OFFSET
0,2
COMMENTS
Computation: Allan C. Wechsler for n <= 5 (manual), Fred Lunnon for n <= 8 (Maple), Michael Kleber for n <= 30 (Mathematica).
LINKS
N. Bedaride, Classification of rotations on the torus T^2, Theoretical Computer Science, 385 (2007), 214-225.
J.-P. Borel, A geometrical Characterization of factors of multidimensional Billiards words and some Applications, Theoretical Computer Science, 380 (2007) 286-303.
L. Vuillon, Balanced Words, Bull. Belg. Math. Soc., 10 (2003), 787-805.
FORMULA
Computation may be expedited by generating only words in which the symbols occur in increasing alphabetic order: this was done in the production version.
EXAMPLE
For n = 5 there are a(5) = 189 words, permutations on the alphabet {1,2,3} of the 32 words
11111, 11112, 11121, 11123, 11211, 11212, 11213, 11231, 12111, 12112, 12113, 12121, 12122, 12123, 12131, 12132, 12212, 12213, 12221, 12222, 12223, 12231, 12232, 12311, 12312, 12313, 12321, 12322, 12323, 12331, 12332, 12333.
MATHEMATICA
(* Number of ways to interleave N elements from 3 arithmetic seqs *)
(* Program due to Michael Kleber, Aug 2010 *)
(* Given a string like "ABCABA", produce a set of inequalities *)
(* about the three arithmetic progressions giving successive A/B/Cs *)
(* The N-th occurrence (1-indexed) of character X corresponds to the value *)
(* BASE[X] + N * DELTA[X] *)
(* In all functions, seq is eg {"A", "B", "C", "A", "B", "A"} *)
(* The arithmetic-progression value of the i-th element of seq *)
value[seq_, i_] := BASE[seq[[i]]] + DELTA[seq[[i]]] * numoccur[seq, i]
numoccur[seq_, i_] := Count[Take[seq, If[i>0, i, Length[seq]+i+1]], seq[[i]]]
(* First element of the seq is greater than anything that would precede it*)
lowerbound[seq_] := (BASE[ # ] < value[seq, 1])& /@ Union[seq]
(* Each element of the seq is greater than the previous one *)
upperbound[seq_] := (value[seq, -1] < value[Append[seq, # ], -1])& /@ Union[seq]
(* Last element of the seq is less than anything that would follow it *)
ordering[seq_] := Table[value[seq, i] < value[seq, i+1], {i, Length[seq]-1}]
ineqs[seq_] := Join[ lowerbound[seq], ordering[seq], upperbound[seq] ]
vars[seq_] := Join @@ ({BASE[ # ], DELTA[ # ]}& /@ Union[seq])
witness[seq_] := FindInstance[ ineqs[seq], vars[seq] ]
witness[s_String] := witness[Characters[s]]
(* All obtainable length-n shuffles of three arithmetic seqs: *)
names = {"A", "B", "C"}
shuf[0] := {""}
candidates[n_] := Flatten[Table[ob<>ch, {ob, shuf[n-1]}, {ch, names}]]
shuf[n_] := shuf[n] = Select[ candidates[n], witness[ # ] != {}& ]
(* Typical session *)
In[18]:= Table[Length[shuf[i]], {i, 0, 12}]
Out[18]= {1, 3, 9, 27, 75, 189, 447, 951, 1911, 3621, 6513, 11103, 18267}
In[19]:= TimeUsed[]/60 Out[19]= 6.73642
CROSSREFS
KEYWORD
nonn,more
AUTHOR
Fred Lunnon, Aug 18 2010
STATUS
approved