Finite sums of arithmetic progressions
Shahram Mohsenipour

TL;DR
This paper provides a combinatorial proof for a generalized theorem related to arithmetic progressions and Hindman's theorem, along with bounds for finite cases.
Contribution
It introduces a new combinatorial proof for a two-fold generalization of classical theorems and establishes tower bounds for finite versions.
Findings
Purely combinatorial proof of the generalization
Tower bounds for finite cases
Extension of van der Waerden-Brauer and Hindman's theorems
Abstract
We give a purely combinatorial proof for a two-fold generalization of van der Waerden-Brauer's theorem and Hindman's theorem. We also give tower bounds for a finite version of it.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsLimits and Structures in Graph Theory · Graph theory and applications · Advanced Graph Theory Research
Finite Sums of Arithmetic Progressions
Shahram Mohsenipour
School of Mathematics, Institute for Research in Fundamental Sciences (IPM) P. O. Box 19395-5746, Tehran, Iran
Abstract.
We give a purely combinatorial proof for a two-fold generalization of van der Waerden-Brauer’s theorem and Hindman’s theorem. We also give tower bounds for a finite version of it.
Key words and phrases:
van der Waerden’s theorem, Hindman’s theorem, finitary Hindman’s theorem
2010 Mathematics Subject Classification:
05D10
1. Introduction
Let be a positive integer and let be an -term arithmetic progression with , we denote the th term of by . Now let and be two -term arithmetic progressions, we define their pointwise sum (or briefly their sum) as the -term arithmetic progression with for . Hence for the -term arithmetic progressions , their finite sum has unambiguous meaning. The following pleasant two-fold generalization of van der Waerden-Brauer’s theorem and Hindman’s theorem, can be deduced form either Furstenberg’s theorem ([3] Proposition 8.2.1) or Deuber-Hindman’s theorem [1].
For any positive integers and , if is -colored, then there exist a color and infinitely many -term arithmetic progressions , such that all of their finite sums (with no repetition) are monochromatic with the color and all the common differences of the above finite sums have also the color too.
It would be pleasant too if we have a purely combinatorial proof of such a statement avoiding topological dynamics as well as the theory of ultrafilters. In Theorems 3.1 and 3.2 of this paper we give such a proof. It is interesting to see whether the method of the proof can be generalized to give a combinatorial proof of Deuber-Hindman’s theorem [1]. We are also interested in a finite version of the above theorem. It is well known that through a compactness argument we can have a finite version. For instance we have the following theorem which is a two-fold generalization of van der Waerden’s theorem and a finite version of Hindman’s theorem.
*For positive integers and there is a positive integer such that whenever is -colored, then there exist -term arithmetic progressions such that is not bigger than and all finite sums of (with no repetition) are monochromatic with the same color.
If we denote the least such by then the proof given through the compactness argument does not give us upper bounds for . But it is not hard to see that the proof given for Theorem 3.1 can be made finitary (which may be regarded as an advantage of the proof over its counterparts using dynamical system or ultrafilters) to give us a primitive recursive upper bound for . To do so we use the finitary Hindman numbers which is a tower function [2]. However due to its iterated use of the function , it gives us an upper bound belonging to the class of WOW functions [5]. In Theorem 4.1, we do a better job by giving a different proof which uses the function just one time and thus obtaining tower bounds for . Also note that according to the Gowers elementary bounds for the van der Waerden theorem, we don’t worry about the van der Waerden part of the proof.
2. Preliminaries
Let’s fix some notations. For a positive integer put . Let be an infinite set, we denote the collection of finite nonempty subsets of by . For a finite set , denotes the collection of nonempty subsets of . Also will denote the set of all finite sums of elements of with no repetition. Let , by we mean that . We also denote the common difference of the arithmetic progression by . We use the following notation for finite sums of arithmetic progressions
[TABLE]
where . Obviously we have
[TABLE]
We define a partial ordering between -term arithmetic progressions by putting whenever for all . Let’s state van der Waerden’s theorem and van der Waerden-Brauer’s theorem [5].
Theorem 2.1** (van der Waerden).**
For positive integers and there is a positive integer such that whenever is -colored, then there is a monochromatic -term arithmetic progression . We denote the least such by .
Theorem 2.2** (van der Waerden-Brauer).**
For positive integers and there is a positive integer such that whenever is a -coloring of , then there are in such that
[TABLE]
We denote the least such by .
We will use the following strong version of Hindman’s theorem [6].
Theorem 2.3**.**
Let be an infinite strictly increasing sequence of positive integers. Let be a positive integer and be -colored. Then there are in such that whenever
[TABLE]
then is monochromatic.
We say that the two positive integers are power-disjoint, if the powers occurring in the expansions of in base are disjoint sets, more precisely if we write and , then the two sets and are disjoint. We denote the set by . We will use the following finitary version of Hindman’s theorem [2] which strengthens the Disjoint Unions Theorem. First we introduce a notation. If is a collection of pairwise disjoint sets, then will denote the set of non-empty unions of elements of .
Theorem 2.4**.**
For positive integers there is a positive integer such that for any -element set of pairwise power-disjoint positive integers, whenever is a -coloring of , then there exist and in such that and for all we have
[TABLE]
Moreover if denotes the least such , then is a tower function.
3. Purely Combinatorial Proofs
In the following theorem we give a purely combinatorial proof of the two-fold generalization van der Waerden’s theorem and Hindman’s theorem mentioned in the introduction.
Theorem 3.1**.**
Let and be positive integers. Let be a -coloring of , then there are -term arithmetic progressions such that
- (i)
,
- (ii)
there is such that for all and all we have
[TABLE]
Proof.
Let and let be a strictly increasing sequence of positive integers with . For we put
[TABLE]
Obviously is an -term arithmetic progression and we have
[TABLE]
In fact it is easily seen that for any in we have
[TABLE]
Now for we inductively define the -term arithmetic progressions so that there are such that the following two conditions are satisfied
- (a)
for all and all we have
[TABLE]
- (b)
for all in we have
[TABLE]
Suppose we have defined with the above properties. We do the job for . The second condition implies that
[TABLE]
Now by Hindman’s theorem there are in such that if we put
[TABLE]
then has a constant value on , which we denote it by . Now we set
[TABLE]
as well as we set
[TABLE]
We check the conditions (a) and (b) for . Let and , hence we have
[TABLE]
where . Suppose , from the induction hypothesis it follows that
[TABLE]
Also for we have
[TABLE]
which implies that
[TABLE]
Now putting (2) and (3) together we deduce
[TABLE]
for . This finishes the proof of the condition (a). Now we turn to checking (b). Let be in . We must show that
[TABLE]
which is equivalent to
[TABLE]
Letting , , we get and (4) becomes
[TABLE]
which is exactly our induction hypothesis. This proves the condition (b).
Now consider and recall that . By construction we have
[TABLE]
Through induced coloring, it follows from van der Waerden’s theorem that there exist and positive integers such that
[TABLE]
We define the desire arithmetic progressions as follows
[TABLE]
It is easily seen by condition (b) that Also for all and all we have
[TABLE]
This finishes the proof of Theorem 3.1. ∎
Now we turn to the two-fold generalization of van der Waerden-Brauer’s theorem and Hindman’s theorem.
Theorem 3.2**.**
Let and be positive integers. Let be a -coloring of , then there are -term arithmetic progressions such that
- (i)
,
- (ii)
there is such that for all and all we have
[TABLE]
Proof.
We start with and a strictly increasing sequence of positive integers with . For , We put . In this case for all and all we will have
[TABLE]
We prove (5) by induction on . First observe that
[TABLE]
Also for , recall the subsets in definition of the arithmetic progressions , so we have
[TABLE]
where . This proves (5). The proof now proceeds as in the proof of Theorem 3.1, in particular (1) can be proved easily for these new . Now recall so that for and we have
[TABLE]
Through induced coloring and this time using we obtain and positive integers such that
[TABLE]
Again define the desire arithmetic progressions by
[TABLE]
Thus for all we have
[TABLE]
Note that in the second and third equations from the end we have respectively used (5) and the easily checked fact . So we conclude that
[TABLE]
and the rest of the proof is the same as the proof of Theorem 3.1. ∎
4. Tower Bounds for the Finite Case
In this section we prove
Theorem 4.1**.**
For positive integers and , let be the least positive integer such that whenever is a -coloring of , then there are -term arithmetic progressions such that
- (i)
,
- (ii)
,
- (iii)
there is such that for all and all we have
[TABLE]
Then is a tower function.
Proof.
Let , we will show that . So from Gower’s elementary bounds for the van der Waerden numbers [4] and Theorem 2.4, it follows that is a tower function. Suppose that and is a -coloring of . We show that satisfies the requirements of the theorem. Put . Let be positive integers defined by . For , we define the -term arithmetic progressions as follows
[TABLE]
Clearly . We claim that for each , the positive integers are pairwise power-disjoint. Let , and with , hence . Also from and
[TABLE]
it follows that
[TABLE]
for . Now to prove the claim it would be enough to show that are pairwise disjoint. In fact we show that
[TABLE]
which easily implies the disjointness of . First observe that
[TABLE]
Also for we have
[TABLE]
thus the claim is proved. Also we have
[TABLE]
Now we define a coloring on as follows. For , we put if for all we have
[TABLE]
Obviously the number of colors is , so from it follows that there are in such that
[TABLE]
which means that for all and all we have
[TABLE]
We denote the above color by . So we have the well-defined function
[TABLE]
Now consider the following -elements set of power-disjoint (due to the claim) positive integers
[TABLE]
From we infer that there exist in and so that for all we have
[TABLE]
The desired arithmetic progressions are defined as follows. For , we set
[TABLE]
Obviously and from it is easily seen that
[TABLE]
Now for and we have
[TABLE]
where . This finishes the proof of the theorem. ∎
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[1] Walter Deuber and Neil Hindman, Partitions and sums of ( m , p , c ) 𝑚 𝑝 𝑐 (m,p,c) -sets , J. Combin. Theory Ser. A 45 (1987), no. 2, 300–302.
- 2[2] Pandelis Dodos and Vassilis Kanellopoulos, Ramsey theory for product spaces , Mathematical Surveys and Monographs, vol. 212, American Mathematical Society, Providence, RI, 2016.
- 3[3] H. Furstenberg, Recurrence in ergodic theory and combinatorial number theory , Princeton University Press, Princeton, N.J., 1981.
- 4[4] W. T. Gowers, A new proof of Szemerédi’s theorem , Geom. Funct. Anal. 11 (2001), no. 3, 465–588.
- 5[5] R.L. Graham, B.L. Rothschild, and J.H. Spencer, Ramsey theory , 2 ed., John Wiley and Sons, 1990.
- 6[6] Alan D. Taylor, A canonical partition relation for finite subsets of ω 𝜔 \omega , J. Combinatorial Theory Ser. A 21 (1976), no. 2, 137–146.
