Forum for the discussion of experiences in the representation and modeling of complex systems with highly non-linear behaviors / Foro para la discusión de temas relacionados con aspectos de la representacion de sistemas complejos y con comportamientos altamente no-lineales (Complex Adaptive Systems -CAS, System Dynamics, Agent Based Modeling) "..donde una inteligencia global emerge de interacciones locales de seres individualmente no inteligentes." -Sociobiology, O.E.Wilson-
Saturday, March 21, 2009
Libro - The Biology of Business
Este es de hecho el primer libro que leí en relación a los sistemas adaptativos complejos y me dio una muy buena introducción a lo que estos sistemas son y los ámbitos en los cuales este enfoque podría ser aplicado.
Este libro constituye una compilación de papers ordenados por tópico: Lo básico de los sistemas CAS, Aplicaciones en la Economía, Liderazgo con CAS (es esto posible en equipos distribuidos de trabajo?), Knowledge Management, Características emergentes y adaptación, por ejemplo.
Este libro sienta las bases para investigaciones posteriores en estas áreas, y lo considero mas un hito filosofico sobre este tema que una guía de aplicación practica de estas tecnicas en empresas existentes.
Altamente recomendable.
Detalles
Paperback: 287 pages
Publisher: Jossey-Bass; 1 edition (October 1, 1999)
Language: English
ISBN-10: 078794324X
ISBN-13: 978-0787943240
Product Dimensions: 9.3 x 6.3 x 1 inches
Link en Amazon
Labels:
Complejidad,
Economia,
General,
Libros
Wednesday, March 18, 2009
Behavioural Economics - Stealing
In this Video, DAn Ariely examines what are the probable bases for stealing. According to his experiments, there are several factors influencing the likelihood that someone will cheat or steal: "Distance" from money, Ingroup/Outgroup cheating incentive, anonimity, etc.
Apparently the capability to cheating is something endemic and widespread, albeit the cheating is normally small.
Labels:
Cooperacion,
Economia,
Evolucion Moral,
TED,
Video
Monday, March 16, 2009
CAS definition
The complexity paradigm uses systemic inquiry to build fuzzy, multivalent, multi-level and multi-disciplinary representations of reality. Systems can be understood by looking for patterns within their complexity, patterns that describe potential evolutions of the system. Descriptions are indeterminate and complimentary, and observer dependent. Systems transition naturally between equilibrium points through environmental adaptation and self-organization; control and order is emergent rather than predetermined.
A CAS behaves/evolves according to three key principles: a.- order is emergent as opposed to predetermined, b.-the system's history is irreversible, and c.-the system's future is often unpredictable.
The basic building blocks of the CAS are agents. Agents are semi-autonomous units that seek to maximize some measure of goodness, or fitness, by evolving over time. Agents scan their environment and develop schema representing interpretive and action rules. These schema are often evolved from smaller, more basic schema. These schema are rational bounded: they are potentially indeterminate because of incomplete and/or biased information; hey are observer dependent because it is often difficult to separate a phenomenon from its context, thereby identifying contingencies; and they can be contradictory. Schema exist in multitudes and compete for survival.
Existing schema can undergo three types of change: first order change, where action is taken in order to adapt the observation to the existing schema; second order change, where there is purposeful change in the schema in order to better fit observations; and third order change, where a schema survives or dies because of the Darwinian survival or death of its corresponding CAS. Schema can change through random or purposeful mutation, and/or combination with other schema. Schema change generally has the effect of making the agent more robust (it can perform in light of increasing variation or variety), more reliable (it can perform more predictably), or grow in requisite variety (in can adapt to a wider range of conditions).
The fitness of the agent is a complex aggregate of many factors, both local and global. The general health or fitness of the agent determines what the probability of change will be. Optimization of local fitness allows differentiation and novelty/diversity; global optimization enhances the CAS coherence as a system and induces long term memory. In general the probability of second order schema change is a nonlinear function of the fitness value.
Schema define how a given agent interacts with other agents surrounding it. Actions between agents involve the exchange of information and/or resources. These flows may be nonlinear. Information and resources can undergo multiplier effects based on the nature of interconnectedness in the system. Agent tags help identify what other agents are capable of transaction with a given agent; tags also facilitate the formation of aggregates, or meta-agents. Meta-agents help distribute and decentralize functionality, allowing diversity to thrive and specialization to occur. Agents or meta-agents also exist outside the boundaries of the CAS, and schema also determine the rules of interaction concerning how information and resources flow externally.
A CAS behaves/evolves according to three key principles: a.- order is emergent as opposed to predetermined, b.-the system's history is irreversible, and c.-the system's future is often unpredictable.
The basic building blocks of the CAS are agents. Agents are semi-autonomous units that seek to maximize some measure of goodness, or fitness, by evolving over time. Agents scan their environment and develop schema representing interpretive and action rules. These schema are often evolved from smaller, more basic schema. These schema are rational bounded: they are potentially indeterminate because of incomplete and/or biased information; hey are observer dependent because it is often difficult to separate a phenomenon from its context, thereby identifying contingencies; and they can be contradictory. Schema exist in multitudes and compete for survival.
Existing schema can undergo three types of change: first order change, where action is taken in order to adapt the observation to the existing schema; second order change, where there is purposeful change in the schema in order to better fit observations; and third order change, where a schema survives or dies because of the Darwinian survival or death of its corresponding CAS. Schema can change through random or purposeful mutation, and/or combination with other schema. Schema change generally has the effect of making the agent more robust (it can perform in light of increasing variation or variety), more reliable (it can perform more predictably), or grow in requisite variety (in can adapt to a wider range of conditions).
The fitness of the agent is a complex aggregate of many factors, both local and global. The general health or fitness of the agent determines what the probability of change will be. Optimization of local fitness allows differentiation and novelty/diversity; global optimization enhances the CAS coherence as a system and induces long term memory. In general the probability of second order schema change is a nonlinear function of the fitness value.
Schema define how a given agent interacts with other agents surrounding it. Actions between agents involve the exchange of information and/or resources. These flows may be nonlinear. Information and resources can undergo multiplier effects based on the nature of interconnectedness in the system. Agent tags help identify what other agents are capable of transaction with a given agent; tags also facilitate the formation of aggregates, or meta-agents. Meta-agents help distribute and decentralize functionality, allowing diversity to thrive and specialization to occur. Agents or meta-agents also exist outside the boundaries of the CAS, and schema also determine the rules of interaction concerning how information and resources flow externally.
Sunday, March 15, 2009
Conway's Game of Life
This is a simple programming game first proposed by British mathematician John Horton Conway in 1970. It is the best-known example of a cellular automaton.
The idea is that you have a 2-dimensional grid with cells. These cells can be alive or dead (on or off,true or false, colored or blank). When the simulation is running the state of a cell is determined by three simple rules:
1.- If the cell has no living neighbors or just one, it will be dead (caused by loneliness);
2.- If four or more of the cells next to it are alive, it will be dead (overcrowding);
3.- If a dead cell has exactly three living neighbors it will come alive (Bitstorm).
The effect of these simple rules can be quite surprising. The starting grid can be very chaotic, but after a while you can see patterns emerge, and even stable situations can be reached where nothing will change any more, or the same patterns are just repeated over and over.
From an Excel implementation of this algorithm, I can describe:
1.- "Main" Sheet
This is where the evolution happens, each evolving cell has the same value:
=IF(Reset,template!L3,IF(M5,CHOOSE(MIN(5,nbors),FALSE,FALSE,TRUE,TRUE,FALSE),nbors=3))
Notes
a.- Reset is the name given to a specific cell in the Main Worksheet, and which can be either False (iteration continues as usual), or True (Cell acquires Value of its equivalent cell in "Template" Sheet, in this particular example, Cell L3.
b.- Usual iteration will determine the number of neighbouring True cells for each cell, and apply the rules as explained above.
2.- "nbors" Sheet
This sheet holds the exact same matrix as in "Main". Each cell here has the value
={SUM(SIGN('1.run'!C3:E5))}
Please note this is a matricial function in Excel.
This is, each cell will sum what happens with each of its 9 surrounding cells. The result will thus be the number of True Cells around it.
3.- "Template" Sheet
Sheet equivalent in dimensions to the other two mentioned above, but which only holds cells with either True or False Values. This matrix will determine the values assumed when Reset is set to True in our "Main" sheet and an iteration is run.
Please feel free to download and play around with an example of such an Excel Database here.
Saturday, March 14, 2009
Stepping outside the Moral Matrix
In this Video Jonathan Heidt proposes al alternative for stepping "outside" the moral matrix (just as in the movie The Matrix) we are all constantly immersed. Several things are proposed here:
1.- Human beings are born with an innitial draft in regard with moral principles, which are modelled afterwards (amplified or diminshed) by the environment. This is reinfoced by several other researchers, whose books I will comment later.
2.- Cooperation is reinforced by punishing the freeriders, and what is more, people will spend resources in this punishment, aiming towards a state of "Fairness". Please consider Axelrod's book Evolution of Cooperation for other good example of this conclusion.
3.- Basic differences in regard with the innitial moral aspects are shown between Liberal, Libertarian and Conservative points of view,
"If you want the thrith to stand clear before you, never be for or against. The struggle between "for" and "against" is the mind's worst disease - Sent ts'an, 700CE
Labels:
Cooperacion,
Evolucion Moral,
TED,
Video
Complex and Adaptive Dynamical Systems: A Primer
An thorough introduction is given at an introductory level to the field of quantitative complex system science, with special emphasis on emergence in dynamical systems based on network topologies. Subjects treated include graph theory and small-world networks, a generic introduction to the concepts of dynamical system theory, random Boolean networks, cellular automata and self-organized criticality, the statistical modeling of Darwinian evolution, synchronization phenomena and an introduction to the theory of cognitive systems.
arXiv document can be seen here
arXiv document can be seen here
Book - The Origin of Wealth
The Origin of Wealth is a bestseller written around the proposal of considering the economic activity as a Complex Adaptive System. To do this Beinkocker first describes the current economic paradigm, showing its limitations, their origin and justification, as well as giving very persuasive arguments on why they are no longer useful in explaining the current situation and its evoultion.
He then jumps into the full description of complex systems with some basic distinguishing ideas of this "complexity economics"
1.-Dynamics: Model describes the economy as a dynamic non-linear system, far from equilibrium in contrast with the traditional model
2.- Agents: Model describes them as individuals making decisions with partial information, subject to errors and biases, and able to learn over time, all this in contrast with the traditional theory.
3.- Networks: Model considers explicit interactions between these individual agents. These Networks also CHANGE over time. The traditional model considers interactions between agents only through established market mechanisms (e.g. auctions)
4.- Emergence: Model makes no distintion between micro and macroeconomics. It searches for behaviours that appear "naturally" in these complex systems. No such alternative considered in the traditional theory.
5.- Evolution: Model considers an evolutionary process of differentiation, selection and amplification, providing the system novelty, and it is a mechanism for attaining order and complexity (flexibility). Traditional theory also considers no such mechanism.
Highly Recommendable.
Product Details
Paperback: 544 pages
Publisher: Harvard Business School Press; 1 edition (September 14, 2007)
Language: English
ISBN-10: 1422121038
ISBN-13: 978-1422121030
Product Dimensions: 9.2 x 6.1 x 1.5 inches
He then jumps into the full description of complex systems with some basic distinguishing ideas of this "complexity economics"
1.-Dynamics: Model describes the economy as a dynamic non-linear system, far from equilibrium in contrast with the traditional model
2.- Agents: Model describes them as individuals making decisions with partial information, subject to errors and biases, and able to learn over time, all this in contrast with the traditional theory.
3.- Networks: Model considers explicit interactions between these individual agents. These Networks also CHANGE over time. The traditional model considers interactions between agents only through established market mechanisms (e.g. auctions)
4.- Emergence: Model makes no distintion between micro and macroeconomics. It searches for behaviours that appear "naturally" in these complex systems. No such alternative considered in the traditional theory.
5.- Evolution: Model considers an evolutionary process of differentiation, selection and amplification, providing the system novelty, and it is a mechanism for attaining order and complexity (flexibility). Traditional theory also considers no such mechanism.
Highly Recommendable.
Product Details
Paperback: 544 pages
Publisher: Harvard Business School Press; 1 edition (September 14, 2007)
Language: English
ISBN-10: 1422121038
ISBN-13: 978-1422121030
Product Dimensions: 9.2 x 6.1 x 1.5 inches
Labels:
Dilema del Prisionero,
Economia,
Libros
Book - The evolution of Cooperation
The Evolution of Cooperation is a book I heard of back in 1991 when I first read "The selfish Gene" by Richard Dawikins. Dawkins devoted a complete chapter to the description of the particular problem for which this book poses an alternative, and his description was so compelling that it eventually led me to buy it.
The experiment is quite simple and at the same time ingenious. It handles around the problem of cooperation. This is, how is cooperation explained in a darwinian context, where all entities are supposed to be modelled as selfish, without central authority, and looking for their own best outcome? It is evident that cooperation can lead to this, that is all cooperators being best off, but, then how is cooperation evolved and configured?
To model this, Axelrod proposed a "Game" of iterated Prisioner's Dilemma between entities (computer functions) with different strategies, set them off to face each other, asigning points per round won, to see which strategy evolved as winner.
Axelrod, did not himself propose all strategies. Instead, he sent out an invitation to several scientists in the english-speaking world who proposed the strategies.
The book is very detailed in understanding first the empiric evolution of the Winning Strategy, TIT_FOR_TAT (Stategy that would mimic the behaviour of the opposite player in the next round), and then its mathematical derivation.
The four suggestions that spring out from this experiment, and as porposed by Axelrod are:
1.- Don't be envious
2.- Don't be the first to defect (betray)
3.- Reciprocate both Cooperation and Defection
4.- Don't be too clever.
I cannot but wonder at the inmense practical applicability of the rules here derived.
Product Details,
Paperback: 264 pages
Publisher: Basic Books; Revised edition (December 4, 2006)
Language: English
ISBN-10: 0465005640
ISBN-13: 978-0465005642
Product Dimensions: 7.8 x 5.2 x 0.7 inches
The experiment is quite simple and at the same time ingenious. It handles around the problem of cooperation. This is, how is cooperation explained in a darwinian context, where all entities are supposed to be modelled as selfish, without central authority, and looking for their own best outcome? It is evident that cooperation can lead to this, that is all cooperators being best off, but, then how is cooperation evolved and configured?
To model this, Axelrod proposed a "Game" of iterated Prisioner's Dilemma between entities (computer functions) with different strategies, set them off to face each other, asigning points per round won, to see which strategy evolved as winner.
Axelrod, did not himself propose all strategies. Instead, he sent out an invitation to several scientists in the english-speaking world who proposed the strategies.
The book is very detailed in understanding first the empiric evolution of the Winning Strategy, TIT_FOR_TAT (Stategy that would mimic the behaviour of the opposite player in the next round), and then its mathematical derivation.
The four suggestions that spring out from this experiment, and as porposed by Axelrod are:
1.- Don't be envious
2.- Don't be the first to defect (betray)
3.- Reciprocate both Cooperation and Defection
4.- Don't be too clever.
I cannot but wonder at the inmense practical applicability of the rules here derived.
Product Details,
Paperback: 264 pages
Publisher: Basic Books; Revised edition (December 4, 2006)
Language: English
ISBN-10: 0465005640
ISBN-13: 978-0465005642
Product Dimensions: 7.8 x 5.2 x 0.7 inches
Labels:
Cooperacion,
Dilema del Prisionero,
Libros
Friday, March 13, 2009
Tuesday, March 10, 2009
Ant Colony Behaviour - BBC Program
The BBC radio show "Frontiers" aired some time ago a very interesting 30 min program on the evolution of self organization, show where ant colonies were initially described and several other examples were shown.
This very well told program shows that emergence of collecive behaviours is apparently inherent of complex systems.
width="290" height="65">
This very well told program shows that emergence of collecive behaviours is apparently inherent of complex systems.
width="290" height="65">
Ant Colony Optimization - Introduction
Ant Colony Optimization (ACO) is a methodology that studies artificial systems taking inspiration from the behavior of real ant colonies and which are used to solve discrete optimization problems. In 1999, the Ant Colony Optimization metaheuristic was defined by Dorigo, Di Caro and Gambardella.
The first ACO system was introduced by Marco Dorigo in his Ph.D. thesis (1992), and was called Ant System (AS). AS is the result of a research on computational intelligence approaches to combinatorial optimization that Dorigo conducted at Politecnico di Milano in collaboration with Alberto Colorni and Vittorio Maniezzo. AS was initially applied to the travelling salesman problem, and to the quadratic assignment problem.
Since 1995 Dorigo, Gambardella and Stützle have been working on various extended versions of the AS paradigm. Dorigo and Gambardella have proposed Ant Colony System (ACS), while Stützle and Hoos have proposed MAX-MIN Ant System (MMAS). They have both have been applied to the symmetric and asymmetric travelling salesman problem, with excellent results. Dorigo, Gambardella and Stützle have also proposed new hybrid versions of ant colony optimization with local search.
For more information you may visit this Wikipedia Website
The first ACO system was introduced by Marco Dorigo in his Ph.D. thesis (1992), and was called Ant System (AS). AS is the result of a research on computational intelligence approaches to combinatorial optimization that Dorigo conducted at Politecnico di Milano in collaboration with Alberto Colorni and Vittorio Maniezzo. AS was initially applied to the travelling salesman problem, and to the quadratic assignment problem.
Since 1995 Dorigo, Gambardella and Stützle have been working on various extended versions of the AS paradigm. Dorigo and Gambardella have proposed Ant Colony System (ACS), while Stützle and Hoos have proposed MAX-MIN Ant System (MMAS). They have both have been applied to the symmetric and asymmetric travelling salesman problem, with excellent results. Dorigo, Gambardella and Stützle have also proposed new hybrid versions of ant colony optimization with local search.
For more information you may visit this Wikipedia Website
Purpose of the Blog
This Blog is intended to serve as record for my continuous research in the compelling field of Complex Adaptive systems. I will ponder on the books I read on the subject and on the different articles that are written on this field.
Subscribe to:
Posts (Atom)