Week 4 – System Dynamics & Behavior

• All systems are dynamic in one or more time domains

• External dynamics – what goes on outside – Behavior of the system

– Activities of the entities in the environment

• Internal dynamics – what goes on inside

• Flows of stuff between components

• Movement of components relative to each other

Motivating Question

• Given our understanding of complex organization, how should we consider dynamical behavior?

– Effects through interactions – causal relations

– Propagation of effects through the network

– Changes in activities on different scales of size

– Rates of changes, measurement and observing – how do we “know” what a system is doing?

Systems are Dynamic in One or More Time Scale

• Consider a living organism such as yourself:

– Metabolism operates in the microsecond scale

– Cell movement operates in the millisecond scale

– Muscle movements operate in the seconds scale

– Physiology operates in the seconds to minutes scales

– Digestion operates in the minutes to hours scales

– Adaptation operates in the seconds to hours to days scales

Causal Relations

• Causality involves a change of state of the system based on prior states and input events

– With no inputs processes decay or fall apart over time

– With inputs (esp. energy) processes proceed and may either grow in size/complexity, or obtain a steady state

Causal Relations – Cont.

• Event A (e.g. an input) causes event B (change of state) if: – A precedes B by time Δt (on some scale) – A never succeeds B except after some long time interval, nΔt – A is connected to B by a force or a flow

• Event A occurring causes a change of state, event B, which can act as an input to another system/process producing a causal chain: A → B → C, events separated by Δt

• Mutual or Circular Causality: A → B → C → A

• Multiple Causality: (B & Y or B OR Y)

• Stochastic Causality: A → B → C (A causes B with probability, x)

Dt Dt

A → B → C → X → Y

P = x P = y

Radius of Effect

• The system of interest, or agent, has a limited range of perception (how far away it can be from an event that will affect it)

• Causal chains can go back far in time and distant in space and still have an impact

• Major source of uncertainty

radius of perception

system of interest radius of hidden causal chains

events from long ago still able to affect the agent

temporal scale

Systems Are Always in Flux

• Dynamics of the Environment – Stochastic – unpredictable in detail – Non-stationary – long term changes in statistical properties – Chaotic – sensitive to initial conditions, no two systems

follow same trajectory

• Systems respond to their environments • Environments respond to their component systems • Adaptive systems are those that have complex, often

redundant mechanisms for dealing with changing environments while maintaining a core constancy – Life as the quintessential example of adaptive systems – Homeostasis and Autopoiesis examples

External Dynamics • Behavior of the system as a whole entity

• Activities of the entities in the environment

• All systems take energy inputs to do internal work, giving off waste heat by the 2nd Law of Thermodynamics (includes concepts in your head!)

• Systems process inputs to produce outputs:

– material products and wastes

– energy consumed and/or transformed

– messages processed for information and knowledge

– forces that alter other processes

External Dynamics

useable energy

raw material

messages

system

waste heat (unusable energy)

output (product)

waste material

sources

sinks

useable energy

raw material

message

system

waste heat (unusable energy)

output (force)

waste material

sources

sinks Behavior of a product process

Behavior of a force process

Black Box Analysis of a System

• Instrument flows into and out of the system

• Record data in time series

measure useable energy flow

measure raw material flow

measure message flow

system

measure heat flow

measure output flow

measure waste flow

sources

sinks

…

t 0

t i

t j

t k

measurements taken at time intervals Δt

energy material message heat product waste

data

Internal Dynamics

• Internal components are often subsystems that are “processes”

• Flows across the boundary and between components – Energy: Essential to do physical work

– Matter: Flows to processes

– Messages: Communications between components that can convey information

• Forces acting between components that cause movements relative to one another

Internal Dynamics in a Flow Network

inputs

output

internal feedback messages

Cooperating Sub-processes

Energy distribution

Process

White Box Analysis

• Instrument internal flows

• Record time series data as in black box analysis

system

useable energy

raw materials

messages

waste heat (unusable

energy)

output (product)

waste material

sources

sinks

stocks

controller

stocks

combining process

internal dynamics

message sender

data from measuring internal flows and stocks

Visualization of Dynamics

• Here is a model of a population representing a real population that was instrumented (births and deaths were recorded)

births deaths

juveniles adults matured

birth rate

matured death rate

adult death rate

maturation rate

transition rate

juvenile death rate

data

Population Dynamics w/wo Negative Feedback

Without stabilizing negative feedback

With stabilizing negative feedback

Here is how we visualize the white box analysis of a three stock population under two different experimental regimes.

Seminar Questions - Dynamics

• Why would we think a rock is a dynamic system? • What does it mean that a system processes

inputs to produce outputs? • How are external and internal dynamics basically

the same (hint: remember organization in structural hierarchies)?

• Are flow changes instantaneously felt in a chain of flows?

• What are the impacts of systems that are dynamic in multiple time scales?

Week 5 – Information, Knowledge, & Computation

• Messages are conveyed from a sender to a receiver

• Messages may or may not convey information

• The receiver may be able to construct a model (knowledge) from the information received

• Processing messages (data) to determine information content and construct knowledge is computation

Motivating Questions

• What is “information?”

– Uses in common language

– Technical definition

• What is “knowledge?”

– When you are “informed”

– Models of the sender and meaning

• How does computation process information and knowledge?

– Transforming messages according to their meaning

Messages • Sender/transducer (encoding in low energy

form)

• Receiver/amplifier

• Processor/constructor

System in the World Behaving

System in the Mind A mental model

Messages sent

Receiver/ amplifier Message

processed

Knowledge construction process

Information

• News of difference that makes a difference

• Measure of surprise in a message

• Amount of information is proportional to the a priori expectation of the receiver (observer) and not something the sender puts into the message – one of the hardest notions to grasp

• Information causes changes in the receiver proportional to the amount in a message