Year-end appeal: Please make a donation to the OEIS Foundation to support ongoing development and maintenance of the OEIS. We are now in our 61st year, we have over 378,000 sequences, and we’ve reached 11,000 citations (which often say “discovered thanks to the OEIS”).
%I #29 Dec 19 2024 11:43:26
%S 1,1,2,4,6,9,13,18,26,36,49,66,88,117,154,202,262,338,435,555,705,891,
%T 1122,1407,1757,2185,2709,3347,4121,5060,6194,7561,9205,11178,13540,
%U 16362,19727,23732,28490,34132,40810,48701,58011,68977,81874,97019,114778,135573
%N Rademacher's partition formula extended to half-integers. a(n) = round(sqrt(48) * (cosh(h(n)) - sinh(h(n))/h(n)) / (24*n + 11)) where h(n) = sqrt(24*n + 11)*(Pi/6).
%C One could say that the sequence gives the "number of partitions of n + 1/2", but that would be a linguistic overstretching. However, it does address Richard Stanley's question: "Let n be any complex number in Rademacher's convergent infinite series for p(n) [the number of partitions of n]. For what n does it converge?"
%C In an answer, Fredrik Johansson points to a cosine-extended version of p(n) that is real-valued on the real line and for which the following also applies: "At half-integers, it appears that all terms in the cosine version of the Rademacher series except the first term vanish, and so one has a trivial closed-form evaluation of p(n + 1/2), n in Z." It is this formula on which the sequence is based.
%H Fredrik Johansson, <a href="https://doi.org/10.1112/S1461157012001088">Efficient implementation of the Hardy-Ramanujan-Rademacher formula</a>, LMS Journal of Computation and Mathematics, 15:341-59, (2012), MR 2988821, S2CID 16580723.
%H Fredrik Johansson, <a href="https://fredrikj.net/blog/2014/03/new-partition-function-record/ ">New partition function record: p(10^20) computed</a>, (March 2014).
%H Fredrik Johansson, <a href="/A376876/a376876.pdf">Python program: Rademacher's partition formula, cos case</a>.
%H Hans Rademacher, <a href="https://doi.org/10.1112/plms/s2-43.4.241">On the partition function p(n)</a>, Proceedings of the London Mathematical Society, Second Series, 43 (4):241-254, (1937), MR 1575213.
%H Richard Stanley, <a href="https://mathoverflow.net/q/366733">Does Rademacher's convergent series for p(n) define an analytic function?</a>, and the <a href="https://mathoverflow.net/q/366805">Answer</a> by Fredrik Johansson, (July 2020).
%H Wikipedia, <a href="https://en.wikipedia.org/wiki/Partition_function_(number_theory)#Approximation_formulas">Partition function, approximation formulas</a>.
%F a(n) ~ exp((Pi*sqrt(6*n + 3) * (Pi^2*(48*n + 23) - 72) + 6*Pi^2 - 108) / (72*Pi^2*(2*n + 1))) / (sqrt(12)*(2*n + 1)).
%e The sequence p(n), {a(n)}, p(n+1), ... (where p(n) = A000041(n)) starts:
%e 1, {1}, 1, {1}, 2, {2}, 3, {4}, 5, {6}, 7, {9}, 11, {13}, 15, {18}, 22, {26}, 30, {36}, 42, {49}, 56, {66}, 77, {88}, 101, ...
%e In turn, the rounded arithmetic mean q(n) = round((a'(n) + a'(n-1))/2), (where a'(n) is a(n) before rounding) can be seen as a simple approximation to the partition numbers, q(n) ~ p(n). For example p(10^20) = 1.838176508344882643646... *10^11140086259, q(10^20) = 1.838176508344882643649... *10^11140086259.
%e The sequence q(n) starts: 1, 1, 2, 3, 5, 7, 11, 16, 22, 31, 42, 57, 77, 103, 136, 178, 232, 300, ... For comparison, see the rounded values of the Hardy-Ramanujan approximate formula A050811.
%e The unrounded value of a(0) is 0.8458241... = A376875.
%p ap := proc(n) local h; h := sqrt(24*n + 11)*(Pi/6):
%p sqrt(48)*(cosh(h) - sinh(h)/h)/(24*n + 11) end:
%p seq(round(evalf(ap(n), 64)), n = 0..47);
%t s[x_] := 1/E^x + E^x + 1/(E^x x) - E^x/x;
%t h[x_] := Sqrt[11 + 24 x] Pi / 6; g[x_] := Sqrt[12]/(11 + 24 x);
%t Table[Round[g[n] s[h[n]]], {n, 0, 47}]
%o (Python) # See links.
%Y Cf. A000041, A050811, A376875.
%K nonn
%O 0,3
%A _Peter Luschny_, Oct 07 2024
%E Dedicated to _N. J. A. Sloane_ on the occasion of his 85th birthday.