Wiki
Wiki

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

Updated


Source. H.-J. Rack and R. Vajda, Optimal cubic Lagrange interpolation: Extremal node systems with minimal Lebesgue constant, Stud. Univ. Babeş-Bolyai Math. 60 (2015), no. 2, 151--171; the edition read is named on the source card.

Statement

Setting. b>1b>1 is the right endpoint of the parameter interval [1,b][1,b] in Theorem 4.2; t=0.4177913013…t=0.4177913013\ldots is the constant (3.5) of the optimal canonical system −1<−t<t<1-1<-t<t<1. In the proof of Theorem 5.2 (p. 165) bb is the point x>1x>1 where the Lebesgue function of that canonical system, x(1−t+t2−x2)/((−1+t)t)x(1-t+t^2-x^2)/((-1+t)t) there (6.6), equals Λ4∗\Lambda_4^*.

Lemma 4.3 (p. 157). bb is the unique positive root of the degree-18 integer polynomial

P18(x)=−121+220x−1014x2+1344x3+3283x4−5166x5+4502x6+15692x7−84178x8+7868x9+210676x10−25694x11−310732x12+34154x13+255377x14−8450x15−124700x16+26875x18,(4.2)\begin{aligned} P_{18}(x)={}&-121+220x-1014x^2+1344x^3+3283x^4-5166x^5+4502x^6\\ &+15692x^7-84178x^8+7868x^9+210676x^{10}-25694x^{11}-310732x^{12}\\ &+34154x^{13}+255377x^{14}-8450x^{15}-124700x^{16}+26875x^{18}, \end{aligned}\qquad(4.2)

and numerically b=1.0433133411…b=1.0433133411\ldots (4.3); (4.6) on p. 158 gives more digits.

Lemma 4.4 (pp. 157-158). bb is given by radicals in terms of tt as

b=b(t)=(p−D)1/3+(p+D)1/3,p=t+t32+2t,D=−127(1+(−1+t)t)3+(t+t3)24(1+t)2,(4.4)b=b(t)=\bigl(p-\sqrt{D}\bigr)^{1/3}+\bigl(p+\sqrt{D}\bigr)^{1/3},\qquad p=\frac{t+t^3}{2+2t},\quad D=-\frac1{27}\bigl(1+(-1+t)t\bigr)^3+\frac{(t+t^3)^2}{4(1+t)^2},\qquad(4.4)

which is the printed (4.4) with its two repeated subexpressions named pp and DD. Substituting (3.5) for tt gives a long expression for bb by radicals alone, printed as (4.5) on p. 158.

Proof pointer

Lemma 4.3: Section 6.4, p. 166. Mathematica's RootReduce applied to the equation (6.6) =Λ4∗=\Lambda_4^*, with Λ4∗\Lambda_4^* and tt given as roots of (3.1) and of Q3∗(x2)Q_3^*(x^2) (input (6.10)), returns P18P_{18} (output (6.11)). Lemma 4.4: Section 6.5, p. 167. With Λ4∗=(1+t2)/(1−t2)\Lambda_4^*=(1+t^2)/(1-t^2) (3.9), the equation becomes the cubic (t+t3)+(1+t3)x+(−1−t)x3=0(t+t^3)+(1+t^3)x+(-1-t)x^3=0 (6.13), solved by Cardan's formula.

Read depth

Claims checked: both lemmas and (4.2) were read on the page image of the print, and the coefficients of (4.2) were matched against the printed output (6.11). The computer-algebra steps were not re-run, and (4.5) was not transcribed or checked. Nothing here is independently reviewed.

Dependencies

The cubic constants (3.1)-(3.9), recorded by the paper from its references [23], [24], [29], [30]; see the Theorem 5.2 page.

Bears on

  • Problem 1129: the lemmas identify the constant bb that bounds the parameter ranges in the paper's description of all minimizing four-node systems (Theorem 5.2).