−3z+3
z-plane: darker means smaller |f(z)|
xy
Z-plane near f(z₀): image of the small circle
XY
z₀ (drag), Hardy’s direction(s), and the image of the circle
part of the circle where |f| < |f(z₀)|
part where |f| ≥ |f(z₀)|
roots; on the right, the circle |Z| = |f(z₀)| and the direction of 0
z0=−1.6+1.5i, m=∣f(z0)∣=15. The first nonzero coefficient in f(z0+ζ)=f(z0)+A1ζ+A2ζ2+⋯ is A1 (k=1, a=∣Ak∣=14.5). Hardy’s direction ϕ with α+kϕ=μ+π: 292°. On the circle of radius 0.25, the least value of ∣f∣ is 11.7, below m.
What to notice
Away from a root, once the radius is small enough the green arc is always there: about half the circle when k=1 (a large circle may have none). The right panel shows why: the image of the small circle is a loop round f(z0), and part of it crosses inside the dashed circle ∣Z∣=∣f(z0)∣, closer to 0. Repeated steps run down into one of the dark wells, which are the roots. At the point where f′(1)=0 (so k=2) the image winds twice round f(z0) and there are two downhill arcs and two uphill ones, but still a way down.
The descent in the widget is not part of Hardy’s proof, which never moves anywhere: it only uses that a downhill direction exists at any point that is not a root. But stepping repeatedly in Hardy’s direction, with a step chosen so that ∣f∣ really decreases, is a version of Newton’s method, and it does find roots in practice.
Precise statements
TheoremContinuous functions on a square attain their boundsHardy: Appendix II (B); cf. §103
Let F(x,y) be continuous on the closed square D: ∣x∣≤R, ∣y∣≤R. Then F is bounded on D, and there are points of D where it takes the values of its upper and lower bounds.
Hardy says this can be proved “much as in the last paragraph of §105”, that is, by repeated bisection, now of squares into four. (His first proof in §105 uses a Dedekind section, which has no analogue in two dimensions.) Continuity in two variables is defined in §108; Hardy does not prove the two-dimensional theorem in the main text.
Lemmad'Alembert's lemmaHardy: Appendix II (B)
Let f be a polynomial and f(z0)=0. Then there is a point z, as close to z0 as we please, with ∣f(z)∣<∣f(z0)∣.
TheoremThe fundamental theorem of algebraHardy: Appendix II (B)
Every polynomial of degree n≥1 with complex coefficients has a root. More precisely, ∣f(z)∣ attains a least value m over the whole plane, and m=0.
Proofs
Proofd'Alembert's lemma
Strategy
Expand f round z0 and keep only the first non-constant term Akζk. Choose the direction of ζ so that Akζk points from f(z0) straight towards the origin, and its length ρ so small that the remaining terms cannot undo the gain.
Expand round z0
Rearranging f(z0+ζ) in powers of ζ gives f(z0+ζ)=f(z0)+A1ζ+A2ζ2+⋯+Anζn, where the A‘s do not depend on ζ and An=a0=0. Let Ak be the first nonzero one among A1,…,An.
Polar forms
Write f(z0)=meiμ with m>0, Ak=aeiα with a>0, and ζ=ρeiϕ.
Make the tail small
Since ∣Ak+1ζk+1+⋯+Anζn∣≤ρk+1(∣Ak+1∣+⋯+∣An∣) when ρ≤1, we can choose ρ so small that aρk<m and the tail g satisfies ∣g∣<21aρk. Then
f(z0+ζ)=meiμ+aρkei(α+kϕ)+g.
Point the main term at the origin
Choose ϕ with α+kϕ=μ+π. Then ei(α+kϕ)=−eiμ, so the first two terms combine to (m−aρk)eiμ, a point on the segment from f(z0) towards 0.
Estimate
∣f(z0+ζ)∣≤(m−aρk)+∣g∣<m−21aρk<m.
Since ρ may be taken as small as we please, there are such points arbitrarily close to z0.
∎
ProofEvery polynomial has a root (proof B)
Strategy
Restrict to a large square D, on whose boundary ∣f∣ is larger than ∣f(0)∣. The least value of ∣f∣ on D is attained at some z0, which must be inside D, and is then the least value over the whole plane. By d’Alembert’s lemma it cannot be positive.
∣f∣ is large far out
As in the previous lesson, ∣f(z)∣=∣a0∣∣z∣n∣1+η∣ with ∣η∣→0 as ∣z∣→∞. So we can choose R such that ∣f(z)∣>∣f(0)∣ for every z with ∣z∣≥R, and in particular on and outside the boundary of the square D of side 2R centred at 0.
A least value is attained in D
F(x,y)=∣f(x+iy)∣ is continuous and non-negative on D, so it attains its lower bound m at some point z0 of D.
z0 is a minimum over the whole plane
m≤∣f(0)∣ because 0 is in D. Points outside D, and points on its boundary, have ∣f∣>∣f(0)∣≥m. So z0 lies inside D, and ∣f(z)∣≥m for every z in the plane.
m=0
If m>0, d’Alembert’s lemma gives points z arbitrarily close to z0 with ∣f(z)∣<m, which contradicts the previous step. So m=0, that is, f(z0)=0.
∎
A hidden ingredient, and Littlewood’s repair
Look again at the step “choose ϕ with α+kϕ=μ+π”. To do it we need a number ζ with ζk in a prescribed direction; in other words we are solving an equation ζk=c. So proof B assumes the fundamental theorem for binomial equationszk=c. We know these are solvable from de Moivre’s theorem (§48), but that rests on the circular functions. Is there a proof that avoids them?
Littlewood showed that there is, by running the same least-value argument on f(z)=zn−c itself.
Reduce to odd n. Square roots can be found by pure algebra: if c=a+ib with b=0 and r=a2+b2, the roots of z2=c are ±(21(r+a)+isgn(b)21(r−a)). If n=2vN with N odd, then v successive square roots reduce zn=c to wN=d.
If the minimum point is not 0. For n odd, let z0 be a point where ∣zn−c∣ is least. If z0=0, then A1=nz0n−1=0, so k=1, and choosing the direction only needs the solution of a linear equation.
If the minimum point is 0. Then f(ζ)=ζn−c. Taking ζ=±ρ and ±iρ, and using that n is odd, f takes the four values −c±ρn and −c±iρn: small steps from −c in each of the four axis directions. One of these brings −c nearer to 0 (unless c=0), so 0 was not a minimum point after all.
Exercises
ExerciseTaylor coefficients at a point●○○
Let f(z)=z3−3z+3. Expand f(1+ζ) in powers of ζ. What are k, a, α and μ at z0=1, and which directions ϕ does Hardy’s rule give? Check directly that ∣f(1+iρ)∣<1 for small ρ>0.
Hint
(1+ζ)3=1+3ζ+3ζ2+ζ3.
Solution
f(1+ζ)=1+3ζ+3ζ2+ζ3−3−3ζ+3=1+3ζ2+ζ3. So f(1)=1 (m=1, μ=0), A1=0 and A2=3: k=2, a=3, α=0. The rule α+2ϕ=μ+π gives ϕ=21π (and ϕ=23π, the other solution modulo 2π): straight up or down. Indeed f(1+iρ)=1−3ρ2−iρ3, and
∣f(1+iρ)∣2=(1−3ρ2)2+ρ6=1−6ρ2+9ρ4+ρ6<1
as soon as 9ρ2+ρ4<6, for instance for 0<ρ<21. Along the real axis, by contrast, f(1±ρ)=1+3ρ2±ρ3>1 for small ρ: on the real line z0=1 looks like a minimum.
ExerciseWhy the real line is not enough●●○
The polynomial x2+1 has no real root, and its least value on the real line, 1 at x=0, is positive. Locate exactly which step of d’Alembert’s lemma fails if we are only allowed to move along the real axis.
Solution
At x0=0: f(0+ζ)=1+ζ2, so k=2, A2=1, α=0, μ=0. Hardy’s rule asks for 2ϕ=π, that is ϕ=21π: a vertical step, ζ=iρ, giving f(iρ)=1−ρ2<1. On the real axis only ϕ=0 and ϕ=π are available, and both give ζ2=ρ2>0, which increases ∣f∣. The lemma needs every direction to be available, so that ζk can point anywhere: this is where the complex numbers are essential.
ExerciseNo hills either●●○
Show that ∣f∣ has no local maximum: if f is a non-constant polynomial and z0 any point, there are points z arbitrarily close to z0 with ∣f(z)∣>∣f(z0)∣.
Hint
Change one sign in the choice of ϕ. Treat the case f(z0)=0 separately.
Solution
If f(z0)=0, any nearby point that is not a root will do, and the roots are finitely many. If f(z0)=meiμ=0, take k, a, α, ρ and g as in the lemma, but choose ϕ with α+kϕ=μ. Then f(z0+ζ)=(m+aρk)eiμ+g, so ∣f(z0+ζ)∣≥m+aρk−∣g∣>m+21aρk>m. (This is the polynomial case of the maximum modulus principle.)
ExerciseA minimum on a closed set, but not on an open one●○○
(a) Show that ∣ez∣ has no least value on the complex plane. (b) Which step of proof B fails for ez in place of a polynomial?
Solution
(a) ∣ex+iy∣=ex>0 for every z, and ex→0 as x→−∞, so the lower bound 0 is not attained.
(b) The first step: ∣ez∣ is not large far out in every direction (it is tiny for large negative x), so we cannot confine the search for a minimum to a bounded square. d’Alembert’s lemma itself does hold for ez (it is never 0, and it has downhill directions everywhere); what fails is the existence of a minimum.
ExerciseSquare roots by algebra●○○
Use the formula in Littlewood’s argument to find the square roots of 3+4i and of −5−12i, and verify that the formula gives a square root of a+ib in general.
Hint
In general, with p=(r+a)/2 and q=sgn(b)(r−a)/2: (p+iq)2=p2−q2+2ipq=a+isgn(b)r2−a2=a+isgn(b)∣b∣=a+ib.
ExerciseFour small steps●●○
Let w=−c=0 and t>0 be small. Show that at least one of w+t, w−t, w+it, w−it is nearer to 0 than w is. (This is the last step of Littlewood’s argument.)
Hint
Write w=u+iv; at least one of u,v is not 0.
Solution
Say u=0 (otherwise swap the roles of the real and imaginary parts and use ±it). Take the sign s=−sgn(u) and 0<t<2∣u∣. Then ∣w+st∣2=(u+st)2+v2=(∣u∣−t)2+v2<u2+v2=∣w∣2, since ∣u∣−t<∣u∣. Moving the real part towards 0 (by less than twice its size) shortens the vector.
Check yourself
Check yourselfEnter a number
For f(z)=z3−3z+3 at z0=−1 (where f′(−1)=0), what is k, the index of the first nonzero coefficient Ak with k≥1?
Why
f(−1+ζ)=5−3ζ2+ζ3: A1=f′(−1)=0 and A2=21f′′(−1)=−3=0. So k=2, and there are two downhill directions, now along the real axis (α=π, μ=0, so 2ϕ=0 or 2π).
Check yourselfTrue or false?
Every continuous function F(x,y)≥0 on the whole plane attains a least value.
Why
False: F=∣ex+iy∣=ex has lower bound 0 and never reaches it. The theorem on bounds needs a closed bounded region. That is why proof B first shows that ∣f∣ is large outside a big square, so that the search for a minimum can be confined to the square.
Check yourselfChoose one
At a point where f(z0)=0 and k=3, in how many of the directions round z0 does ∣f∣ decrease to first order, and how are they arranged?
Why
To first order f(z0+ρeiϕ)≈f(z0)+aρ3ei(α+3ϕ), and the main term points “towards 0” (has a component opposite to f(z0)) for three arcs of ϕ, each of width π/3, alternating with three uphill arcs. Hardy’s three directions are the middles of the downhill arcs. In the widget, k=2 shows two and two.
Check yourselfChoose one
Which fact does proof B, as Hardy first gives it, quietly assume?
Why
Choosing ϕ with α+kϕ=μ+π means finding ζ with ζk pointing in a given direction, i.e. solving ζk=c. De Moivre’s theorem provides this; Littlewood’s argument shows how to avoid it. No differentiability of ∣f∣ is used, and coefficients may be complex.
Hardy’s language
Summary
Summary
Far out ∣f(z)∣ is large, so ∣f∣ attains a least value m at some point z0 of a big square, and this is its least value over the whole plane.
d’Alembert’s lemma: if f(z0)=meiμ=0 and Ak=aeiα is the first nonzero Taylor coefficient, a small step ρeiϕ with α+kϕ=μ+π gives ∣f∣<m−21aρk.
So m=0: the least value is a root.
The step uses complex directions (it fails on the real line) and assumes ζk=c is solvable; Littlewood’s variant removes that assumption using square roots and four axis-directions.