[FRIAM] Automata with FFT

[glen]

There's an observability assumption underlying any sort of assessment of frequencies. I've whipped the horse before about pre-emptive registration; the idea being that once you *plop* into an observing frame, you may not be able to get out of it. Even the most agnostic, objective way of targeting some "system" raises indexicality and universality issues.

So, sure, once one registers some thing as "that's an agent", puts a boundary around it, formalizes flux across that boundary, etc., then you can identify agent behaviors and (nearly) distinguish/register those behaviors, and quantify their frequencies (within some ε similarity between any 2 behaviors).

Of course, where one defines agents explicitly, up-front, frequency analysis is standard practice, even in monitoring operating systems and the execution of arbitrary software. But with underlying (global) logics like Lenia's updater, all you're doing is shunting off the physics to an approximation. I suppose you might find a *compression* of Lenia's kernel as applied over space and time. That seems to be what Chan is after, similar to "theoretical biology".

On 9/27/22 18:50, Steve Smith wrote:
> 
> On 9/26/22 9:30 AM, Jon Zingale wrote:
>> One thing that got me interested in this work is the possibility of organizations of agents in different frequency spaces coming together to make agents over the total frequency domain. Also, I would be curious about the robustness of agents under frequency filtering. There appear to be possibilities here that weren't necessarily available in other "smooth" approaches.
> 
> I'm really glad you tossed this one into the ring, BTW.
> 
> I don't know how this maps onto your own thoughts but I have been cogitating on systems of systems (e.g. ecosystems, sociopoliticaleconomic, etc) for some time and *they* definitely represent huge scale ranges in any dimension you can identify/measure.  Multi-scale features abound in the these domains with all sorts of frequency-dependent coupling/mixing models implied.  " If a tick carrying lyme disease falls off a grass-blade in a field and lands on a pet dog with a ... human master who ... etc., does a butterfly flap it's wings in a typhoon?"  and "for want of a nail..." both come to mind.
> 
> I regularly have Herb Simon's prophetic statement <https://www2.econ.iastate.edu/tesfatsi/ArchitectureOfComplexity.HSimon1962.pdf> ringing in my ears when I see/hear/think about the modeling of systems of systems:
> 
>     /NEARLY DECOMPOSABLE SYSTEMS In hierarchic systems, we can distinguish between the interactions among subsystems, on the one hand, and the interactions within subsystems -i.e., among the parts of those subsystems~n the other. The interactions at the different levels may be, and often will be, of different orders of magnitude./
> 
> Agent modeling carries a strong bias toward quantized entities but makes spatial and time frequency coupling more obvious/easy/explicit than some other forms...  I think of CA (including the myriad sub-species that Chan offers in Table 1) as specialized/constrained cases of Agent-models, just with specific spatio-temporal regularity imposed?

-- 
ꙮ Mɥǝu ǝlǝdɥɐuʇs ɟᴉƃɥʇ' ʇɥǝ ƃɹɐss snɟɟǝɹs˙ ꙮ