Erdos Discrepancy Problem

John reminded that a native of South Australia, Terence Tao, has only last year finally solved the Erdos Discrepancy Problem. Here is an excerpt from Wikipedia about the problem:

Here is a link that explains it in less mathematical terms.

Anti-divisors

I'd never heard of an anti-divisor until I reached my 24434th day on Earth. Here is my tweet for that day:

24434: 2×19×643; member of OEIS A258786: numbers \(n\) whose sum of anti-divisors is a permutation of their digits 

Of course, I needed to find out exactly what an anti-divisor was. Fortunately, I came across a subsite on OEIS that explained the concept. Here is the explanation:

The anti-divisor, or unbiased non-divisor, is a concept related very closely to the concept of prime numbers, and the concept of a divisor. 

Every integer is either prime, or has two or more prime factors, for example 61 is prime, but 63 can be written as 3 x 3 x 7. 

Every integer is then said to be the product of some factors. A divisor is a combination of these factors, for example 63 has the divisors 1, 3, 7, 9, 21 and 63, as 63 divided by any of these numbers yields an integer. 

It logically follows from this that any number that is not a divisor of an integer is a non-divisor. So, we say that the non-divisors of 63 are all the integers less than or equal to 63 except for 1, 3, 7, 9, 21 and 63. 

The definition of an anti-divisor follows on from this - an anti-divisor is a non-divisor such that doesn't divide the number in the most unbiased way possible. For example, we say 42 is an anti-divisor of 63, as 42 surrounds 63 with a gap of 21 on either side. 41 on the other hand, has gaps of 22 and 18, so the gap of 22 is larger than the 18. 41 is called a biased non-divisor of 63. 

There are two distinct mathematical definitions of anti-divisors, one for even anti-divisors, and one for odd anti-divisors. The two definitions are very similar, however we need two definitions because a even anti-divisor candidate defines exactly one number, and an odd anti-divisor candidate defines exactly two numbers. 

Both definitions do share a common feature. An anti-divisor of n must lie in the region [2, \(n\)-1]. 1 is never an anti-divisor, its exclusion is determined by the phrase 'non-divisor', and the fact that no integers are contained between two successive integers with 1 as a divisor. Also note that there are 'larger-than '\(n\)' anti-divisors, namely 2\(n\)-1, 2\(n\) and 2\(n\)+1, but these are considered to be trivial. 

For k even:

\(k\) is an even anti-divisor of \(n\) when \(n \equiv k/2 \bmod{k} \) or to put it another way, \(k\) is an even anti-divisor of \(n\) when \(k(x+1/2)=n\), for some \(x\) greater than or equal to 1. For example 10 is an even anti-divisor for 15, 25, 35, 45, 55, etc,.. and no other numbers. This is because \(10(x+1/2)=10x+5\), and so the numbers are generated over \(x>1\). Another example, 8 is an even anti-divisor for 12, 20, 28, etc..., and no other numbers. 

For k odd: 

\(k(x+1/2)\) falls between 2 integers, and we claim that both of these integers are anti-divisors for \(n\).

Hence the basic definition is the same for odd anti-divisors, except that \( k(x+1/2) \) itself is never an integer, so we define the odd anti-divisors as the two numbers surrounding \(k(x+1/2) \). So \(k\) is an odd anti-divisor of \(n\) when we have either \(k(x+1/2)-1/2\) or \(k(x+1/2)+1/2\) for some \(x\) greater than or equal to 1.  

In mod terminology, \(k\) is an odd anti-divisor of \(n\) when \(n \equiv (k-1)/2 \bmod{k} \) or \( n \equiv (k+1)/2 \bmod{k} \). So, 7 is an odd anti-divisor for 10, 11, 17, 18, 23, 24, etc.., and no other numbers. This is because we have the two generating functions \(7x+3\) and \(7x+4\). Again, 11 is an odd anti-divisor for \(11x+5\), \(11x+6\), i.e. 16, 17, 27, 28, etc..., and no other numbers. 

Putting it together 

Now we have defined an anti-divisor, let's consider an integer n, e.g.10. We then find all the anti-divisors of 10, in this case 3, 4 and 7. The pattern of anti-divisors is as random and incomprehensible as that with prime numbers. 

See another post from 28th February 2021 titled More on Anti-divisors. 

Gaps between Twin Primes

Today is a prime day for me, the prime being the larger half of the twin prime pair 24419 and 24421. Just as the gaps between primes are variable, so too are the gaps between successive twin primes. However, at certain points records are set regarding how big these gaps are. It just so happens that 24421 marks one of those points. The next prime pair is 24917 and 24919, and the gap of 496 between 24421 and 24917 sets a record. A list of the initial record intervals is attached.

24421 is a member of OEIS A036061: increasing gaps among twin primes, the largest prime of the starting twin pair. The record gaps between primes was treated in this earlier post.