Arithmetic Density and Related Concepts: 1 Counting Arithmetic Progressions (Aps)

Arithmetic Density and Related Concepts
J. R. Wheat
06/18/2023
1 Counting Arithmetic Progressions (APs)

Given an arbitrary increasing sequence of positive integers S there is associated
with it, a counting function αS (n) which determines the number of arithmetic
progressions[1] up to sn ∈ S. The counting function for the positive integers,
and consequently all infinite arithmetic progressions, is given by the formula
listed under A330285 (OEIS)[2] :
n Xk
X k−1
.
j=1
j+1
k=1
We may further generalize from this AP counting function, αZ+ , the following
formula which applies to any arbitrary integer sequence,
n
X
αZ+ (k)π(n, k)
k=1
where π(n, k) is the number of primitive progressions of length k, up to the n-th

element of S. A primitive progression is defined as any subset
{s, s + d, . . . , s + (k − 1)d} ⊂ {s1 , s2 , . . . , sn }
such that
{s − d, s + kd} ∩ {s1 , s2 , . . . , sn } = ∅,
{s, s + d, . . . , s + (k − 1)d} ̸⊂ {s′ , s′ + d′ , . . . , s′ + (j − 1)d′ }
for k − 1 ≤ j and d′ < d.
2 Relative Arithmetic Density

There are two other uses for αZ+ which will be discussed in both this section
and the following section, respectively. The first such use involves computing
the relative arithmetic density of our sequence S, but before proceeding it is
necessary that we elaborate on the formulation of ”partial” densities.
1
Let us begin by observing that the number of APs up to sn is merely a
fraction of all sub-sequences over said interval. The partial density DS (n) is
thus given by
αS (n)
2n − Tn − 1
where Tn is the n-th triangular number and the denominator counts sub-sequences
with more than two elements. The infinite sum of partial densities over Z+
is approximately 2.89563562435821120303, and will be referred to as Layman’s
constant[3] . The exact value of the infinite series of partial densities requires fur-
ther analysis, in the meantime we can still use Layman’s constant to construct
a definition for relative densities in general by comparing arbitrary number se-
quences with the positive integers themselves. The relative arithmetic density
is thus described by the ratio:
∞
X
DS (n)
n=1
D(S) = ∞ ,
X
DZ+ (n)
n=1
which is just some real number less than or equal to one.
3 Natural Arithmetic Density

The second use for these counting functions involves a more naturalistic inter-
pretation of arithmetic density, as far as real number valuations are concerned.
We can arrive at such a definition by taking advantage of the fact that the
counting function for any infinite progression has maximal growth. It then be-
comes an arbitrary choice of which complementary sequence we might use to
construct a suitable infinite series for our valuation. The author has chosen the
sequence of factorials because firstly they converge for all AP counting func-
tions, and secondly because they allow for more convenient algebraic analysis.
The formula for the natural arithmetic density of S is then defined to be
∞
X αS (n)
δ(S) = .
n=1
(n − 1)!
This series is approximately 1.42638919894259387857 when the counting func-

tion is taken over any infinite AP. For the rare case when there is only one
progression of length three at the beginning of the given sequence, the series
above is equal to e − 2.
2
4 The Partitive Convergence Theorem
4.1 Proof of the Theorem
Assuming that the infinite series
∞
X 1
= ∞,
s
n=1 n
then there exists some sub-sequence {s′1 , s′2 , ... } such that
n1 n2 n3
X 1 X 1 X 1
< < < ...
sk sk sk
k=n1 k=n1 +1 k=n2 +1
and
n1 2 −1
nX 3 −1
nX
X 1 1 1
> > > ... ,
sk sk sk
k=n1 k=n1 +1 k=n2 +1
for n1 = 1 and s′m = snm . We can deduce that this sequence of finite series
must be constructable, otherwise the infinite sum of reciprocals would not di-
verge. What is not immediately evident, however, is that each of these finite
series should converge to the same value from below. We denote this divergence
constant as C and we will further expand on its significance in the following
subsection.
In order to establish our proof, let us first draw our attention to the terminal
values at the end of each finite series. Without the addition of these single
terms, any particular series falls short of those before it according to the second
inequality above. Therefore, if all of the finite series are to diverge from any
particular constant, that would imply each terminal value is greater than the
1 ′
previous one, or s1′ < sm+1 . This cannot be the case since our infinite sub-
m
sequence is monotonically increasing, and so its reciprocal elements must tend
toward zero. As an example of the proof, let it be noted that the divergence
constant for the harmonic series is log(3).
Before leaving the present topic, there are two other constants which should
be briefly discussed for future reference. These are the criticality constant κ and
the co-criticality constant κ̄. These can be calculated using the infinite series
∞
X 1
n
m=1 m
and
∞
X 1
,
s′
m=1 m
respectively. In the case where S = Z+ these two constants are equal, and their
value is
∞
log( 32 ) − Ψ 13 (1)

X 2
=2
m=1
3m − 1 log(3)
3
where Ψq is the q-digamma function[4] .
4.2 Algebraic Derivations

A further consequence of the theorem is that partial sums of divergent series
can now be represented using the divergence constant mentioned above. To
do so requires the introduction of an error function which adjusts the constant
according to the interval over which the function is taken, i.e
nm
X 1
= C − Err(m) .
sk
k=nm−1 +1
According to this definition, it follows that the partial sum up to s′m is given by
the formula
nm m
X 1 X
= Cm − Err(k)
sk
k=1 k=1
We can then expand upon this equation by introducing balancing coefficients

βm,k for nm−1 + 1 ≤ k ≤ nm , which allow us to compute partial sums over
the m-th interval {snm−1 , snm−1 +1 , ... , snm }. These are implemented in the
following manner
k
X 1
Cβm,k − Err(m) = ,
j=n
s
+1 j
m−1
which then implies

∇n m −1 nm ∇n m −1
X X X ∇nm − j
C βm,j − (∇nm − 1)Err(m) =
j=1 j=1
sk
k=nm−1 +1
where ∇ is the backward difference operator[5] .
4
REFERENCES
[1] Weisstein, Eric W. ”Arithmetic Progression.” From Mathworld –A Wol-

fram web resource. (http://mathworld.wolfram.com/ArithmeticProgression.html)
[2] ”A330285.” The On-Line Encyclopedia of Integer Sequences. (http://oeis.org/A330285)
[3] ”A05133.” The On-Line Encyclopedia of Integer Sequences. (http://oeis.org/A051336)
[4] Weisstein, Eric W. ”q-Polygamma Function.” From Mathworld –A Wol-
fram web resource. (https://mathworld.wolfram.com/q-PolygammaFunction)
[5] Weisstein, Eric W. ”Backward Difference.” From Mathworld –A Wolfram
web resource. (http://mathworld.wolfram.com/BackwardDifference)

Arithmetic Density and Related Concepts: 1 Counting Arithmetic Progressions (Aps)

Uploaded by

Copyright:

Available Formats

Arithmetic Density and Related Concepts: 1 Counting Arithmetic Progressions (Aps)

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Arithmetic Density and Related Concepts: 1 Counting Arithmetic Progressions (Aps)

Uploaded by

Copyright:

Available Formats

Arithmetic Density and Related Concepts

1 Counting Arithmetic Progressions (APs)

where π(n, k) is the number of primitive progressions of length k, up to the n-th

{s, s + d, . . . , s + (k − 1)d} ⊂ {s1 , s2 , . . . , sn }

2 Relative Arithmetic Density

which is just some real number less than or equal to one.

3 Natural Arithmetic Density

This series is approximately 1.42638919894259387857 when the counting func-

4.2 Algebraic Derivations

We can then expand upon this equation by introducing balancing coefficients

which then implies

where ∇ is the backward difference operator[5] .

[1] Weisstein, Eric W. ”Arithmetic Progression.” From Mathworld –A Wol-

You might also like