login

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”).

a(n) counts anti-chains of size two in "0,1,2" Motzkin trees on n edges.
5

%I #47 Feb 23 2024 17:43:41

%S 0,0,1,5,23,91,341,1221,4249,14465,48442,160134,523872,1699252,

%T 5472713,17520217,55801733,176942269,558906164,1759436704,5522119250,

%U 17285351782,53977433618,168194390290,523076690018,1623869984706

%N a(n) counts anti-chains of size two in "0,1,2" Motzkin trees on n edges.

%C "0,1,2" trees are rooted trees where each vertex has outdegree zero, one or two. They are counted by the Motzkin numbers.

%C From _Petros Hadjicostas_, Jun 02 2020: (Start)

%C Let A(r,n) be the number of ordered pairs (T, s), where T is a "0,1,2" tree (Motzkin tree) with n edges and s is an r-element anti-chain in T. See Definition 42, p. 30, in Salaam (2008) but we use different notation here.

%C An r-element anti-chain in a tree is a set of r nodes such that, for every two nodes u and v in the set, u is neither an ancestor nor a descendant of v.

%C For the current sequence, a(n) = A(r=2, n) for n >= 0.

%C Let A[r](z) = Sum_{n >= 0} A(r,n)*z^n be the g.f. of the sequence (A(r,n): n >= 0) for fixed r >= 1.

%C In Theorem 44 (p. 33), Salaam proved that A[r](z) = c_{r-1} * z^(2*r-2) * L(z)^(r-1) * V(z)^r, where c_r = (1/(r + 1))*binomial(2*r, r) is the r-th Catalan number in A000108, L(z) = T(z) = 1/sqrt(1 - 2*z - 3*z^2) is the g.f. of the central trinomial numbers A002426, and V(z) = T(z)*M(z), where M(z) = (1 - z - sqrt(1 - 2*z - 3*z^2))/(2*z^2) is the g.f. of the Motzkin numbers A001006.

%C It follows (see Table 2.4, p. 39) that A[r](z) = c_{r-1} * z^(2*r-2) * T(z)^(2*r-1) * M(z)^r for fixed r >= 1.

%C For r = 1, A[r=1](z) = Sum_{n >= 0} A(r=1, n)*z^n = T(z)*M(z) = V(z) is the g.f. of the total number of vertices in all "0,1,2" trees with n edges (i.e., the g.f. of the sequence (A005717(n+1): n >= 0)).

%C For r = 2, A[r=2](z) = z^2 * T(z)^3 * M(z)^2 is the g.f. of the current sequence. (End)

%H G. C. Greubel, <a href="/A178834/b178834.txt">Table of n, a(n) for n = 0..1000</a>

%H Lifoma Salaam, <a href="https://search.proquest.com/docview/193997569">Combinatorial statistics on phylogenetic trees</a>, Ph.D. Dissertation, Howard University, Washington D.C., 2008; see Definition 42 (p. 30), Theorem 44 (p. 33), and Table 2.4 (p. 39).

%F G.f.: z^2 * M(z)^2 * T(z)^3, where M(z) = (1 - z - sqrt(1 - 2*z - 3*z^2))/(2*z^2) is the g.f. of the Motzkin numbers and T(z) = 1/sqrt(1 - 2*z - 3*z^2) is the g.f. of the central trinomial numbers.

%F D-finite with recurrence: -(n-2)*(n+2)*a(n) + (4*n^2-n-8)*a(n-1) + (2*n^2-n-12)*a(n-2) - 3*n*(4*n-3)*a(n-3) - 9*n*(n-1)*a(n-4) = 0. - _R. J. Mathar_, Jun 14 2016

%F a(n) ~ 3^(n + 3/2) * sqrt(n) / (4*sqrt(Pi)) * (1 - sqrt(3*Pi)/sqrt(n)). - _Vaclav Kotesovec_, Mar 08 2023

%e For n = 3, we have a(3) = 5 because there are 5 two-element anti-chains on "0,1,2" Motzkin trees on 3 edges.

%t M:= (1-z -Sqrt[1-2*z-3*z^2])/(2*z^2); T:= 1/Sqrt[1-2*z-3*z^2]; CoefficientList[Series[z^2*M^2*T^3, {z, 0, 30}], z] (* _G. C. Greubel_, Jan 21 2019 *)

%o (PARI) z='z+O('z^33); M=(1-z-sqrt(1-2*z-3*z^2))/(2*z^2); T=1/sqrt(1-2*z-3*z^2); v=Vec(z^2*M^2*T^3+'tmp); v[1]=0; v

%o (Magma) m:=30; R<x>:=PowerSeriesRing(Rationals(), m); [0,0] cat Coefficients(R!( (1-x-Sqrt(1-2*x-3*x^2))^2/(4*x^2*Sqrt(1-2*x-3*x^2)^3) )); // _G. C. Greubel_, Jan 21 2019

%o (SageMath) ((1-x-sqrt(1-2*x-3*x^2))^2/(4*x^2*sqrt(1-2*x-3*x^2)^3)).series(x, 30).coefficients(x, sparse=False) # _G. C. Greubel_, Jan 21 2019

%Y Cf. A000108, A002426, A001006, A005717.

%K nonn

%O 0,4

%A _Lifoma Salaam_, Dec 27 2010