Biases among Congruence Classes for Parts in k-regular Partitions

From MaRDI portal
Publication:6404493

arXiv2207.04352MaRDI QIDQ6404493FDOQ6404493


Authors: Faye Jackson, Misheel Otgonbayar Edit this on Wikidata


Publication date: 9 July 2022

Abstract: For integers k,tgeq2 and 1leqrleqt let Dk(r,t;n) be the number of parts among all k-regular partitions (i.e., partitions of n where all parts have multiplicity less than k) of n that are congruent to r modulo t. Using the circle method, we obtain the asymptotic [ D_{k}(r,t;n) = frac{3^{frac{1}{4}}e^{pisqrt{frac{2Kn}{3}}}}{pi t 2^{frac{3}{4}}K^{frac{1}{4}}n^{frac{1}{4}}sqrt{k}}left(log k + left(frac{3sqrt{K}log k}{8sqrt{6}pi} - frac{tpi(k-1)K^{frac{1}{2}}}{2sqrt{6}}left(frac{r}{t}- frac{1}{2} ight) ight)n^{-frac{1}{2}} + O(n^{-1}) ight), ] where K=1frac1k. The main term of this asymptotic does not depend on r, and so if Pk(n) is the total number of parts among all k-regular partitions of n, we have that fracDk(r,t;n)Pk(n)ofrac1t as noinfty. Thus, in a weak asymptotic sense, the parts are equidistributed among congruence classes. However, inspection of the lower order terms indicates a bias towards the lower congruence classes; that is, for 1leqr<sleqt we have Dk(r,t;n)geqDk(s,t;n) for sufficiently large n. We make this inequality explicit, showing that for 3leqkleq10 and 2leqtleq10 the inequality Dk(r,t;n)geqDk(s,t;n) holds for all ngeq1 and the strict inequality Dk(r,t;n)>Dk(s,t;n) holds for all ngeq17.




Has companion code repository: https://github.com/fayealephnil/kregularbiases









This page was built for publication: Biases among Congruence Classes for Parts in k-regular Partitions

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6404493)