Re: What is emergence?

[Russell Standish]

This subject suffers from a great deal of confusion. I am
contemplating writing a paper myself as an effort to clear this up. My
own view of what emergence is as follows:

Consider a system described by 2 languages which we call "microscopic"
and "macroscopic". Furthermore, we suppose that the system is
completely described by the microscopic language - true in the case of
an agent based simulation for example, arguably less true of a real
world system, where the microscopic description is probably expressed
in terms of an approximation like classical mechanics.

So far, this is in agreement with Edmund and Sipper, although I would
attribute these ideas back to Denbigh and Denbigh in the early '80s

If some concept or property in the macroscopic language is not
contained within the microscopic description, then it is emergent. I
would disagree with Edmund and Sipper regarding the element of
surprise. An example of this is entropy. According to Boltzman's
formula for entropy, S=-k log W, where W is the number of microstates
within \epsilon of a constant energy surface, i.e. someone has
identified all microstates lying within a small distance of some
energy surface as a particular macrostate. This choice is not given by
the microdescription, and is in fact "arbitrary". epsilon is not given
(it cannot be allowed to approach zero, as entropy tends to infinity),
and equally a different equivalence class could be employed (eg
macrostate i corresponds to i particles being in the left side of the
container, and N-i particles in the right - the resulting entropy also
obeys the H-theorem and various other nice properties, but is
numerically different from the conventional definition). So the
concept of entropy cannot be contained in the microscopic language -
it is an emergent property of the system.

Of course, the arbitrariness of emergent properties will undoubtedly
worry some people. The thing to realise, of course, is that not all
macroscopic descriptions are created equal. Some are simpler than
others (hence preferred by Occam's razor), and some are more accurate
(or predictive) than others. In general, there is a trade-off between
accuracy and simplicity, although it is entirely possible to have a
macroscopic description that is niether simple nor accurate (which
would make it worthless).

So one could say that a property is emergent if it is contained in a
macroscopic language that is effective (has a reasonable balance of
simplicity and accuracy), but is not contained in the underlying
microscopic language.

                                                Cheers

hanjing wrote:
> 
> Hello,
> 
> Althought I have been in the area of Alife for 3 years, I still have
> a question: 
> 
>     What is emergence? It seems not so clear to me. 
> 
>     Is there any definition? 
> 
>     Are there any criteria to judge whether a phenomenon is "emergence"? 
> 
> I have heard some oponions such as "unpredictable","unexpected",
> but it may be caused by lack of knowledge or data... Because after we get 
> this "unpredictable emergence", we can find a way to explain why it 
> occurs -- so it is predictabl. 
> 
> This question has been in my mind for a long time, and I found it very 
> important. 
> 
> 
> I will be very happy to get any answers.
> 
> Thanks a lot and sorry for my question!
> 
> Han Jing
> 
> 
> 
> 
> 
> 
> 
> 
> 
>                   ==================================
>    Swarm-Modelling is for discussion of Simulation and Modelling techniques
>    esp. using Swarm.  For list administration needs (esp. [un]subscribing),
>    please send a message to <address@hidden> with "help" in the
>    body of the message.
>                   ==================================
> 



----------------------------------------------------------------------------
Dr. Russell Standish                     Director
High Performance Computing Support Unit, Phone 9385 6967                    
UNSW SYDNEY 2052                         Fax   9385 6965                    
Australia                                address@hidden             
Room 2075, Red Centre                    http://parallel.hpc.unsw.edu.au/rks
----------------------------------------------------------------------------


                  ==================================
   Swarm-Modelling is for discussion of Simulation and Modelling techniques
   esp. using Swarm.  For list administration needs (esp. [un]subscribing),
   please send a message to <address@hidden> with "help" in the
   body of the message.
                  ==================================