Re: [Swarm-Modelling] lifecycle requirements

[Marcus G. Daniels]

Scott Christley wrote:
> This is often call the state explosion problem, and it becomes a 
> problem when state change is more than changing the value of a 
> variable but implies functional change as well.  If you design a model 
> where you explicitly handle all of those states, that means you need 
> to explicitly write code for the function of the agent for each 
> states.  What if the number of states is exponential?
I think that's overstating it a bit.  A larger space means more things 
are possible, but the interpreter of that space can still be simple.   A 
codon is simple and stable in spite of the fact a larger genomes can 
code for more complex things.   Similarly, the instruction set of a CPU 
is small and finite and relatively straightforward to get working 
compared to the virtually infinite number of programs than can be 
written with that set of opcodes.  

> This might be no big deal, but another complexity is that under 
> different circumstances, the splicing process produces different RNA.  
> So the basic idea that a single gene (DNA) produces a single Protein 
> is not true, there are multiple proteins produced through alternative 
> splices.
Ok,1) the notion of alternatives and 2) information about how to 
regulate the alternatives
> In biology, a protein's function is defined by its structure but that 
> structure is not fixed, enzymes and other molecules often change that 
> structure, so a protein is considered to have different conformations.  
and 3) context dependency
> Now you are faced with a scenario that the hyperspace is so large that 
> you cannot code in all of the possible interactions ahead of time.  
But it's mainly about computer runtime not human time (e.g. too many 
coding details).  
> Well (one of) the key problem is going to be exactly how do two 
> unknown protein 3 dimensional structures interact.  If you take the 
> naive approach you say, okay we will model the atoms of the two 
> structures, calculate all the forces, and make some determination on 
> whether they bind, etc etc.  But you cannot use that approach when 
> attempting to get your higher level patterns because it is too 
> computationally complex for a system with lots of interacting proteins.
Some have been known to take a swing at this:

 http://www.t10.lanl.gov/kys
 http://mdgrape.gsc.riken.jp
> What you need then is to do that naive approach once, then translate 
> the result into a rule.
Probably more like a few hundred times per different perspectives and 
assumptions to know when that rule is really there and not just 
statistical noise, or some event correlated to another that is the real 
cause.   :-)

> However that rule has two parts, the protein X and Y is okay but what 
> about the do something part.  What if you don't know all the possible 
> "do somethings" ahead of time?  In a really cool evolutionary model 
> you would not, this would be new functionality that was acquired.
A common objection to agent modeling is along the lines of "You know, I 
can think of N different ways you could get that result -- you found one 
using a liberal tolerance for error on your inputs."   An evolutionary 
model may help you find more, but what do they have to do with the real 
world?   E.g. how much error is involved in the conformational estimates 
of X and Y and how does that impact the extent to which that "something" 
means anything?