The network repeats?

[Jay Hannah]

I've updated perl/ (Committed revision 161) and noticed something I
had not seen before.

http://jays.net/wiki/ChemChains_sandbox
The network repeats?

Have you seen this sort of network repetition in the C++ version?

Thanks,

j

[Tomas Helikar]

Jay, you're most likely getting repeats (or cycles) due to a couple of things. 1) you're not iterating the inputs after T+0 and 2) the lack of chaos would cause the inputs to be thrown on a cycle, which MIGHT cause the entire network to settle on one.

So are your inputs set to a fixed value then?

Tom

[Jay Hannah]

On Aug 19, 2008, at 4:52 PM, Tomas Helikar wrote:
> Jay, you're most likely getting repeats (or cycles) due to a couple
> of things. 1) you're not iterating the inputs after T+0 and 2) the
> lack of chaos would cause the inputs to be thrown on a cycle, which
> MIGHT cause the entire network to settle on one.
>
> So are your inputs set to a fixed value then?

Yes. perl/ currently interprets "Input:ECM:False:R:C:1:0-100:1999" to
mean it should start as 0 (False) and then the node never varies
because Inputs have no Inputs to change their value, so it remains in
its initial state forever.

Like most of my code, I'm sure this is a misunderstanding and/or
laziness on my part. :)

I'm hoping to get a deterministic set of network states (T 0..100)
out of cpp/ so other implementations (and future cpp/
implementations) can be tested against the gold standard and
corrected. I'll attempt more reading of cpp/ to see if/how this may
be possible.

Thanks,

j