Wiki
Wiki

Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.

Updated


We prove the claim. The change z=λwz=\lambda w, followed by division of PP by λ4\lambda^4, multiplies the renormalized length by λ\lambda. It therefore suffices to treat

∣a∣≤1,∣b∣≤1,max⁡(∣a∣,∣b∣)=1.(1)|a|\le1,\qquad |b|\le1,\qquad \max(|a|,|b|)=1. \tag{1}

The real constant cc is unrestricted. All constants below are independent of it.

The projection inequality

For α∈R/πZ\alpha\in\mathbb R/\pi\mathbb Z and s∈Rs\in\mathbb R let

N(α,s)=#{t∈R:Re⁡P(eiα(t+is))=0}.N(\alpha,s)=\#\{t\in\mathbb R: \operatorname{Re}P(e^{i\alpha}(t+is))=0\}.

Except for finitely many directions and a null set of tangent lines, this is a number between zero and four. If AA is any measurable set of such lines, the projection inequality is

J(P)≤−12∫A(4−N(α,s)) ds dα.(2)J(P)\le-\frac12\int_A(4-N(\alpha,s))\,ds\,d\alpha. \tag{2}

Here J(P)J(P) exists: centering gives eight ends with positions rei(π/8+jπ/4)+O(r−1)re^{i(\pi/8+j\pi/4)}+O(r^{-1}) and tangent derivatives differing from those of rays by O(r−2)O(r^{-2}). Each end therefore has a finite length excess over its radial coordinate. The compact portion of the real algebraic curve has finite length.

For clarity, (2) is the normalization of the projection argument in the proof of Fryntov and Nazarov's Lemma 1, pp. 5--6 of the preprint identified on the source card. Project the plane centrally onto the lower hemisphere of a sphere of radius RR tangent to it at zero. Spherical Crofton gives projected total length 4πR−DR4\pi R-D_R, with

DR=12∫0π∫R(4−N(α,s))(1+s2/R2)−3/2 ds dα.D_R=\frac12\int_0^\pi\int_{\mathbb R} (4-N(\alpha,s))(1+s^2/R^2)^{-3/2}\,ds\,d\alpha.

For all sufficiently large rr, every circle of radius ρ≥r\rho\ge r meets HH at least eight times, by the leading term Re⁡z4\operatorname{Re}z^4. The radial coordinate on the hemisphere therefore bounds the length of the projected part outside DrD_r from below by 4πR−8Rarctan⁡(r/R)4\pi R-8R\arctan(r/R). Its part inside has length at most 8Rarctan⁡(r/R)−DR8R\arctan(r/R)-D_R. Letting R→∞R\to\infty, restricting the nonnegative deficit integral to AA, and then letting r→∞r\to\infty gives (2). Equivalently the unoriented Euclidean line measure is 12 ds dα\tfrac12\,ds\,d\alpha, which fixes the factor in (2).

A quadratic coefficient bounded away from zero

Suppose ∣a∣≥1/100|a|\ge1/100. There is a direction

α0∈{−π/12,π/12,π/2−π/12,π/2+π/12}\alpha_0\in\{-\pi/12,\pi/12,\pi/2-\pi/12,\pi/2+\pi/12\}

for which cos⁡4α0=1/2\cos4\alpha_0=1/2 and Re⁡(ae2iα0)≥∣a∣/2\operatorname{Re}(ae^{2i\alpha_0})\ge|a|/2. Indeed, writing a=x+iya=x+iy, the first two choices yield 3x/2±y/2\sqrt3x/2\pm y/2, and the other two change their signs. The largest absolute value of the first two is at least 3x2/4+y2/4≥∣a∣/2\sqrt{3x^2/4+y^2/4}\ge|a|/2.

For ∣α−α0∣≤1/100|\alpha-\alpha_0|\le1/100, the elementary Lipschitz estimates for the exponential give, with A4=cos⁡4αA_4=\cos4\alpha and B=Re⁡(ae2iα)B=\operatorname{Re}(ae^{2i\alpha}),

A4≥12−4100>25,B≥∣a∣2−2∣a∣100>2∣a∣5.(3)A_4\ge\frac12-\frac4{100}>\frac25, \qquad B\ge\frac{|a|}{2}-\frac{2|a|}{100}>\frac{2|a|}{5}. \tag{3}

In the real polynomial q(t)=Re⁡P(eiα(t+is))q(t)=\operatorname{Re}P(e^{i\alpha}(t+is)) the leading, cubic and quadratic coefficients are respectively

A4,−4ssin⁡4α,B−6s2A4.A_4,\qquad -4s\sin4\alpha,\qquad B-6s^2A_4.

After division by A4A_4 and removal of the cubic term, the quadratic coefficient is exactly

BA4−6s2A42.(4)\frac{B}{A_4}-\frac{6s^2}{A_4^2}. \tag{4}

For ∣s∣≤∣a∣/10|s|\le\sqrt{|a|}/10 it is positive, since A4B>4∣a∣/25>6∣a∣/100A_4B>4|a|/25>6|a|/100 by (3). A monic real quartic with four real roots and zero cubic coefficient has quadratic coefficient −12∑jxj2≤0-\tfrac12\sum_j x_j^2\le0. Thus qq has a nonreal conjugate pair, and N(α,s)≤2N(\alpha,s)\le2 throughout this rectangle, apart from irrelevant null sets.

The rectangle has angular width 1/501/50 and offset width ∣a∣/5\sqrt{|a|}/5. Substitution in (2) gives

J(P)≤−∣a∣250≤−12500.(5)J(P)\le-\frac{\sqrt{|a|}}{250}\le-\frac1{2500}. \tag{5}

Angles are taken modulo π\pi; crossing an endpoint of a fundamental interval does not change the rectangle's line measure.

A substantial linear coefficient

It remains to treat ∣a∣<1/100|a|<1/100, so (1) gives ∣b∣=1|b|=1. Choose α0∈{0,π/2}\alpha_0\in\{0,\pi/2\} such that ∣Re⁡(beiα0)∣≥1/2>7/10|\operatorname{Re}(be^{i\alpha_0})|\ge1/\sqrt2>7/10. For ∣α−α0∣≤1/100|\alpha-\alpha_0|\le1/100 and ∣s∣≤1/100|s|\le1/100, write

q(t)=A4t4+dt3+et2+ft+g.q(t)=A_4t^4+d t^3+e t^2+f t+g.

Its coefficients satisfy

910<A4≤1,∣d∣≤125,∣e∣≤1100+610000=535000,∣f∣≥69100−210000−4106>23.(6)\begin{aligned} &\frac9{10}<A_4\le1,\qquad |d|\le\frac1{25}, \qquad |e|\le\frac1{100}+\frac6{10000}=\frac{53}{5000},\\ &|f|\ge\frac{69}{100}-\frac2{10000}-\frac4{10^6} >\frac23. \end{aligned} \tag{6}

Here d=−4ssin⁡4αd=-4s\sin4\alpha, e=Re⁡(ae2iα)−6s2A4e=\operatorname{Re}(ae^{2i\alpha})-6s^2A_4, and f=Re⁡(beiα+2iase2iα−4is3e4iα)f=\operatorname{Re}(be^{i\alpha}+2ias e^{2i\alpha}-4is^3e^{4i\alpha}). The constant coefficient plays no role.

Divide by A4A_4 and remove the cubic term. The resulting quadratic and linear coefficients, B0B_0 and C0C_0, obey

∣B0∣≤∣e∣A4+3d28A42<150,∣C0∣≥∣f∣A4−∣de∣2A42−∣d∣38A43>12.(7)\begin{aligned} |B_0|&\le\frac{|e|}{A_4}+\frac{3d^2}{8A_4^2}<\frac1{50},\\ |C_0|&\ge\frac{|f|}{A_4} -\frac{|de|}{2A_4^2}-\frac{|d|^3}{8A_4^3}>\frac12. \end{aligned} \tag{7}

These inequalities follow directly by inserting (6); the two error terms in the second line sum to less than 1/10001/1000. If the depressed quartic had four real roots xjx_j, Newton's identities would give

∑jxj=0,∑jxj2=−2B0≤125,∣3C0∣=∣∑jxj3∣≤(∑jxj2)3/2≤1125.\sum_jx_j=0,\qquad \sum_jx_j^2=-2B_0\le\frac1{25}, \qquad \left|3C_0\right|=\left|\sum_jx_j^3\right| \le\left(\sum_jx_j^2\right)^{3/2}\le\frac1{125}.

This contradicts (7). Hence again N(α,s)≤2N(\alpha,s)\le2. The angular and offset widths are both 1/501/50, so (2) gives J(P)≤−1/2500J(P)\le-1/2500. Together with (5), this proves the normalized estimate. Restoring the scale λ\lambda proves the claim.