[seqfan] Re: 3-dimensional arrays?

[Chris Starling]

Hi Felix,

Thank you for your reply.

Yes I quite agree about the ease of translation; the line by line scans
seem more intuitive and more straightforward to program.  There might be
some argument for a zig-zagging scan of lines in a triangle, since each
line end would stay contiguous when it was morphed into a sequence, sort of
like an unfolding measuring stick.

But something tells me that "carriage return" scanning, as I call it, is
even more intuitive and appropriate.  The advantage there is that all
segments of contiguous entries can be read in the same direction, except
trivially, at the very beginning of each triangular layer.  And indeed, I
was able to get a simple spreadsheet to produce a sequence from a
trivariate multinomial.  Meanwhile the spirals don't even have the
advantage of longer contiguous segments of entries.

But I grin as I note that the questions still remain of what corner to
start from, and what corner to scan toward, in each layer.

Carry on,
-Chris.