Fourier Analysis: A New Result

Rajesh Dachiraju · 2025

Read Fourier Analysis: A New Result by Rajesh Dachiraju (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-11465.txt.

Introduction

When a wave-like function jumps at a point, its Fourier series overshoots nearby, and Rajesh Dachiraju proves a new way to read the jump's size straight from the series. His 2025 paper builds a weighted average of the companion conjugate series, weighted by the size of each frequency's coefficient. At smooth points the average fades to nothing.

At a jump point it settles to the jump height, scaled by a fixed constant, a fact he proves by splitting the two cases and leaning on a classical theorem of Lukacs. A second result tracks the total variation of the same average across any interval. The work sharpens the basic theory of Fourier series at discontinuities.

Catalog record

CREATORRajesh Dachiraju
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE6 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-11465

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Fourier Analysis: A New Result
Rajesh Dachiraju

Abstract
. This article contains a new result in Fourier analysis concerning jump type discontinuities.



2020 Mathematics Subject Classification
Primary 42A99;

1.
Notations

Let
f
be a
2\pi
periodic BV function whose derivative is also BV.Let the amount of jump at a point
x
is denoted as
\lfloor f\rfloor(x)=f(x+0)-f(x-0)
Define function
J:\mathbb{R}\to\mathbb{R}
, such that
J(x)=\lfloor f\rfloor(x)
. Define the set
L=\{x/x\in(0,2\pi)\wedge J(x)\neq 0\}
. Let there not be any jumps at
0,\pi,2\pi
. and all jumps at rational multiples of
2\pi
.

Let the Fourier series of
f
be defined as

(1.1)

\displaystyle S[f]=\sum_{\nu=-\infty}^{\nu=\infty}c_{\nu}e^{i\nu x}

and the conjugate Fourier series as

(1.2)

\displaystyle\tilde{S}[f]=-i\sum_{\nu=-\infty}^{\nu=\infty}sign(\nu)c_{\nu}e^{i\nu x}

Let the Fourier partial sum of
f
be defined as

(1.3)

\displaystyle S_{n}(x;f)=\sum_{\nu=-n}^{\nu=n}c_{\nu}e^{i\nu x}

and the conjugate partial sum of
f
as

(1.4)

\displaystyle\tilde{S}_{n}(x;f)=-i\sum_{\nu=-n}^{\nu=n}sign(\nu)c_{\nu}e^{i\nu x}

Define

G(n)=\sum_{i=0}^{n}\left|c_{i}\right|

The Dirichlet kernel
D_{n}(x)
and conjugate Dirichlet kernel
\tilde{D}_{n}(x)
are defined as

(1.5)

\displaystyle D_{n}(x)=\frac{1}{2}+\sum_{\nu=1}^{n}\cos(\nu x)=\frac{\sin(n+\frac{1}{2})x}{2\sin(\frac{1}{2}x)}

(1.6)

\displaystyle\tilde{D}_{n}(x)=\sum_{\nu=1}^{n}\sin(\nu x)=\frac{\cos(\frac{x}{2})-\cos(n+\frac{1}{2})x}{2\sin(\frac{1}{2}x)}

Let

(1.7)

\displaystyle Y_{n}(x)=-\frac{1}{\log(n)G(n)}\sum_{k=1}^{n}\tilde{S}_{k}(x;f)\left|c_{k}\right|

2.
Results

Lemma 2.1
.

(2.1)

\displaystyle\lim_{n\to\infty}Y_{n}(x)=K\frac{J(x)}{2\pi}

where
K
is a constant

Proof.

Case 1 Points of no jump.

\tilde{S_{n}}(x;f)\sim O(1)
and hence
Y_{n}(x)\sim 0
, as
J(x)=0
for no jump the result is proved.

Case 2 Points of jump.

By Lukacs theorem, at points of jump,
\tilde{S_{n}}(x;f)\sim\frac{J(x)}{\pi}\log(n)
,
So

(2.2)

\displaystyle Y_{n}(x)=\frac{1}{\log(n)G(n)}\sum\limits_{k=1}^{n}\frac{J(x)}{\pi}\log(k)|c_{k}|

Noting
|c_{k}|\sim\frac{1}{k}
(as we are assuming points of jump),

(2.3)

\displaystyle Y_{n}(x)=\frac{1}{\log(n)K\log(n)}\sum\limits_{k=1}^{n}\frac{J(x)}{\pi}\log(k)K^{2}\frac{1}{k}

where
K
is a constant.

on taking summation as integration, we get

(2.4)

\displaystyle\frac{1}{K(\log(n))^{2}}K\frac{J(x)}{\pi}\int_{0}^{n}\log(k)K\frac{1}{k}dk

Substituting
\log(k)=t
in the integral, we get

(2.5)

\displaystyle\frac{1}{(\log(n))^{2}}\frac{J(x)}{\pi}\int_{0}^{t}Ktdt

Hence we get

(2.6)

\displaystyle\lim_{n\to\infty}Y_{n}(x)=K\frac{J(x)}{2\pi}

equality in lemma holding only upto a constant factor.

∎

Lemma 2.2
.

Given any
(a,b)\subseteq(0,2\pi)
, show that

(2.7)

\displaystyle\lim_{n\to\infty}V_{a}^{b}(Y_{n})=K\frac{1}{\pi}\sum_{x\in L\cap(a,b)}\left|J(x)\right|

where
K
is a constant.

note:
V_{a}^{b}(f)
is the variation of
f
in
(a,b)
.

Proof.

Let

(2.8)

\displaystyle V_{n}=V_{a}^{b}(Y_{n})=\int_{a}^{b}\left|\frac{1}{\log(n)G(n)}\sum_{k=1}^{n}\tilde{S}^{\prime}_{k}(x;f)\left|c_{k}\right|\right|dx

(2.9)

\displaystyle V_{n}\leq\frac{1}{\log(n)G(n)}\sum_{k=1}^{n}\int_{a}^{b}\left|\tilde{S}^{\prime}_{k}(x;f)\left|c_{k}\right|\right|dx

using a lemma from
[
1
]

(2.10)

\displaystyle V_{n}\leq\frac{1}{\log(n)G(n)}\sum_{k=1}^{n}\int_{a}^{b}\left|\tilde{S}_{k}(x;f^{\prime})-\sum_{y\in L}\frac{J(y)}{\pi}\tilde{D}_{k}(x-y))\left|c_{k}\right|\right|dx

(2.11)

\displaystyle V_{n}\leq\frac{1}{\log(n)G(n)}\sum_{k=1}^{n}\int_{a}^{b}\left|\tilde{S}_{k}(x;f^{\prime})-\sum_{y\in L}\frac{J(y)}{\pi}\tilde{D}_{k}(x-y))\left|c_{k}\right|\right|dx

(2.12)

\displaystyle V_{n}\leq\frac{1}{\log(n)G(n)}(\sum_{k=1}^{n}\int_{a}^{b}\left|\tilde{S}_{k}(x;f^{\prime})\left|c_{k}\right|\right|dx-\sum_{k=1}^{n}\int_{a}^{b}\left|\sum_{y\in L}\frac{J(y)}{\pi}\tilde{D}_{k}(x-y))\left|c_{k}\right|\right|dx)

Since
f^{\prime}
is BV,

(2.13)

\displaystyle\int_{a}^{b}\left|\tilde{S}_{k}(x;f^{\prime})\right|dx\sim O(1)

,So

(2.14)

\displaystyle\sum_{k=1}^{n}\int_{a}^{b}\left|\tilde{S}_{k}(x;f^{\prime})\left|c_{k}\right|\right|dx\sim\sum_{k=1}^{n}(O(1))\frac{V}{k}\sim O(\log(n))

(2.15)

\displaystyle\sum_{k=1}^{n}\int_{a}^{b}\left|\sum_{y\in L}\frac{J(y)}{\pi}\tilde{D}_{k}(x-y))\left|c_{k}\right|\right|dx\leq\sum_{k=1}^{n}\sum_{y\in L}\frac{J(y)}{\pi}\left|c_{k}\right|\int_{a}^{b}\left|\tilde{D}_{k}(x-y)\right|dx

(2.16)

\displaystyle\sum_{k=1}^{n}\sum_{y\in L}\frac{J(y)}{\pi}\left|c_{k}\right|\int_{a}^{b}\left|\tilde{D}_{k}(x-y)\right|dx

\displaystyle=\sum_{k=1}^{n}\sum_{y\in L\setminus(a,b)}\frac{J(y)}{\pi}\left|c_{k}\right|\int_{a}^{b}\left|\tilde{D}_{k}(x-y)\right|dx

\displaystyle+\sum_{k=1}^{n}\sum_{y\in L\cap(a,b)}\frac{J(y)}{\pi}\left|c_{k}\right|\int_{a}^{b}\left|\tilde{D}_{k}(x-y)\right|dx

It is known that

(2.17)

\displaystyle\int_{a}^{b}\left|\tilde{D}_{k}(x-y)\right|dx\sim 2\log(k)

if
y\in(a,b)
and

(2.18)

\displaystyle\int_{a}^{b}\left|\tilde{D}_{k}(x-y)\right|dx\sim O(1)

if
y\notin(a,b)
.

(2.19)

\displaystyle V_{n}\leq\frac{1}{\log(n)G(n)}(\sum_{k=1}^{n}\int_{a}^{b}\left|\tilde{S}_{k}(x;f^{\prime})\left|c_{k}\right|\right|dx

\displaystyle-\sum_{k=1}^{n}\sum_{y\in L\setminus(a,b)}\frac{J(y)}{\pi}\left|c_{k}\right|\int_{a}^{b}\left|\tilde{D}_{k}(x-y)\right|dx

\displaystyle+\sum_{k=1}^{n}\sum_{y\in L\cap(a,b)}\frac{J(y)}{\pi}\left|c_{k}\right|\int_{a}^{b}\left|\tilde{D}_{k}(x-y)\right|dx

Therefore

In the first term on the RHS

(2.20)

\displaystyle\int_{a}^{b}\left|\tilde{S}_{k}(x;f^{\prime})\right|dx\sim O(1)

and hence first term goes to zero.

Coming to second term,
y\notin(a,b)
, so

(2.21)

\displaystyle\int_{a}^{b}\left|\tilde{D}_{k}(x-y)\right|dx\sim O(1)

and hence second term also vanishes.

In the third term
y\in(a,b)
and hence

(2.22)

\displaystyle\int_{a}^{b}\left|\tilde{D}_{k}(x-y)\right|dx=2\log(k)

.Also
|c_{k}|\sim\frac{1}{k}
, so after summation via integration and knowing
G(n)\sim\log(n)
, we get the result.

Read the complete text