Jon’s PhD Journal

January 29, 2007

Monday — complexity ….

Filed under: Notes — JDE @ 10:02 am

And yet more from Roger Lewin’s Complexity book:

  • p115 Lovelock: “… homeostasis [of Earth] emerged as a property of the system”
  • p115 Lovelock: “… the biota opeates in such a way to ensure optimum physical conditions for itself” — i.e. the actions of life adapts Earth to benefit life (From Ford Doolittle’s review of Lovelock’s 1979 book “It is not novel to suggest that life has profoundly changed the Earth, but it is novel and daring to suggest that it has done so in a seemingly deliberately adaptive way, in order to ensure its own existence”.
  • p116 Lovelock: “… like all complex systems in the universe, it has a tendecny to produce stability and survive” & ” I need to show that the stability emerges from the properties of the system, not from some purposeful guiding hand” — see Daisyworld
  • a particular Daisyworld simulation had 20 species of Daisy, five of rabbits and three foxes: a decrease in the population led to dips in the fox and rabbit population also
  • p117 quote from Alfred Lotka The Elements of Physical Biology (1925): “But it will be apparent (snip) that the physciual aws governing evolution in all probabilty take on a simpler form when referred to the system as a whole thatn to any part thereof”
  • p117 “… the Gaia hypothesis did satisfy some of the criteria of complex adaptive systems that Stu Kauffman had outlined (specifically) the emergence of homeostatic mechanisms … possibly a consequence of a system adapting to the edge of chaos”.
  • p125 from work by Jim Drake (maybe at Purdue): “the global property of persistence, arising from interaction among species in the community, and not particularly special species at that”
  • Stu Kauffman: rugged landscape idea?? (I think this could be fitness landscapes)
  • what is a power law?
  • p127 “… whatver the communities do, they do as a result of internal dynamics, not in response to anything external … an emergent proprty of a dynamical system”
  • p132 Stu Kauffman: “… there’s a price to pay in becoming more complex; the system is more lifely to break, fo instance”
  • Book: William Poundstone, The Recursive Universe — about cosmology and the game of life
  • p133 Dan McShea (University of Michigan, maybe)
  • p135 John Tyler Ponner (Princeton) attempted to develop an objective measure of (biological?) complexity: count the number of different cells in an organism. This should give a sense of the number of specialised functions the organism can perform
  • Measuring complexity is difficult
  • p137 Norman Packard quote “I view organisms as complex dynamical systems, and what drives their evolution is increased computational ability” and “I’m not syaing that every organism need itself become more complex; second, the sytesm as a whole undoubtedly becomes more complex”
  • p137 1977 text on evolution by Theodosius Dobzhansky, Francisco Ayala, G. Ledyard Stebbins and James Valentine : “‘the ability to gather and process information’ is siad to have increased through evolutionary history, and, indeed, to be a mark of progress”
  • p138 Francisco Ayala: “the ability to obtain and process information about the environmant and to react accordinly is an important adaptation because it allows that organism to seek out suitable environments and resrouces and to acoid unsuitable ones.”
  • p138 Ed Wilson “… also considers information processing a a measure of complexity” and “There’s been a gneral increase in information processing over the last 550 million years, particularly the last 150 million years”
  • seems some organisms increase in complexity to adapt to the environment. I guess that some don’t, namely the ones who’s environmental niche is small i.e. lower competition, and they have an advanteage on those approaching their niche ?

January 25, 2007

Thursday: more complexity ….

Filed under: Notes — JDE @ 7:22 pm

Notes from Roger Lewin’s Complexity :

  • p84: Tom Ray — came up with Tierra, which is a simulated world. Basically the virtual organisms in the simulation have a few simple rules, and they can reproduce, spawn, etc., as a result of these rules.
  • probably need to look at power-law distributions again … may be important
  • p109: Kauffman talking about Gaia”… (James) Lovelock is arguing that Earth’s biologicl and physcial systems are tightly coupled in a giant homeostatic system. My coevolutionary model is a clue that coevolving entities such as he talks about can control the structure of their (fitness) landscapes and how richly couped they are”
    • e.g. mammals breathe in O2 and CO2 out as a waste product, whereas plants breathe in CO2 and O2 out as a waste product — on a gloabl scale it balances, so where did the functional integration come from
  • p109 Kauffman continues: “it’s not unreasonable to think there might be an attractor to the meta-dynamics of the system. The adaptive agents collectively make the worklds they live in congenial to themselves, and are drawn to that characteristic structure, the edge of chaors, where their interests are mutually balanced. Now that’s homeostasis”
    • NB the edge of chaos appears to be the bit before truly chaotic behaviour, but moving away from stable behaviour
    • intermittency in flow of fluids could be an analogy to this (p101)
  • p109 Kauffman to see if Gaia exists “…need to know whether systems are coupled together … You need to know how extensive the links are, because if they are short, you won’t get a global system. You need to get a sense of a dunamical system that has emergent properties, properties that might lead to global homeostatic mechanisms”

January 24, 2007

Wednesday: continuing today’s complexity …

Filed under: Notes — JDE @ 7:11 pm

And some more, continuing from today’s earlier post:

  • p57 fitness landscapes:
    • fitness of an individual related to the height of a peak it would create if a landscape view was created
    • highest peaks = greater fitness
    • “if an indivudal happens to be in a fitness vallyey, then mutation and selection might push it up a local peak, representing a rise in fitness. Once on the local peak it may, metaphorically, gaze enviously at a nearby peak, but be unable to reach it because that would require crossing a valley of lower fitness”
  • p57 Stuart Kauffman working with John Maynard Smith:
    • used fitness landscapes to show that natural selection unable to move a species towards fitness peaks, and that the dynamics of the genetic system itself may exert a strong influence in this respect
  • p58-59 Red Queen effect (names by Leigh Van Valen, University of Chicago):
    • predator and prey are coupled together: one makes an evolutionary advancement, increasing it’s fitness and lowering the fitness of the other, the other must react/respond to keep the fitness status quo
    • if you have hundreds of these coupled landscapes, do not get chaos: instead system moves through activity states (frozem, chaotic, etc), then comes to rest, with fitness optimised “posised at the edge of chaos”
    • the individuals are acting selflishly: “Collective adaptiation to selfish ends produces the maximum average fitness, each species in the contenxt of other. As if by an invisible hand (…) collective good is ensured”
      • Phil Anderson called this “Mini-Gaia”
    • p60: … an ecosystem brought itself to a collectively beneficial state, controlled through vast networks of interactions
  • P60 – 61: Per Bak: developed hypothesis that large interactive systems — dynamical systems — naturally evolve toward a critical state.
  • p72: Brian Goodwin: mechanics of embryological development are tightly constrained, and this greatly limits the kinds of structures and the kins of species that can arise.
    • in the language of complex dynamical systems, the space of morphological possibilities is thinly populared by attractors, those states to which dynamcial systems eventually settles

Robocode …

Filed under: Coding — JDE @ 6:12 pm

A quick post whilst I think about it: came across RoboCode on the internet today, which is a tool used to teach Java to people. You program a tank and leave it running on it’s own to fight other tanks.

Whilst maybe an interesting learning opportuniyt, may also be used and adapted as a protoyping arena for boids?

Wednesday: more complexity …

Filed under: Notes — JDE @ 10:22 am

Further notes from Richard Dawkins’ Complexity book:

  • p47 “Notice that global structure emerges from flocal activity rules, a charcteristic of complex systems”
  • p48: “mathematicians already knew that many dynamical systems exhibit 3 classes of bejhavior: fixed point, periodic and chaotic”
    • Wolfram found a 4th class studying cellular automata that was somewhere between chaotic and fixed or periodic behaviour
  • p50 lambda parameter: lambda is a mathematical device that sets the rules of the cellular automaton and allows the consequences to be monitored across a continuum
  • Names:
    • Stuart Kauffman
    • Christopher Langton
    • Steve Wolfram
    • Norman Packard
    • Doyne Farmer

January 23, 2007

Tuesday: Complexity …

Filed under: Notes — JDE @ 7:02 pm

NB Yesterday tidies up the emergence paper and sent off to Will.

Today, currently looking Complexity by Roger Lewin. Notes:

  • complexity is somewhere between order and chaos (p10)
  • highly nonlinear systems
    • small inputs can => lrg consequences
    • very slight diferences in inital conditions provde very different outcomes
    • p13: does an emergent macro behaviour feed back into the micro level interactions?
    • p17 similar to Strogatz’s comments about phase transitions being rapid: water -> ice, etc
    • p20: “msot complex systems exhbit what mathematicians call attractors, states to which the system evenutally settles, depending onf the properties of the system”
    • p24: self-organisation is a natural property of complex genetic systems
    • p27: Boolean networks – some sort of attractor?

… from reading the first 2 chapters, it seems that the argument is that there are natural states, inherent in systems, which systems tend to gravitate towards. (Energy states and protein folding is ringing in my ears). This is used to explain how the eye formed as a biological construct: not merely chance through evolution, but because if you put the building blocks for an eye together, they will “tend” towards forming an eye-like structure.

January 20, 2007

Saturday: Java-ing the tardis* …

Filed under: Coding — JDE @ 11:59 am

*or something…

Chapter 2 = done!

NB notes being taken in hard-copy book, instead of online

Started on chapter 3: did primitive variables, and started to look at object references variables.

Plus worked out how to use classes in Eclipse, which works pretty well actually!!

January 19, 2007

Coding Friday: new weapon of choice, plus path forwards …

Filed under: Coding — JDE @ 7:11 pm

Following discussions with Will on Thursday, still uhmming and ahhing about choosing a development language. I can see the arguments for both sides, unfortunately. But I think I need to basically pick one, and make it work. Java has got more of a community around it (I’d say), so it makes a lot of sense to write a simulaton in Java. (Python, to be fair, is supposed to be reasonably easy to pick up, and it has been in my experience to date. As such, I could always pick that up in the future as required).

So the next question is to work out what to do next in the coding stream. The final goal for this is to create a simulation. (Typing as thinking here, ergo a potential ramble) This simulation needs to be able for me to get location data out of it, and for me to change the logic the agents use.

I don’t know that much Java however … but I do have some books to learn with. I wonder if I spent 2 hours each Friday, and 15min every other weekday (=> 3 hours per week), how long would it take me to learn enough to go forwards …? I would have thought at that rate not more than 2 months, tops (~24 hrs worth).

Today is the 19th January 2007. Java learnt by 16-March then. Should hopefully know enough to start hacking the simulation I’ve found to bits.

(later) Chapter 1 = done !! Only 17 to go !!

January 18, 2007

Thursday blogging …

Filed under: Notes — JDE @ 3:04 pm

Still on the complexity trail today. No post yesterday as a particular browser (with an ironically apt name) should be holding it’s head in shame, seeing as it *kept* crashing. I was going through the PNAS references found earlier in the week, but can’t remember where I got up to — ergo new ground to be covered today!!

 (… 20 min later) realised I’m sitting in a library for a change. With books. Probably on complexity. Found a couple of titles from the library system, and gone for a mooch on what they have in them.

Notes from meeting with Will, 18-Jan-07 (F2F) …

Filed under: Meetings — JDE @ 12:52 pm

Some notes:

  • !! get emergence report out of the drawer and send to Will !!
  • Attractors:
    • point: leads to convergence to single point
    • limit cycle: more interesting (would this produce geese skeins?)
    • chaos: flocking behaviour
  • books mentioned:
    • Deep Simplicity
    • James Lovelock books about Gaia
  • look at Richard Mitchell (@ Reading) website: explains the Gaia work (plus information about Gaia)
  • Coding
    • consider learning Java => less fiddling on the edge of Python development ;-) 
    • you can use someone else’s model SO LONG AS YOU CITE IT!!
    • highlight all assumptions
      • e.g. the flocking model and the fact that although it appears to be visual comparable to real bird flocks, is it really flocking?
  • NB mature students get less slack when talking at conferences
    • people assume you are a young academic
    • highlight your status when you talk
  • design of experimentation
    • look at particle swarm optimisation PhDs:
      • look at methods used
      • look at experimental content
      • i.e. how did other people experiment using simulations
  • IMPORTANT to compare work against other work in field
    • the plan about comparing your efforts to Reynolds’ model seems a good approach
  • Transfer report
    • Timelines
      • Sep-07: presentation and viva
      • Jun-07: hand in report
      • (start thinking about getting there after PSO report, expected Mar-07 ?)
    • Will has examples of the transfer report
Older Posts »

Blog at WordPress.com.