Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.
Updated
Statement
For the weighted norm of a sequence is given by (p. 8)
is the spectrum of a sequence, as on the [[analysis/borichev_et_al_2017_spectra_stationary_processes_z/theorem_1|Theorem 1]] page.
Lemma 3 (p. 8). Let , and let be an open arc with . Then there are and such that every sequence with and satisfies
The paper traces the lemma to Szegő and calls it the main ingredient of the proof of [[analysis/borichev_et_al_2017_spectra_stationary_processes_z/theorem_4|Theorem 4]] (p. 8).
The two lemmas it uses
Lemma 1 (p. 7). For a closed arc and there is a polynomial with and . The paper proves it in two lines from Runge's theorem.
Lemma 2 (p. 7). For a closed arc there is a constant such that every polynomial of degree satisfies . This is V. S. Videnskii's Bernstein inequality on an arc, cited from Borwein and Erdélyi, Polynomials and polynomial inequalities (Section 5.1.E19.c); the paper remarks that any bound polynomial, or even subexponential, in would do.
Proof pointer
Pp. 8--9. After a rotation the arc is ; a smooth cutoff equals on and off a slightly larger arc . Lemma 1 gives with and , and with has the form and is exponentially small on , with its derivatives controlled by Lemma 2. Since vanishes near , the prediction error equals , and the Cauchy--Schwarz inequality against the weighted norm bounds it by a constant depending on and times , which is below once .
Used by. [[analysis/borichev_et_al_2017_spectra_stationary_processes_z/theorem_4|Theorem 4]].
Source. Alexander Borichev, Mikhail Sodin, Benjamin Weiss, Spectra of stationary processes on , arXiv:1701.03407v1 (12 January 2017), identified on the [[analysis/borichev_et_al_2017_spectra_stationary_processes_z/_index|source card]]; labels and pages are that version's.
Read depth. Claims checked: Lemmas 1--3 and the proof of Lemma 3 were read clause by clause on pp. 7--9; Lemma 2 is recorded as the paper cites it and was not checked against its source. Nothing here is independently reviewed.
Bears on. Problem 1150 (context only): the source card notes that this lemma, with the proof of Theorem 4, would make a stationary-limit argument quantitative when a fixed arc contains the spectrum. The paper does not relate the prediction length to the maximum modulus of a polynomial, and it does not mention the problem.