a(0)=a(1)=a(2)=a(3)=a(4)=a(5)=1 , and a(n)=(a(n-1)*a(n-5)+a(n-2)*a(n-4)+a(n-3)*a(n-3))/a(n-6) if n > 5 .
The first few terms are: 1,1,1,1,1,1,3,5,9,23,75,421,1103,5047,41783,281527 and is listed as A006722 in Sloane's Online Encyclopedia of Integer Sequences (OEIS). For more information about this and related sequences, see article "SomosSequence" in Eric Weisstein's MathWorld.
The sequence is one of a large class of sequences of numbers that satisfy a non-linear recurrence relation depending on previous terms. It is also one of the class of sequences which can be computed from a theta series, hence I call them theta sequences. Here is the details:
Fix the following seven constants: c1 = 0.875782749065950194217251..., c2 = 1.084125925473763343779968..., c3 = 0.114986002186402203509006..., c4 = 0.077115634258697284328024..., c5 = 1.180397390176742642553759..., c6 = 1.508030831265086447098989..., and c7 = 2.551548771413081602906643... . Consider the doubly indexed series: f(x,y) = c1*c2^(x*y)*sum(k2, (-1)^k2*sum(k1, g(k1,k2,x,y))) , where g(k1,k2,x,y) = c3^(k1*k1)*c4^(k2*k2)*c5^(k1*k2)*cos(c6*k1*x+c7*k2*y) . Here both sums range over all integers. Then the sequence defined by a(n) = f(n-2.5,n-2.5) is the Somos 6 sequence. I announced this in 1993.Based on my earlier work, in 1989 I started to publicize this sequence at the Computers and Mathematics conference held at MIT.