cp's OEIS Frontend

This is a front-end for the Online Encyclopedia of Integer Sequences, made by Christian Perfect. The idea is to provide OEIS entries in non-ancient HTML, and then to think about how they're presented visually. The source code is on GitHub.

A378514 Number of partitions of 2^n-1 into two summands >= 0 having a common divisor > 1.

Original entry on oeis.org

0, 1, 1, 4, 1, 14, 1, 64, 40, 212, 56, 1184, 1, 2900, 2884, 16384, 1, 61088, 1, 284288, 159520, 776800, 89264, 5070848, 577216, 11195732, 10375168, 67834880, 1522240, 269570912, 1, 1073741824, 813199072, 2863486292, 917553184, 21299044352, 308159200, 45813683540
Offset: 1

Views

Author

Dmytro Inosov, Nov 29 2024

Keywords

Comments

a(n) counts binary numbers (A007088) of length n that are not coprime with their bitwise inverse as integers in base 2.
Equivalently, m from A007088 is counted toward a(A055642(m)) iff GCD(m, A002275(A055642(m)) - m) > 1, assuming base 2 in the calculation of GCD. Therefore a(n) is the base-2 analog of A378511.
For any n > 1, a(n) > 0 because the trivial partition A002275(n) = A002275(n) + 0 always counts toward a(n): GCD(A002275(n), 0) = A002275(n) > 1.
a(n) = 1 iff n is a Mersenne exponent (A000043). Indeed, if a partition 2^n-1 = k + m exists with GCD(k, m) = q > 1, then 2^n-1 itself is divisible by q. Whenever 2^n-1 is a Mersenne prime (A000668), this is only possible for q = 2^n-1, therefore the only such partition is the trivial one, {2^n-1, 0}. The inverse is also true. If n is not a Mersenne exponent, 2^n-1 has a nontrivial divisor q, and the partition 2^n-1 = q + (2^n-1-q) counts toward a(n) because GCD(q, 2^n-1-q) = q > 1. Therefore, a(n) > 1.

Examples

			a(2) = 1 because there are only 2 possible partitions of 2^2-1 = 3 into a sum of two nonnegative integers, namely: 3 = 3 + 0 and 3 = 2 + 1. The partition {3, 0} counts toward a(2) since GCD(3,0) = 3 > 1. The partition {2, 1} does not count since GCD(2,1) = 1.
a(4) = 4 because among the 8 possible partitions of 2^4-1 = 15 into a sum of two nonnegative integers, the summands are non-coprime in exactly 4 cases:
----------------------------------------------
partition   binary vectors       GCD
(base 10)      (base 2)
----------------------------------------------
15 = 8 + 7   {1000, 0111}   GCD(8, 7) = 1;
15 = 9 + 6   {1001, 0110}   GCD(9, 6) = 3;
15 = 10 + 5  {1010, 0110}   GCD(10, 5) = 5;
15 = 11 + 4  {1011, 0100}   GCD(11, 4) = 1;
15 = 12 + 3  {1100, 0011}   GCD(12, 3) = 3;
15 = 13 + 2  {1101, 0010}   GCD(13, 2) = 1;
15 = 14 + 1  {1110, 0001}   GCD(14, 1) = 1;
15 = 15 + 0  {1111, 0000}   GCD(15, 0) = 15.
----------------------------------------------
a(7) = 1 because 7 is a term in A000043.
		

Crossrefs

Programs

  • Maple
    a:= n-> (m-> ceil((m-numtheory[phi](m))/2))(2^n-1):
    seq(a(n), n=1..38);  # Alois P. Heinz, Nov 29 2024
  • Mathematica
    CountNonCoprimes2[i_] := Table[If[!CoprimeQ @@ #, #, ##&[]] &[{n, 2^i-1-n}], {n, 2^(i-1), 2^i-1}] // Length; Table[CountNonCoprimes2[i], {i, 25}]
    (* Version that uses the built-in EulerPhi[] function *)
    Table[Ceiling[(# - EulerPhi[#])/2] &[2^m-1], {m, 100}]
  • SageMath
    def a(n): return 2^(n-1) - euler_phi(2^n-1) / 2 if n > 1 else 0
    print([a(n) for n in range(1, 39)])  # Peter Luschny, Nov 29 2024

Formula

a(n) <= A000325(n-1) = 2^(n-1) - n + 1;
a(A000043(n)) = 1;
From Alois P. Heinz, Nov 29 2024: (Start)
a(n) = A082023(2^n-1) + signum(n-1).
a(n) = ceiling((2^n-1 - phi(2^n-1))/2). (End)
From Peter Luschny, Nov 30 2024: (Start)
a(n) = A000079(n-1) - A056742(n) for n > 1.
a(n) = 2^(n - 1) - phi(2^n - 1)/2 for n > 1. (End)

Extensions

a(32)-a(38) from Alois P. Heinz, Nov 29 2024