*n*-dimensional vector. Each element of the vector can contain any one of [tex]\lambda=3[/tex] values (-1, 0 or +1). Then the number of possible vectors is simply:

[tex]\lambda^n[/tex]

If we place the additional restriction that the vector must contain exactly [tex]k[/tex] non-zeros, then it becomes:

[tex]p=(\lambda-1)^{k}\times\binom{n}{k}=\frac{n!(\lambda-1)^{k}}{k!(n-k)!}[/tex]

If we change the restriction so that it must contain

*at most*[tex]k[/tex] non-zeros and

*at least*1 non-zero, then it becomes:

[tex]p=\sum_{k'=1}^{k}\left[(\lambda-1)^{k'}\times\binom{n}{k'}\right]=\sum_{k'=1}^{k}\frac{n!(\lambda-1)^{k'}}{k'!(n-k')!}[/tex]

Are my equations correct? Is there a more compact way of expressing this last equation, to get rid of the summation?