Скачать презентацию Politecnico di Torino — Ceris Hermes — Скачать презентацию Politecnico di Torino — Ceris Hermes —

62c75a2330151089c5e471aef54c7e8c.ppt

  • Количество слайдов: 42

Politecnico di Torino - Ceris – Hermes - AEIT Electricity market performance under physical Politecnico di Torino - Ceris – Hermes - AEIT Electricity market performance under physical constraints Agent based simulation and electricity market Pietro TERNA, Department of Economic and Financial Science, University of Torino terna@econ. unito. it, http: //web. econ. unito. it/terna September 25 th, 2007 Real Collegio Carlo Alberto 1

____________________ Building models: three ways ____________________ September 25 th, 2007 Real Collegio Carlo Alberto ____________________ Building models: three ways ____________________ September 25 th, 2007 Real Collegio Carlo Alberto 2

Three different symbol systems: • verbal argumentations • mathematics • computer simulation (agent based) Three different symbol systems: • verbal argumentations • mathematics • computer simulation (agent based) September 25 th, 2007 Real Collegio Carlo Alberto 3

____________________ How to use agents in simulation models: a radical view ____________________ September 25 ____________________ How to use agents in simulation models: a radical view ____________________ September 25 th, 2007 Real Collegio Carlo Alberto 4

The radical characterization of an ABM must be found into the possibility of real The radical characterization of an ABM must be found into the possibility of real – direct or indirect (via the environment) – interaction amid the agents … avoiding simplifications coming from the use aggregate (simultaneous) equations (going back to the Walrasian auctioneer) September 25 th, 2007 Real Collegio Carlo Alberto 5

____________________ Agent based simulation and real world representation ____________________ September 25 th, 2007 Real ____________________ Agent based simulation and real world representation ____________________ September 25 th, 2007 Real Collegio Carlo Alberto 6

Social simulation as a computer based way to execute complex mental experiments, dels mo Social simulation as a computer based way to execute complex mental experiments, dels mo d but also as a via to represent the complexity of real world ase nt-b age sim September 25 th, 2007 n = latio u Real Collegio Carlo Alberto 7

____________________ A dictionary ____________________ September 25 th, 2007 Real Collegio Carlo Alberto 8 ____________________ A dictionary ____________________ September 25 th, 2007 Real Collegio Carlo Alberto 8

A dictionary, from Conte R, Edmonds B, Moss S. , Sawyer R. K. , A dictionary, from Conte R, Edmonds B, Moss S. , Sawyer R. K. , Sociology and Social Theory in Agent Based Social Simulation: A Symposium Computational & Mathematical Organization Theory 7, 183 -205, 2001 “ 1. The purpose of Agent Based Social Simulation (ABSS) is to analyse the properties of social systems defined by dense patterns of interaction among autonomous, cognitive individuals. 2. The same modelling techniques that are intended to represent real social systems can also represent software systems such as the Internet and large federated data bases populated by intelligent information agents or, indeed, any other large, complex multi agent system. Multi agent based simulations (MABS) of such systems share the techniques of ABSS. ” My note: • we use frequently the name of Agent Based Model (ABM) or Agent Computational Ecomics (ACE) instead of ABSS; • in computer science the attention is devoted to Multi Agent Systems, MAS; adding “simulation” we have MABS and, in some way, ABSS. September 25 th, 2007 Real Collegio Carlo Alberto 9

____________________ A general structure for agent-based simulation models, the ERA scheme ____________________ September 25 ____________________ A general structure for agent-based simulation models, the ERA scheme ____________________ September 25 th, 2007 Real Collegio Carlo Alberto 10

ERA, ment, n Enviro ents g ules, A R NN CS GA Avatar http: ERA, ment, n Enviro ents g ules, A R NN CS GA Avatar http: //web. econ. unito. it/terna/ct-era. html September 25 th, 2007 Real Collegio Carlo Alberto 11

____________________ Tools ____________________ September 25 th, 2007 Real Collegio Carlo Alberto 12 ____________________ Tools ____________________ September 25 th, 2007 Real Collegio Carlo Alberto 12

Swarm, http: //www. swarm. org SLAPP, Swarm-Like Agent Protocol in Python, temporary at http: Swarm, http: //www. swarm. org SLAPP, Swarm-Like Agent Protocol in Python, temporary at http: //eco 83. econ. unito. it/terna/slapp ; Python at www. python. org JAS, http: //jaslibrary. sourceforge. net/ Ascape, http: //www. brook. edu/dynamics/models/ascape/ Repast, http: //repast. sourceforge. net/ Star. Logo, http: //education. mit. edu/starlogo/ didactical perspective nearly videogames Star. Logo TNG, http: //education. mit. edu/starlogo-tng/ Net. Logo, http: //ccl. northwestern. edu/netlogo/ SDML (based upon Small. Talk, as a declarative programming tool): http: //www. cpm. mmu. ac. uk/sdml/ See also ABLE, http: //www. research. ibm. com/able/ JADE, http: //jade. tilab. com/ or DAML, www. daml. org September 25 th, 2007 Real Collegio Carlo Alberto 13

We have also specialized agent based simulators, like j. ES and j. ESOF http: We have also specialized agent based simulators, like j. ES and j. ESOF http: //web. econ. unito. it/terna/jes/ a simulator useful to reproduce enterprises behavior September 25 th, 2007 Real Collegio Carlo Alberto 14

____________________ The surprising world of the Chameleons, with SLAPP From an idea of Marco ____________________ The surprising world of the Chameleons, with SLAPP From an idea of Marco Lamieri, a project work with Riccardo Taormina ____________________ September 25 th, 2007 Real Collegio Carlo Alberto 15

The reinforcement learning algorithm A direct tool: the reinforcement learning. We have a set The reinforcement learning algorithm A direct tool: the reinforcement learning. We have a set of states S, related to an environment; a set of possible actions A; a set of scalar rewards, in R. At any time t we have an agent in a state st of S and we can chose the action a in A(st). After the action it will be in st+1 with a reward rt+1. Reward are summed over time with a discount rate factor. Our agent develops the capability of mapping all the possible actions A in a state S to all the related rewards. September 25 th, 2007 Real Collegio Carlo Alberto 16

The metaphorical models we use here is that of the changing color chameleons We The metaphorical models we use here is that of the changing color chameleons We have chameleons of three colors: red, green and blue When two chameleons of different colors meet, they both change their color, assuming the third one (If all chameleons get the same color, we have a steady state situation) The metaphor is interpreted in the following way: an agent diffusing innovation or ideas (or political ideas) can change itself via the interaction with other agents: as an example think about an academic scholar working in a completely isolated context or interacting with other scholars or with private entrepreneurs to apply the results of her work September 25 th, 2007 Real Collegio Carlo Alberto 17

A simple rule for a complex environment y a l p t e L A simple rule for a complex environment y a l p t e L September 25 th, 2007 Real Collegio Carlo Alberto 18

Running or chasing for identity! • Reinforcement learning and pattern recognition, with bounded rationality Running or chasing for identity! • Reinforcement learning and pattern recognition, with bounded rationality • Agent brain built upon 9 Artificial Neural Networks September 25 th, 2007 Real Collegio Carlo Alberto 19

____________________ The surprising world of the Chameleons, with Net. Logo ____________________ September 25 th, ____________________ The surprising world of the Chameleons, with Net. Logo ____________________ September 25 th, 2007 Real Collegio Carlo Alberto 20

September 25 th, 2007 Real Collegio Carlo Alberto 21 September 25 th, 2007 Real Collegio Carlo Alberto 21

____________________ Electricity market ____________________ September 25 th, 2007 Real Collegio Carlo Alberto 22 ____________________ Electricity market ____________________ September 25 th, 2007 Real Collegio Carlo Alberto 22

ion ats igh e ork w esf T L September 25 th, 2007 Real ion ats igh e ork w esf T L September 25 th, 2007 Real Collegio Carlo Alberto 23

ion ats igh e ork w esf T L September 25 th, 2007 Real ion ats igh e ork w esf T L September 25 th, 2007 Real Collegio Carlo Alberto 24

ion ats igh e ork w esf T L September 25 th, 2007 Real ion ats igh e ork w esf T L September 25 th, 2007 Real Collegio Carlo Alberto 25

ion ats igh e ork w esf T L September 25 th, 2007 Real ion ats igh e ork w esf T L September 25 th, 2007 Real Collegio Carlo Alberto 26

ion ats igh e ork w esf T L September 25 th, 2007 Real ion ats igh e ork w esf T L September 25 th, 2007 Real Collegio Carlo Alberto 27

ion ats igh e ork w esf T L September 25 th, 2007 Real ion ats igh e ork w esf T L September 25 th, 2007 Real Collegio Carlo Alberto 28

ion ats igh e ork w esf T L September 25 th, 2007 Real ion ats igh e ork w esf T L September 25 th, 2007 Real Collegio Carlo Alberto 29

ion ats sf rk wo Te h eig L September 25 th, 2007 Real ion ats sf rk wo Te h eig L September 25 th, 2007 Real Collegio Carlo Alberto 30

____________________ From spin glasses to videogames: representing our worlds and their complexity via the ____________________ From spin glasses to videogames: representing our worlds and their complexity via the use of simple simulation tools, with agents ____________________ September 25 th, 2007 Real Collegio Carlo Alberto 31

____________________ Scratch http: //weblogs. media. mit. edu/llk/scratch/ ____________________ September 25 th, 2007 Real Collegio ____________________ Scratch http: //weblogs. media. mit. edu/llk/scratch/ ____________________ September 25 th, 2007 Real Collegio Carlo Alberto 32

September 25 th, 2007 Real Collegio Carlo Alberto 33 September 25 th, 2007 Real Collegio Carlo Alberto 33

The credit multiplier (money_supply. sb) September 25 th, 2007 Real Collegio Carlo Alberto 34 The credit multiplier (money_supply. sb) September 25 th, 2007 Real Collegio Carlo Alberto 34

____________________ Star. Logo TNG http: //education. mit. edu/starlogo/ ____________________ September 25 th, 2007 Real ____________________ Star. Logo TNG http: //education. mit. edu/starlogo/ ____________________ September 25 th, 2007 Real Collegio Carlo Alberto 35

Star. Logo TNG September 25 th, 2007 Real Collegio Carlo Alberto 36 Star. Logo TNG September 25 th, 2007 Real Collegio Carlo Alberto 36

Sperimentiamo la complessità con Star. Logo TNG termites 2 September 25 th, 2007 Real Sperimentiamo la complessità con Star. Logo TNG termites 2 September 25 th, 2007 Real Collegio Carlo Alberto 37

Sperimentiamo la complessità con Star. Logo TNG Fish and Plankton September 25 th, 2007 Sperimentiamo la complessità con Star. Logo TNG Fish and Plankton September 25 th, 2007 Real Collegio Carlo Alberto 38

____________________ Squeak http: //www. squeak. org/ ____________________ September 25 th, 2007 Real Collegio Carlo ____________________ Squeak http: //www. squeak. org/ ____________________ September 25 th, 2007 Real Collegio Carlo Alberto 39

September 25 th, 2007 Real Collegio Carlo Alberto 40 September 25 th, 2007 Real Collegio Carlo Alberto 40

____________________ Second Life http: //secondlife. com/ ____________________ September 25 th, 2007 Real Collegio Carlo ____________________ Second Life http: //secondlife. com/ ____________________ September 25 th, 2007 Real Collegio Carlo Alberto 41

September 25 th, 2007 Real Collegio Carlo Alberto 42 September 25 th, 2007 Real Collegio Carlo Alberto 42