OFFSET
1,2
COMMENTS
As an interspersion (and dispersion), the array is, as a sequence, a permutation of the positive integers. Column k consists of the numbers m such that the least summand in the tribonacci representation of m is T(1,k). For example, column 1 consists of numbers with least summand 1. This array arises from tribonacci representations in much the same way that the Wythoff array, A035513, arises from Fibonacci (or Zeckendorf) representations.
From Abel Amene, Jul 29 2012: (Start)
(Row 1) = A000073 (offset=4) a(0)=0, a(1)=0, a(2)=1
(Row 2) = A001590 (offset=5) a(0)=0, a(1)=1, a(2)=0
(Row 3) = A000213 (offset=4) a(0)=1, a(1)=1, a(2)=1
(Row 4) = A214899 (offset=5) a(0)=2, a(1)=1, a(2)=2
(Row 5) = A020992 (offset=6) a(0)=0, a(1)=2, a(2)=1
(Row 6) = A100683 (offset=6) a(0)=-1,a(1)=2, a(2)=2
(Row 7) = A135491 (offset=4) a(0)=2, a(1)=4, a(2)=8
(Row 8) = A214727 (offset=6) a(0)=1, a(1)=1, a(2)=2
(Row 9) = A081172 (offset=8) a(0)=1, a(1)=1, a(2)=0
(column 1) = A003265
(column 2) = A353083
(End) [Corrected and extended by John Keith, May 09 2022]
FORMULA
T(1,1)=1, T(1,2)=2, T(1,3)=4, T(1,k)=T(1,k-1)+T(1,k-2)+T(1,k-3) for k>3. Row 1 is the tribonacci basis; write B(k)=T(1,k). Each row satisfies the recurrence T(n,k)=T(n,k-1)+T(n,k-2)+T(n,k-3). T(n,1) is least number not in an earlier row. If T(n,1) has tribonacci representation B(k(1))+B(k(2))+...+B(k(m)), then T(n,2) = B(k(2))+B(k(3))+...+B(k(m+1)) and T(n,3) = B(k(3))+B(k(4))+...+B(k(m+2)). (Continued shifting of indices gives the other terms in row n, also.)
EXAMPLE
Northwest corner:
1 2 4 7 13 24 44 81 149 274 504
3 6 11 20 37 68 125 230 423 778
5 9 17 31 57 105 193 355 653
8 15 28 51 94 173 318 585
10 19 35 64 118 217 399
12 22 41 75 138 254
14 26 48 88 162
16 30 55 101
18 33 61
21 39
23
MAPLE
# maximum index in A73 such that A73 <= n.
A73floorIdx := proc(n)
local k ;
for k from 3 do
if A000073(k) = n then
return k ;
elif A000073(k) > n then
return k -1 ;
end if ;
end do:
end proc:
# tribonacci expansion coeffs of n
A278038 := proc(n)
local k, L, nres ;
k := A73floorIdx(n) ;
L := [1] ;
nres := n-A000073(k) ;
while k >= 4 do
k := k-1 ;
if nres >= A000073(k) then
L := [1, op(L)] ;
nres := nres-A000073(k) ;
else
L := [0, op(L)] ;
end if ;
end do:
return L ;
end proc:
A278038inv := proc(L)
add( A000073(i+2)*op(i, L), i=1..nops(L)) ;
end proc:
A135175 := proc(n, k)
option remember ;
local a, known, prev, nprev, kprev, freb ;
if n =1 then
A000073(k+2) ;
elif k>3 then
procname(n, k-1)+procname(n, k-2)+procname(n, k-3) ;
else
if k = 1 then
for a from 1 do
known := false ;
for nprev from 1 to n-1 do
for kprev from 1 do
if procname(nprev, kprev) > a then
break ;
elif procname(nprev, kprev) = a then
known := true ;
end if;
end do:
end do:
if not known then
return a ;
end if;
end do:
else
prev := procname(n, k-1) ;
freb := A278038(prev) ;
return A278038inv([0, op(freb)]) ;
end if;
end if;
end proc:
seq(seq(A135175(n, d-n), n=1..d-1), d=2..12) ; # R. J. Mathar, Jun 07 2022
CROSSREFS
KEYWORD
nonn,tabl
AUTHOR
Clark Kimberling, Dec 18 2007
EXTENSIONS
T(3, 4) corrected and more terms by John Keith, May 09 2022
STATUS
approved