|
|
A100967
|
|
Least k such that binomial(2k + 1, k - n) >= binomial(2k, k).
|
|
3
|
|
|
3, 9, 18, 29, 44, 61, 81, 104, 130, 159, 191, 225, 263, 303, 347, 393, 442, 494, 549, 606, 667, 730, 797, 866, 938, 1013, 1091, 1172, 1255, 1342, 1431, 1524, 1619, 1717, 1818, 1922, 2029, 2138, 2251, 2366, 2485, 2606, 2730, 2857, 2987, 3119, 3255, 3394, 3535
(list;
graph;
refs;
listen;
history;
text;
internal format)
|
|
|
OFFSET
|
1,1
|
|
COMMENTS
|
From the formula, if we know k, we can estimate n as approximately 0.83 sqrt(k).
Open question: Does binomial(2*a(n) + 1, a(n) - n) = binomial(2*a(n), a(n)) for any n? An affirmative answer would settle whether there exists an odd term greater than 3 in A003016. - Danny Rorabaugh, Mar 16 2016
|
|
LINKS
|
|
|
FORMULA
|
Round(0.3807 + 1.43869 n + 1.44276 n^2) is an exact fit for the first 50 terms.
As n -> infinity, we have a(n) = (n^2+n)/log(2) + o(n). - Robert Israel, Mar 16 2016
|
|
MAPLE
|
F:= proc(n) local Q, LQ, k, k0;
LQ:= -ln(GAMMA(k-n+1))-ln(GAMMA(k+1+n))-ln(k+1+n)+ln(2*k+1)+2*ln(GAMMA(k+1));
k0:= floor(fsolve(LQ, k=n..max(2*n^2, 9)));
if (2*k0+1)*binomial(k0, n) >= (n+1)*binomial(k0+1+n, n+1) then
while (2*k0-1)*binomial(k0-1, n) >= (n+1)*binomial(k0+n, n+1) do k0:= k0-1 od
else
while (2*k0+1)*binomial(k0, n) < (n+1)*binomial(k0+1+n, n+1) do k0:= k0+1 od;
fi;
k0;
end proc:
|
|
MATHEMATICA
|
k=1; Table[While[Binomial[2k+1, k-n] < Binomial[2k, k], k++ ]; k, {n, 50}]
|
|
PROG
|
|
|
CROSSREFS
|
Cf. A000984, A003015 (numbers that occur 5 or more times in Pascal's triangle).
|
|
KEYWORD
|
nonn
|
|
AUTHOR
|
|
|
STATUS
|
approved
|
|
|
|