Improvements on Waring s Problem

Similar documents
Improvements on Waring s Problem

Two Approaches to Proving. Goldbach s Conjecture

Small signal analysis

Chapter 11. Supplemental Text Material. The method of steepest ascent can be derived as follows. Suppose that we have fit a firstorder

Harmonic oscillator approximation

Pythagorean triples. Leen Noordzij.

An improved lower-bound on the counterfeit coins problem

and decompose in cycles of length two

Team. Outline. Statistics and Art: Sampling, Response Error, Mixed Models, Missing Data, and Inference

Statistical Properties of the OLS Coefficient Estimators. 1. Introduction

Complex Numbers. x = B B 2 4AC 2A. or x = x = 2 ± 4 4 (1) (5) 2 (1)

Introduction to Interfacial Segregation. Xiaozhe Zhang 10/02/2015

Foundations of Arithmetic

Chapter 6 The Effect of the GPS Systematic Errors on Deformation Parameters

Some basic inequalities. Definition. Let V be a vector space over the complex numbers. An inner product is given by a function, V V C

Linear Approximating to Integer Addition

Additional File 1 - Detailed explanation of the expression level CPD

2.3 Nilpotent endomorphisms

Section 3.6 Complex Zeros

Beyond Zudilin s Conjectured q-analog of Schmidt s problem

arxiv: v1 [math.co] 12 Sep 2014

Math 217 Fall 2013 Homework 2 Solutions

A METHOD TO REPRESENT THE SEMANTIC DESCRIPTION OF A WEB SERVICE BASED ON COMPLEXITY FUNCTIONS

The multivariate Gaussian probability density function for random vector X (X 1,,X ) T. diagonal term of, denoted

Estimation of Finite Population Total under PPS Sampling in Presence of Extra Auxiliary Information

Design of Recursive Digital Filters IIR

Scattering of two identical particles in the center-of. of-mass frame. (b)

Some Results on the Counterfeit Coins Problem. Li An-Ping. Beijing , P.R.China Abstract

Specification -- Assumptions of the Simple Classical Linear Regression Model (CLRM) 1. Introduction

Finding Dense Subgraphs in G(n, 1/2)

Math 261 Exercise sheet 2

Weak McCoy Ore Extensions

J. Number Theory 130(2010), no. 4, SOME CURIOUS CONGRUENCES MODULO PRIMES

Lecture Space-Bounded Derandomization

THE CHINESE REMAINDER THEOREM. We should thank the Chinese for their wonderful remainder theorem. Glenn Stevens

Solution of Linear System of Equations and Matrix Inversion Gauss Seidel Iteration Method

A Study of Quantum Strategies for Newcomb s Paradox

Restricted divisor sums

Start Point and Trajectory Analysis for the Minimal Time System Design Algorithm

Bezier curves. Michael S. Floater. August 25, These notes provide an introduction to Bezier curves. i=0

a new crytoytem baed on the dea of Shmuley and roved t rovably ecure baed on ntractablty of factorng [Mc88] After that n 999 El Bham, Dan Boneh and Om

A FORMULA FOR COMPUTING INTEGER POWERS FOR ONE TYPE OF TRIDIAGONAL MATRIX

On the average number of divisors of the sum of digits of squares

= = = (a) Use the MATLAB command rref to solve the system. (b) Let A be the coefficient matrix and B be the right-hand side of the system.

Some congruences related to harmonic numbers and the terms of the second order sequences

Root Locus Techniques

The Expectation-Maximization Algorithm

A Computational Method for Solving Two Point Boundary Value Problems of Order Four

Anti-van der Waerden numbers of 3-term arithmetic progressions.

General viscosity iterative method for a sequence of quasi-nonexpansive mappings

Method Of Fundamental Solutions For Modeling Electromagnetic Wave Scattering Problems

NUMERICAL DIFFERENTIATION

Ballot Paths Avoiding Depth Zero Patterns

Appendix B. Criterion of Riemann-Stieltjes Integrability

Example: (13320, 22140) =? Solution #1: The divisors of are 1, 2, 3, 4, 5, 6, 9, 10, 12, 15, 18, 20, 27, 30, 36, 41,

Randić Energy and Randić Estrada Index of a Graph

( N) Chun-Xuan Jiang. P. O. Box 3924, Beijing , P. R. China

Chapter 12. Ordinary Differential Equation Boundary Value (BV) Problems

8.6 The Complex Number System

n α j x j = 0 j=1 has a nontrivial solution. Here A is the n k matrix whose jth column is the vector for all t j=0

Complex Numbers Alpha, Round 1 Test #123

Introduction to Vapor/Liquid Equilibrium, part 2. Raoult s Law:

Difference Equations

Batch RL Via Least Squares Policy Iteration

A new Approach for Solving Linear Ordinary Differential Equations

A combinatorial problem associated with nonograms

Zhi-Wei Sun (Nanjing)

U.C. Berkeley CS294: Spectral Methods and Expanders Handout 8 Luca Trevisan February 17, 2016

On Partial Opimality by Auxiliary Submodular Problems

Y. Guo. A. Liu, T. Liu, Q. Ma UDC

AGC Introduction

Generalized Linear Methods

International Mathematical Olympiad. Preliminary Selection Contest 2012 Hong Kong. Outline of Solutions

Power-sum problem, Bernoulli Numbers and Bernoulli Polynomials.

Another converse of Jensen s inequality

College of Computer & Information Science Fall 2009 Northeastern University 20 October 2009

ON A DETERMINATION OF THE INITIAL FUNCTIONS FROM THE OBSERVED VALUES OF THE BOUNDARY FUNCTIONS FOR THE SECOND-ORDER HYPERBOLIC EQUATION

3.1 Expectation of Functions of Several Random Variables. )' be a k-dimensional discrete or continuous random vector, with joint PMF p (, E X E X1 E X

Errors for Linear Systems

1 GSW Iterative Techniques for y = Ax

PHYS 705: Classical Mechanics. Calculus of Variations II

On the SO 2 Problem in Thermal Power Plants. 2.Two-steps chemical absorption modeling

Section 8.3 Polar Form of Complex Numbers

Lecture 13 APPROXIMATION OF SECOMD ORDER DERIVATIVES

Week3, Chapter 4. Position and Displacement. Motion in Two Dimensions. Instantaneous Velocity. Average Velocity

Solutions to exam in SF1811 Optimization, Jan 14, 2015

VARIATION OF CONSTANT SUM CONSTRAINT FOR INTEGER MODEL WITH NON UNIFORM VARIABLES

Lecture 10 Support Vector Machines II

A Simple Proof of Sylvester s (Determinants) Identity

A Result on a Cyclic Polynomials

Problem #1. Known: All required parameters. Schematic: Find: Depth of freezing as function of time. Strategy:

Binomial transforms of the modified k-fibonacci-like sequence

1 cos. where v v sin. Range Equations: for an object that lands at the same height at which it starts. v sin 2 i. t g. and. sin g

arxiv: v6 [math.nt] 23 Aug 2016

A New Upper Bound on 2D Online Bin Packing

Chapter 5. Solution of System of Linear Equations. Module No. 6. Solution of Inconsistent and Ill Conditioned Systems

Variable Structure Control ~ Basics

Calculating Jacobian coefficients of primitive constraints with respect to Euler parameters

Lectures - Week 4 Matrix norms, Conditioning, Vector Spaces, Linear Independence, Spanning sets and Basis, Null space and Range of a Matrix

A summation on Bernoulli numbers

Transcription:

Improvement on Warng Problem L An-Png Bejng, PR Chna apl@nacom Abtract By a new recurve algorthm for the auxlary equaton, n th paper, we wll gve ome mprovement for Warng problem Keyword: Warng Problem, Hardy-Lttlewood method, recurve algorthm, auxlary equaton

Introducton Warng problem now to fnd G ( ), the leat nteger, uch that each uffcent large nteger may be repreented a um of at mot th power of natural number The Hardy-Lttlewood method, that, o called crcle method man analy method, whch propoed by Hardy Ramanujan and Lttlewood n about, whch have been appled uccefully n olvng ome problem of number theory, eg Warng problem and Goldbach problem The nown bet reult for Warng problem up to now are a followng For uffcently large (Wooley [] ), G ( ) log( log ) O () And for maller, G(), G(), G(), G(), () For the detal referred to ee the Vaughan and Wooley urvey paper [] In th paper, by a new recurve algorthm, we wll gve ome mprovement for G ( ) Theorem For uffcently large, G ( ) o ( ) () Theorem For, let F ( ) be a n the Lt, then G ( ) F ( ) () F ( ) F ( ) F ( ) F ( ) Lt In ecton, there further progre

The Proof of Theorem Suppoe that P a uffcent large nteger, C ( P) a ubet of [, P ], a gven nteger, conder the equaton x x y y, x, y C ( P), () Denoted by S ( C ( P)) the number of oluton of (), or mply S ( P ), when the electon C ( P) clear n context The equaton above called auxlary equaton of Warng problem In the followng, we wll tae ue of teratve method to contruct C ( P) Suppoe that a real number, Let P P, P a et of prme number p n nterval [ P /, P ], wrte P Z, defne C( P ) xp xc ( P), pp () Wth repect to the contructon, we wll alo conder followng a relatve equaton p x x y yq x x y y, x, y C ( P), where p, qp, p q Denote by T, ( p, q) the number of oluton of (), and T, ( q) T, ( p, q) p () Lemma For nteger,, t ha S P Z T q () ( ), ( ) Proof A uual, wrte ex ( ) e x, let f( ) e x, f(, p) e p x, f( ) e y xc ( P) xc ( P) yc ( P) Then clearly, f ( ) e p x f(, p) pp xc ( P) pp Applyng Hölder nequalty, t ha S ( P ) f( ) d f( ) f( ) d pp qp f(, p) f(, q) d

d pq pq pq, P, pq ( ) ZS( P) Z f(, pf ) (, q) d pq, P, pq Z f( ) d ( f(, p) f(, q) ) d It clear that Z S ( P ) mnor for S ( P ) Moreover, for a non-negatve nteger, let (,, ) (, ) p q f p f(, q), then by Cauchy nequalty, t ha f (, p) f(, q) d (, p, q) (, q, p) d (, pq, ) d (,, ) q p d / / And / / f(, p) f(, q) d (,, ) (,, ) p q d q p d pq, P, pq pq, P, pq pq, P, pq (, pq, ) d Clearly, the nner ntegral the number of oluton of equaton () Denote by P [ ab, ] the et of prme number n the nterval[ ab, ] Smlar to contruct (), let,, be real number, /,, whch wll be determned later, and let Z P, P P [ Z /, Z], P P / Z, H P/ Z, Recurvely defne C ( P) x p xc ( P ), pp,( p, x),,, Smply wrte C C ( P),,,, In the followng, t wll be ued the notaton of dfference of a functon: A uual, for an nteger coeffcent of polynomal ( x), recurvely defne the forward dfference ( ( x ), t ) ( xt ) ( x ), ( ( x), h, h,, h, h ) ( ( x), h, h,, h), h,, Suppoe that t h m, m a contant, then we now that m ( ( x t) ( x)), n th cae we

defne modfed dfference ( ( ), ; ) ( ) ( ) x hm m x hm x ( ( x), h,, h, h ; m,, m, m ) ( ( x), h,, h; m,, m), h ; m Smply wrte, ( x, h,, h; p,, p ) ( x, h,, h; p,, p ),,, And defne f ( ) ex ( ), f(, p) epx ( ), pp, xc xc,( x, p) g ( ; h,, h; p,, p ) e( ( x) ), x And () () hh hh pp pp F(, q) g ( q ) F (, q) g ( q ) hh hh pp pp F (, q) g ( q ) Let p hh hh pp pp ( ) f(, p) f(, q) d ( ) () f ( ) F (, q) d, ( ) f ( ) F(, q) d, () () () Lemma ( ) () pc f (, p) F (, q) d ( Z ) P Z H S ( P ) P( Z ) P () Proof A uual, for a number x, denote by x wth that x x mod p Smply wrte ( x) ( x), and For a p P, let, p p p P p P h h h D(, p) x xc, ( x) mod p, ( p) C \ D (, p) Then the um of g ( q ) can be dvded two part, that, normal and ngular part, e one

wth ( x, y) ( p), and the other one not It not dffcult to demontrate that the ntegral n the ngular part econdary, for the mplcty, we ave the nvetgaton, and n the followng acquece n the normal part It clear that for equaton ( x) ( y) n mod p there are at mot O( p ) oluton wth that ( x, y) ( p), hence we can dvde ( p) nto ( ) O p, ayl clae, F p ( ) l, uch that n each cla F ( p) the equaton ha at mot two oluton mod p And denote by ( x) card y ( x, y) F ( p) It clear that j j j ( x) P Hence, t ha jl ( ) () ( ) (, ) (, ) (, ) (, ) pc p h pc ( ) f (, p) e(( ( x) ( y)) q ) d p h pc jl ( x, y) Fj( p) ( ) l f (, p) e(( ( x) ( y)) q ) d p h pc jl ( x, y) F j ( p) f p F q d f p g q d l f (, p) ( x) e( ( ( x), h; p) q ) l p h ( ) Fj j pc jl h x e( ( ( x), h; p) ( ( y), h ; p)) q ) d ( x, y) F j h, h ( ) l f (, p) Fj( p) d p h pc jl p l f (, p) e( ( ( x), h; p) ( ( y), h ; p)) q ) d p p l f (, p) ( ) f (, p) j( x) e( ( ( x), h; p)) q ) d h pc jl h x ( ) h pc jl ( x, y) Fj h, h ( ) lp Z f ( ) d h ( ) pc h x ( ) p h pc h x lp Z Z H S( P ) ( ) jl p h pc h x P e( ( ( x), h; p)) q ) d l f ( ) e( ( ( x), h; p) q ) d P f ( ) e( ( ( x), h; p)) q ) d lh p h ( ) f (, p) e( ( ( x), h; p) q ) d p h pc h x

lp Z ZHS( P ) Pl P From the proof above, we can now that, p ( ) T ( q) f(, p) f(, q) d Z P ZS( P) Z P ( P) P () Bede, / / ( ) ( ) ( ) f( ) F(, q) d f( ) d f( ) F(, q) d S P H Z / / / ( ) ( ) () where H H, Z Z j j j j In general, we have Lemma U V W () U S ( P) Z ( Z H ) ( lp Z Z H S ( P )), / ()/ / / V S ( P) / ()/ / / Z ZH Pl ( ) ( ), W S ( P) Z ( Z H ) ( P ) / ()/ / / Proof / / ( ) () ( ) ( ) () f ( ) (, ) ( ) ( ) ( (, )) F q d f d f F q d ( ) / () S ( P) f (, p) ( F (, q)) d pp / ( ) () S( P) Z f (, p) ( F (, q)) d pp / ( )/ ( ) () S( P) Z Z H f (, p) F (, q) d pp / ()/ / S ( P) Z ( Z H ) ( lp Z Z H S( P )) ( Pl ) ( P ) U V W / / / /

Let U V H P W H P, () / / / /,, wll be decded later Hence, t ha / ( Z) P ZZ HS( P) / / / / / / / / / PZ ( ) S ( P) ( HZ ) ( H) P ( P) ( H) P And t follow / / / / / ( Z) P ZZ H ( ) ( ) S P P H P / P ( H ) P That, / / / ( Z ) ZZ H S ( P) P( H) P () On the other hand, U V W U ( H ) P / / S ( P) Z ( Z H ) ( lp Z Z H S ( P )) ( H ) P / ()/ / / / / Combne the two equalte, ( Z ) Z Z H S ( P) P( H ) P S ( P) Z ( Z H ) ( lp Z Z H S ( P )) ( H ) P It follow / ()/ / / / / / ()/ / / ( ) ( ) ( ) ( ) / / ( Z ) ( H ) P S P Z H l H S ( P ) e () S ( P) Z ( H ) ( Z ) ( H ) () S P Z H P () ( ) ( ) ( ) Let ( ) S X X, () become P ( P/ Z ) ( P/ Z ) ( Z ) P Z P () Z ( ) e ( ) Z Z P Z Z And ( ) ( ) ( ) ( ) ( ) ( )

Denote by a () ( ), b, t ha ( ) j j a a j b a, and a ab a Bede, by (), t ha / / / / ( Z ) P Z Z H S ( P ) P ( H ) P It eay to now that Z HPS ( ) P, o t ha ( Z ) P Z Z H S ( P ) PZ H P S ( P )( H ) P And ( H ) P ( P/( Z) ) P ( ) Defne (), there a ab a a ( ) ( ) ( ) ( ) ( ) Epecally, a ( ) ( ) ( ) ( ) ( ) On the other hand, by Lemma wth, t ha S ( P ) Z ( P) Z U H P Z Z P ZS ( P) H P and / / / / ( )/ ( / ) ( ) ( ) Or, () ( ) () ( /) () ( ) ( ) ( ) When greater, t may ha ( ) () ( ) ( ) And let ( ) ( ) ( ), ( ) () Subttutng () and () n (), t follow

() ( ), ( ( ) ) ( ) where ( ) And then () ( ) ( ( ) ) ( ( ) ) So, ( d ) ( d) ( ( ) ) () d, or, or other It nown that,, e (), () Hence, ( ), or ( ( ) ) Let (, ) log log ( ) ( ( ) ) () From (), we can now that ( ) wll approach zero a tend to (, ) Hence t ha Lemma For uffcent large, and arbtrary mall, there ( (, ) ) uch that () Bede, for greater, t ha ( ( ) ) (, ) log a () Moreover, we now that (ee [] or []) ( u) G ( ) u ˆ where log( / ) log log ˆ, log log log O ( ) log Tae u ( (, ) ), and let, Theorem proved

The proof of Theorem For the maller, and () may be followed by recuron () and () from the ntal d, d,, or any nown better d by choong optmal value and n turn Wth nvolve earchng of two parameter and, and the retrcton of the ablty of PC, we have leen the earch range only n four dgt The reult n Lt are obtaned by PC () () Lt For, let ( ) f a power of, or /ele From the nown reult (ee [], []), we now that for two potve nteger tv,, f atfyng ) tv ( ), ) v ( t), then Wth Lt, we tae v a n the followng lt G ( ) t v () v( ) Lt And Theorem followed Further Improvement

Shortly after the paper appeared, we realze that the method of parameterzed recuron appled n the ecton, alo avalable for the recurve proce appled n paper [],and t unexpected that the reult are even better than the prevou one, the new reult are that Theorem For uffcent large, G ( ) o( ), f not a power of, ele () Theorem For, let F ( ) be a n the Lt, then G ( ) F ( ) () F ( ) F ( ) F ( ) F ( ) Lt The Proof of Theorem : Let, ( ) ( ) J T ( q) f(, p) f(, q) d, p J f ( ) F(, q) d, () Lemma of paper [] wll be ued n the followng proof, we retate here Lemma J ZPS ( P) J, J U V, U S ( P) Z P( H Z ) Z S ( P ) / ()/ V S ( P) Z ( H Z J ),, / ()/ / Where H H, Z Z j j j j / ()

Let U V H P,,, () / /, The parameter, wll be decded later Hence, t ha PH ( Z ) ZS ( P) ( H Z J) H P And J P( H Z ) Z S ( P) H P () On the other hand, / J U H P S ( P) Z P( H Z ) Z S ( P ) H P () / / / ()/ / / Combne the two equalte, t ha / PH ( Z ) ZS ( PH ) P S ( P) Z PHZ ( ) Z S ( P ) H P It follow, / ( )/ / / S ( P) Z ( H ) H, S P P H P / ()/ / / / / ( ) and S ( P) Z ( H ) H () S P P () ( ) Let ( ) S X X, t ha P ( P/ Z ) ( P/ Z ) P () Z P e ( Z ) Z P Z Z and ( ) ( ) Denote by a (() ) () ( ) ( ) (), b, t ha ( ) j j a a j b, and a Moreover, by () wth, t ha a ab a ()

/ PH ( Z ) ZS ( P) ( H Z J) H P / / / And t eay to now that J PH Z S ( ) P, hence PH ( Z ) ZS ( P) ( H Z PHZS ( P)) H P H P P H ( P/ Z ) ( ) Z P ( ) Let () defned a before, there a ab a a ( ) ( ) ( ) ( ) ( ) Epecally, a ( ) ( ) ( ) ( ) ( ) On the other hand, by Lemma wth, t ha S ( P ) Z J ( P) Z U H P Z PZS ( P) H P And / / / / ( / ) ( / / ) ( ) ( ) ( ) Or, () ( ) ( /)( ) () () ( ) ( ) ( ) When greater, t may ha ( ) () ( ) ( ) And let ( ) ( ) ( ), ( ) () Subttutng () and () n (), t follow () ( ) () ( ( ) ) ( ( ) ) where ( ) And then

() ( ) ( ( ) ) ( ( ) ) For (), hence ( ), () ( ( ) ) Let (, ) log log ( ) ( ( ) ) () From (), we can now that ( ) wll approach zero a tend to (, ) Hence t ha Lemma For uffcent large, and arbtrary mall, there (, ) uch that () Clearly, for greater, there (, ) log () Denote by x /, and let, whch a root of the equaton x log x x x Hence, (, ) / log, a x We now that ( u) G ( ) u ˆ where e log( / ) log log ˆ, log log log O ( ) log G ( ) u( o()) ( u ) log Tae u (, ), and let x, Theorem proved

The Proof of Theorem : For maller, () may be followed by recuron () and () and the ntal () by choong optmally value and n turn Wth the retrcton of the ablty of PC, we have leened the earch range of parameter, only n fve dgt The reult n Lt are obtaned by PC A completed lt ncludng ntermedate reult poted behnd a appendx () () Lt A the proof of Theorem, wth () and Lt, we tae v ( ) a n the followng lt v( ) Lt And Theorem followed Further Improvement () In th ecton, we wll preent further mprovement when larger Theorem For uffcent large, G ( ) o( ), f not a power of,, otherwe ()

The Proof of Theorem : The notaton and ymbol ued here wll be ame a before Defne F (, q) e(( ( x) ( y)) q ), () () hh hh pp pp x, y F (, q) e(( ( x) ( y)) q ), () () hh hh pp pp xy Let ( ) J f(, p) f(, q) d, p ( ) J f ( ) F(, q) d, There () Lemma J ZPS ( P) J, J U V, U P HZS ( P), V S ( P) Z ( HZJ ),, / / ()/ / where H H, Z Z j j j j () Proof By Cauchy nequalty, ( ) ( ) ( ) ( ) (, ) ( ) ( ) (, ) / / ( ) ( ) () f ( ) d H Z f ( ) (, ) F q d / / J f F q d f d f F q d ( ) / ( ) ( ) () f( ) d ( HZ ) f( ) ( ) ( ) (, ) PHZ d f F q d S ( P) ( H Z ) / / / / ( ) / / () S ( P) ( H Z ) P( H Z ) S ( P) f (, p) F(, q) d pp ( ) () PHZ ( ) S ( P) Z f (, p) F(, q) d pp P H Z S ( P) S ( P) ( Z ) ( H Z ) f ( ) F (, q) d / / / / ( ) P H Z S P S P Z / / / / / ( ) ( ) ( ) ( HZ ) ( J ) / / / Let

( ) / /, U V H P The parameter, wll be determned later So, t follow () Z () PH ( Z ) S ( P ) J ( H ) P () Bede, J U ( H ) P P H Z S ( P)( H ) P () / / / / / Combne the two dentte above, t follow S ( P ) P H ( H ) () S P H P e / / () Z / ( ) ( ) ( Z ) Z P ( P/ Z ) ( P/ Z ) Or, ( ) ( )/ / ( () ( )) ( ) / / P P P And, (() ( )) ( ) () ( ) ( ) Denote by a, b Then t ha j j a a a b, and a b a a j Moreover, by () wth, t ha PH ( Z ) S ( P ) S ( P ) Z ( H Z J)( H) P e ( ) PH Z S P Z J H P ( ) ( ) ( ) ( ) It eay to now that J PH Z S ( ) P, hence PH ( Z ) S ( P ) Z PHZS ( P)( H) P And S S e ( ) ( P ) ( ) Z H( H) P ( P ) ( ) ( ) ( ) Z P

It follow () ( ) Hence, ( ) ( ) ( ) a ( ) ( ( ) )( ( ) ) Epecally, ( ) ( ) ( ) a ( ) ( ( ) )( ( ) ) Furthermore, by Lemma wth, t ha S ( P ) Z J ( P) Z U H P Z PZS ( P) H P And / / / / ( / ) ( / / ) ( ) ( ) ( ) e () ( ) ( /)( ) () () ( ) ( ) ( ) Let ( ) ( ), ( ) () Then t follow ( ) ( ) () ( ), () ( ( ) ) ( ( ) ) where ( ), And then ( ) ( ) () ( ) ( ( ) ) ( ) ( ) ( ) ( ( ) ) e ( ) ( ) ( ) ( d ) ( d) () ( ) ( ( ) ) ( )

Let ( ) ( ) (, ) log log ( ( )) ( d) ( ) ( ( ) ) () From (), we can now that ( d) wll approach zero a tend to (, ) Hence t ha Lemma For uffcent large, and arbtrary mall, there (, ) d uch that () It clear that, when greater, ( ) ( ) ( d) (, ) log () ( ) ( ) We chooe ( d) uch that ( d) /, then we can tae, and Hence, t ha o( ) And then let, by (), t follow (, ) ( d) () On the other hand, from (there tae, ), we can now that d log () We have nown that (ee []) G ( ) u( o()) ( u ) log Tae u d (, ), Theorem proved In fact, Theorem can be alo proved along the way of ecton a followng The Second Proof of Theorem : By () and (), there ( ) ( ) () ( ) ( ( ) ) ( ( ) ) Where ( ) ( ),,,

And then, ( ) ( ) ( ) ( d ) ( d) () ( ) ( ( ) ) ( ) Let ( ) ( ) (, ) log log ( d)( ( ) ) ( ) ( ( ) ) () From (), we can now that ( d) wll approach zero a tend to (, ) Clearly, when greater, ( ) ( ) ( d) (, ) log () ( ) ( ) We chooe ( d) uch that ( d) /, then t can be taen that, and o( ), and then let, by (), t follow (, ) ( d) The ret ame a the frt proof Tae

Reference AP L, A note on Warng Problem, arxv: RC Vaughan, The Hardy-Lttlewood method, Cambrdge Unverty Pre, RC Vaughan, A new teratve method n Warng problem, Acta Math (), - RC Vaughan and TD Wooley, Warng problem: A urvey, Number Theory for the Mllennum III, A K Peter,, pp TD Wooley, Large mprovement n Warng problem, Ann of Math (), -

Appendx Intermedate Reult of Recuron for Theorem () - - - - - - - - - - - - - - - - - - - -

- - - - - - - - - - - - - - - - - - - -

- - - - - - - - - - - - - - - - - - - - - - - -

- - - - - - - - - - - - - - - - - - - - -

- - - - - - - - - - - - - - - - - -

- - - - - - - - - - - - - - - - - - - - - - - - -

- - - - - - - - - - - - - -