BCS APSG Theory of Systems
1 like244 views
The document discusses systems theory and provides definitions and principles about systems. It defines a system as a collection of components bound more strongly to each other than their environment. Systems can exist because of stable components and binding forces. Complex systems can exhibit emergent behaviors from simple local rules operating at a large scale. All complex adaptive systems use some form of computation, and the theory of evolution describes how selective pressure favors replication of better adapted systems in large ecosystems of variable systems.
BCS APSG Theory of Systems
- 1. The Theory of Systems: An Information Oriented Approach Geoff Sharman 1
- 2. Why? • We recognise systems in our experience: we are surrounded by them: • physical (mechanical, electrical, electronic, software, etc) • cosmological • chemical • biological • social (ethnic, organisational, governmental, etc) • General belief that a system is “more than the sum of its parts” • But what is a system? What makes the difference and how do systems work? Are there any general principles? What can we learn about building systems? • We want to present a synthesis based on known scientific principles 2
- 3. Dramatis Personae • This synthesis is based on the work of, amongst others: • Sir Isaac Newton • Sir Benjamin Thompson/Count Rumford • Ludwig Boltzmann • Charles Darwin • Alan Turing • Richard Feynman • and others …
- 4. Definition • A system is a collection of components which are more strongly bound to each other than they are to their environment (identity property) • A component may be anything, including another system (hierarchy property) • BUT note that many of these words (e.g. “component”, “bound”, “environment”) are not precisely defined, so further explanation is needed 4
- 5. Why can Systems Exist? • Ultimately, systems can exist because: • there are stable, long lived components with which to build them (e.g. protons, nuclei, atoms) • there are forces which can bind components together (e.g. strong, weak, electromagnetic, gravitation) • BUT what are the equivalents of force in a biological or social system? 5
- 6. Locality • The essence of a system is that it is local and its rules are local … as are scientific laws • Every system has a boundary, which separates what is “in the system” from what is “outside the system” • Every system has an “environment”, which is just another name for “everything outside the system” • So we can draw a diagram to show a system and its boundary 6
- 7. Schematic of a System • Boundary • Environment = everything outside system Components and local rules 7
- 8. What is the Environment? • The “environment” is everything else in the Universe and it has an effect on any system: • E.g. all systems exchange heat with their surroundings and are subject to gravitational forces • When the environmental effects are small, systems tend towards equilibrium • Every system also affects its environment and leaves its “footprint” on it 8
- 9. Binding vs. Containment • How does a system maintain its identity? • It must have: • internal binding forces which are stronger than those exerted by the environment, OR • a physical envelope which contains components and separates them from the environment • or both 9
- 10. Inputs and Outputs • Energy Energy • Matter Matter • Information Information Components 10
- 11. Systems Principle #1 • Outputs from a system must balance inputs • In Physics, this is known as the conservation of mass/energy • Systems move and change, but you cannot build a perpetual motion machine! • Sometimes also known as the First Law of Thermodynamics 11
- 12. How Systems Change • In small scale systems, change is deterministic and reversible (e.g. collisions between particles) • Given the state of the system at a given time, we can calculate a subsequent state or an earlier state using equations of motion • By “state”, we mean a complete specification of the momentum (energy & direction) of each component (e.g. particle) in the system • This state is known as information; momentum is conserved, therefore information is conserved (usually) 12
- 13. Large Scale Systems • By large scale, we mean systems with many components and, especially, many similar components (e.g. molecules of gas) • The equations of motion for a large system cannot be solved, so motion appears random • The effect of many interactions between components (e.g. collisions between gas molecules) is to share energy between the components, leading to an equilibrium state (e.g. energy profile) with many indistinguishable “microstates” • We characterise this by statistical measures, e.g. temperature and pressure 13
- 14. Ideal Gas as a System • Pressure • Temperature Volume • PV = NkT (Ideal Gas Law) Gas molecules 14
- 15. Systems Principle #2 • In large scale systems, change is irreversible • Systems are fragile! • This occurs because all microstates are equally likely but only a few are simple and “useful” • In Physics, this is known as the Second Law of Thermodynamics 15
- 17. Lack of Information • It’s impossible to have full knowledge of the state of a large scale system: calculation/measurement could take longer than the age of the universe • We perceive this as “disorder” • The Second Law of Thermodynamics is a consequence of this lack of information, usually called entropy, which always increases • S = k logW (W = number of micro states) 17
- 18. Systems Encode Information • All large scale systems have information encoded within them: • at the low level (e.g. direction of motion, magnetisation, nuclear spin, etc.) • at larger scale, a record of irreversible change • This is what makes geology, archaeology, forensic science and memory devices possible 18
- 19. Maxwell’s Sorting Demon • Maxwell suggested that the supposedly irreversible changes to a gas could be reversed by the action of a “sorting demon” • the demon would manipulate a “trap door” separating two chambers of gas • open it to allow high velocity molecules into one chamber, but close it to low velocity molecules • gradually accumulate high velocity (hot) molecules in one chamber and slow (cold) molecules in the other • Is this possible? What would it take to implement it? 19
- 20. The Sorting Demon cold gas hot gas 20
- 21. Resolving the Paradox • For almost 100 years, the sorting demon was thought to be paradoxical because it suggested that entropy could be decreased • We now know in principle how to implement the demon with sensors and a memory device • But operating the demon requires energy • And resetting the memory after each particle detection has an energy cost • This tells us that total entropy increases in the environment 21
- 22. Open vs. Closed Systems • An open system accepts inputs and outputs • A closed system is isolated from its environment, so does not accept inputs or outputs • But both these are contradictory: • a fully open system would lose its identity • a fully closed system is impossible • so most systems are partially open, partially closed 22
- 23. Systems Principle #3 • There are no closed systems • In Physics, this is known as the Third Law of Thermodynamics • It’s usually expressed as “entropy tends to zero at absolute zero” or as “it’s impossible to reach absolute zero” • In other words, you cannot isolate a system 23
- 24. Complex Systems
- 25. Non-Equilibrium Systems • Most interesting real systems (e.g. plants, animals, machines, and planets) operate in a range of states which are far from equilibrium • They take inputs of energy/matter from the environment in order to operate • They produce outputs of waste energy/matter to the environment • They have complex internal structure and may occur in large scale communities 25
- 26. Large Scale Behaviour • Some behaviour of large scale systems can be predicted from low level properties, or described by simple laws e.g. • Mass of system = sum (masses of components) acting at centre of gravity • Motion of rigid bodies with a centre of gravity is described by Newton’s Law s = vt + 1/2 at2 • Behaviour of a gas is given by Boyle’s Law PV = RT • These predictions are only valid over certain ranges 26
- 27. • Systems with many levels of nested components can exhibit complex behaviours, e.g. actuating mechanisms, moving around, sending messages, making things, etc. • They can operate as functional processes, transforming specific inputs into specific outputs • They can also protect themselves against changes in the environment and exhibit resilience • They can repair themselves and reconfigure in response to external conditions Complex Systems 27
- 28. Emergence • Sometimes, large scale complex systems exhibit surprising and unexpected behaviour, e.g. • Phase change from gas to liquid or solid, crystallisation into geometric shapes, complex motion of water waves • Chemical processes, e.g. catalysis • Biological processes, e.g. life, consciousness • We use multiple levels of description to document these emergent properties 28
- 29. The Problem of Emergence • Higher level behaviour appears to show order. The basic problem is to explain how this arises from apparent disorder in large scale systems • Some, e.g phase change, is due to micro-level interactions between atoms and molecules, as a substance is cooled • But some natural systems show self organisation, e.g. growth of living things. How does this happen? 29
- 30. Philosophical Explanations • Philosophers traditionally use a number of approaches to explain complex behaviour • Holism => i.e. systems can only be understood as a whole • Reductionism => i.e. systems can be fully understood from their basic parts • Vitalism => living systems are made of different stuff from inorganic systems • None of these explanations is satisfactory: • We gain insight by taking systems apart and understanding subsystems • We have no evidence for vitalism; living systems are made of the same basic stuff as simple systems • But can all complex behaviours, e.g. life, brains, consciousness, etc. be understood using just basic physics? 30
- 31. Emergence from Scale • Experiments with cellular automata, e.g. Game of Life, have found self sustaining patterns, showing: • Simple local rules applied over a large scale can lead to complex patterns of behaviour • Some patterns include gates and memory, making them equivalent to a computer • Some natural processes appear to be exhibit symmetry and computation, e.g. crystal growth 31
- 32. Mechanism for Emergence • What is the difference between a working system and a pile of components? or between a living creature and a dead one? • Some patterns of behaviour are independent of the actual system components or even their type, e.g. motion of fluids. How can we explain that? • The answer to these question appears to be information and computation, which enables the parts to be organised and specifies interactions 32
- 34. Systems Principle #4 • Complex systems use some form of computation • Three classes of system provide models of computation: • Finite state machines (finite fixed memory) • Pushdown automata (finite stack memory) • Turing Machines (infinite two stack memory) and equivalents, e.g. CAs • Turing showed that: • Finding the output of a TM is an undecidable problem • All TMs have equivalent capability, no matter how designed. We call this the Universality of Computation: any computer can do the same as any other 34
- 35. Finite State Machines • A finite state machine (automaton) is an abstract system which can can exist in a limited number of states, e.g. vending machine, lift • The states provide a form of memory • It can accept a defined set of input signals (often thought of as a string of symbols in some defined grammar) which cause transition from one defined state to another • It provides a restricted model of computation, with less capability than a Turing machine - but then most real computers have finite memory 35
- 36. Computation in Nature • Studies of one dimensional cellular automata suggest that some natural processes are equivalent to automata, e.g. turbulent fluids, weather systems • We know that biological cells use a computing mechanism based on DNA • Therefore, natural systems may perform computation and may exhibit universality • Although we cannot predict their behaviour (through lack of information), we may be able to simulate it 36
- 37. Small Automata • Margenstern studied Turing Machines with small numbers of symbols and states • He asked what is the smallest automaton that exhibits universality? • The answer is surprisingly small: a TM with 5 symbols and 5 states shows universality • Studies of small Cellular Automata also show universality 37
- 38. Universality of TMsno.ofsymbols 0 5 10 15 20 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 no. of states TMs known to be universal
- 40. Complex Adaptive Systems • Real world complex systems have a number of common features: • protective envelope, with input/output ports • static hierarchical structure of components with dynamic operational relationships • the capability to control their function in response to external conditions • the capability to exchange messages between components and with the external environment • the capability to model themselves and the external environment • and sometimes, the capability to replicate themselves 40
- 41. System Structure • Static hierarchical nesting of components follows from the definition of a system. Can be represented by a tree • Dynamic networked structure depends on how components “invoke” each other by passing messages. Can be represented by a graph • Graphs can include cycles (loops) and recursion 41
- 42. Control Systems • Control systems (e.g. thermostat, speed controller) can maintain some parameter at a constant value (steady state) via a negative feedback loop. Requires two data values plus an algorithm: • Desired value of parameter • Actual measured value of parameter • Algorithm for returning parameter to desired value • Most complex systems and all living things include control systems 42
- 43. Modelling • Most complex systems (e.g. computers, living things) exist within communities of similar systems • Their environment is hazardous and competitive, but may offer rewards (e.g. food) • To function successfully, they need to know about the environment by modelling it and monitoring it • Information is encoded within the system and associated with algorithms for updating the model, responding to changes, and reacting externally 43
- 44. Self Modelling • Complex systems also need to monitor and control themselves, to: • manage inputs, workload, etc. • maintain outputs with varying inputs • ensure correct operation of internal components • adapt to failures and external changes • For this, they need a model of their own structure 44
- 45. Replication • Because systems are fragile, the capability to adapt and respond will degrade with time. All systems become defunct eventually • Longer term, only systems with the capability to replicate themselves will survive • This means copying all the internal and external model information and creating a physical replica, as in the reproductive cycle of life • Artificial hardware systems are rarely capable of self replication but software often is, e.g. viruses 45
- 46. Ecosystems • Darwin recognised that animals and plants exist within competitive ecosystems, where there are many complex predator/prey relationships: “everything is connected to everything else” • He also recognised that individual plants and animals are subject to natural variation • And an ecosystem functions as a system which exerts selective pressure on individuals within it 46
- 47. Systems Principle #5 • In a large scale ecosystem of variable systems, selective pressure will favour replication of the best adapted systems • This is the Theory of Evolution. It describes a mechanism for learning about the ecosystem • It also applies to economic markets, software releases, and other systems where each iteration incorporates new adaptations • The selective pressure is customer requirements! 47
- 48. Conclusions
- 49. What have we Learnt? • All systems are subject to known scientific principles • Many natural systems show aspects of computation • All complex adaptive systems are computers • Evolution occurs in systems with many similar components, not in individual component systems • Systems development is the iterative process of learning customer requirements 49