Complexity in EngineeringAnatoly Levenchuk

Open University SkolkovoGlobal Challenges

12-nov-2015

2

Measures of Complexity a non--exhaustive listSeth Lloyd

http://web.mit.edu/esd.83/www/notebook/Complexity.PDF

1. Difficulty of description. Typically measured in bits. Information; Entropy; Algorithmic Complexity or Algorithmic Information Content; Minimum Description Length; Fisher Information; Renyi Entropy; Code Length (prefix-free, Huffman, Shannon- Fano, error-correcting, Hamming); Chernoff Information; Dimension; Fractal Dimension; Lempel--Ziv Complexity.

2. Difficulty of creation. Typically measured in time, energy, dollars, etc. Computational Complexity; Time Computational Complexity; Space Computational Complexity; Information--Based Complexity; Logical Depth; Thermodynamic Depth; Cost; Crypticity.

3. 3. Degree of organization. This may be divided up into two quantities: a) Difficulty of describing organizational structure, whether corporate, chemical, cellular, etc.; b) Amount of information shared between the parts of a system as the result of this organizational structure.a) Effective Complexity: Metric Entropy; Fractal Dimension; Excess Entropy; Stochastic Complexity; Sophistication; Effective Measure Complexity; True Measure Complexity; Topological epsilon-machine size; Conditional Information; Conditional Algorithmic Information Content; Schema length; Ideal Complexity; Hierarchical Complexity; Tree subgraph diversity; Homogeneous Complexity; Grammatical Complexity.b) Mutual Information: Algorithmic Mutual Information; Channel Capacity; Correlation; Stored Information; Organization.

There are a number of related concepts that are not quantitative measures of complexity per se, but that are closely related: Long--Range Order; Self--Organization; Complex Adaptive Systems; Edge of Chaos.

3

Our definition of complexityComplex system – the one that does not fit in the sole engineer’s head, thus collaboration of a team and automation of a knowledge work are mandatory.

E.g.: • Aircraft• programming-in-the small vs.programming in the large• VLSI – very large scale integration, more than 1000 transistors

on a single chip (now transistor count is more than 20bln. – FPGA Virtex-Ultrascale XCVU440)

• Artificial neural network – 16bln. parameters.

4

Sources of ideas for fighting complexity

• Software engineering / computer hardware engineering

• Banking, insuarance, security market• Retail industry (one of the leaders now!)• Transport engineering (aerospace, railway,

automotive)• Other mechanical engineering• Civil engineering

5

Engineering: complexity is about number of independent parts

PP&P – process, power & petroleumPLM – product life-cycle management

From Dassault Systemes presentation

How to make such people?

Hunting and gathering Settled farming

Systems Engineering: dealing with complexity.

7

Systems Engineering (SE) is an interdisciplinary approach and means to enable the realization of successful systems. It focuses on holistically and concurrently understanding stakeholder needs; exploring opportunities; documenting requirements; and synthesizing, verifying, validating, and evolving solutions while considering the complete problem, from system concept exploration through system disposal.

http://sebokwiki.org/wiki/Systems_Engineering_%28glossary%29

https://en.wikipedia.org/wiki/Apollo_programApollo landings (1969-1972)

Apollo Program• 24 astronauts orbited Moon• 12 astronauts walked on Moon• 382kg of lunar soil and rocks

returned to Earth

8

System approach in systems engineering standards and public documents

• BKCASE, Body of Knowledge and Curriculum to Advance Systems Engineering (2015), http://www.bkcase.org/

• IEC 81346 (2009), Industrial systems, installations and equipment and industrial products -- Structuring principles and reference designations -- Part 1: Basic rules

• ISO/IEC/IEEE 15288 (2015) Systems and software engineering - System life cycle processes,

• ISO 15926-2 (2003), Industrial automation systems and integration -- Integration of life-cycle data for process plants including oil and gas production facilities -- Part 2: Data model.

• ISO/IEC/IEEE 42010 (2011), Systems and software engineering - Architecture description,

• OMG Essence (2014) – Kernel and Language for Software Engineering Methods, specification http://www.omg.org/spec/Essence/Current

9

Complexity: divide and conquer

• System holonic structure• Separation of concerns• Abstracton (modeling-generation)• Learning (autoencoder-decoder)• Cognitive load management (expression

problem)• …

System in the eyes of the beholders (stakeholders).

Theatre metaphor

Stakeholder is role vs. actor/performer, office/position, rank

System approach 2.0, based on human action

11

Holonpart-whole relationship

System of interest(using system)(system in operation environment)(subsystem)

Subsystem(System of interest)(Using system)(system in operation environment)

Using system(system-of-interest)(system in operation environment)(subsystem)

Enabling system

System of Systemsconditional part-whole relationship

Enable system

Holarhyzoom – select

Leidraadse (2008), Guideline Systems Engineering for Public Works and Water Management, 2nd edition, http://www.leidraadse.nl/

There are 4 systems here:System of interest

Requirements

System of interest

Constraints(Architecture)

Using system

Stakeholder needs

14

1 2

4Enabling system

System in operation environment

3

Generations of engineering(modeling development for checking, simulation and generation)

15

Effectiveness

Time

III generationModel-based engineering: formal languages («executable code»)

II generationContemporary («classic») engineering: diagrams and drawings («pseudocode»)

I generation«Alchemy-like engineering»: informal texts and sketches

199018601400

IV generationArtificial intelligence: formal+informal computations

2020

Interdisciplinary Plurality(on one system level, even without holarchy)

On base of Fig.3ISO 81346-1

-Module

=Component

+Location

All specialties• Mechanics• Cinematics• Electrics• Electronics• Control software• Fluid dynamics• Strength• Temperature• Noise• Vibration• …

All life cycle stages• Inception• Design• Construction,

manufacturing• Operation• Maintenance• Modernization• Retirement

PLM/ALM, ERP, EAM• Product model• Project model

16

17

System definition and system descriptionISO 42010 + OMG Essence

Basic system structuresISO 81346

• =Components• -Modules• +Locations

• Multiple variants of representations of each system aspect.• This is only basic system aspects, there are multiple other

system structure types!• Rare completely separated. Usually presented in hybrid form.

18

Hybrid diagrams• There are few ontology engineers, you should not expect too much

formalism. • Most of system descriptions are hybrid (with components and modules

are mixed).• Terminology can differ (e.g. “component” can be “functional element”

and even “module”).

19

Component diagrams (principal schemas)

20

21

Principal schema complexity

• Great metamodels (discipline)• Modelers (collaboration)• Model checking (formalization)• Generate!• Simulation

Module diagram examples (1)

22

FR160B PCB 2-Layer USB Portable Power Module -- - Green (3.5 x 2.6 x 1.5cm)

Model FR160BQuantity 1Color GreenMaterial PCB

Features

Input: 5V/800mA; Output: 5V/1A; LED lightening; With protection board on COB; Output current limited protection

Application Great for DIY project

Other ON (Press button) / OFF (Automatically)

Packing List 1 x Module

Module diagram examples (2)

Intellect stack1. Application2. Cognitive architecture 3. Learning algorithm4. Numerical libraries and frameworks5. Scientific computing programming language6. Hardware acceleration of computations 23

http://www.slideshare.net/Techtsunami/cn-prt-iot-v1

http://www.w3.org/2001/12/semweb-fin/w3csw

http://ailev.livejournal.com/1210678.html

Semantic web stack

Networking Layer Comparison

24

Modules: key for complexity

• Modularity: links have a price! The more links, the more price! (http://arxiv.org/abs/1207.2743)

• Modules: black-boxes with functions, available via interfaces

• Interfaces: communications. Conway law, reverse Conway maneuver.

• Optimization: DSM

Logical and physical architectures matchingISO 81346-1

Figure 7

25

Logical architecture (component structure, functional decomposition) iteratively match with physical architecture (module structure, work product decomposition).

Most complex part: modular synthesis

26

Multiscale * beyond life cycle

<<< Inception Architecture Non-architecture part of design

Manufacturing Operation>>>

Usingsystem

IT-1 IT-2 IT-3 IT-4 IT-5

Macro IT1 IT2 IT3 IT4 IT5

Meso IT6 IT7 IT8 IT9 IT10

Micro IT11 IT12 IT13 IT14 IT15

Nano IT16 IT17 IT18 IT19 IT20

Specialization/professionalization in each cell, plus expansion to neighborsIntegration at a product level: overall table («enabling eco-system»!)

CAD/CAM/codes/PLM/CAE/ERP/EAM/… configuration and change management!

Substance (system) levels * realization (life cycle) levels

27

Expression problem

• Programming-in-the small vs. programming in the large• Granularity & modularity• Packages (Modula)• Object-oriented approach• Data bases/queries• Julia: multiple dispatch

• Functional programming – Johan van Bethem (in https://www.illc.uva.nl/Research/Publications/Reports/PP-2005-22.text.pdf): «much of logic is about a balance between the expressive power of formal languages and the complexity of performing natural tasks for them, such as model checking for truth, consistency maintenance, or valid inference. This is the thrust of many meta-theorems, including Gödel's and Tarski's celebrated result about the limitations of first-order logic. The 'Golden Rule' of logic says that gains in expressive power are lost in higher complexity».

Practice = discipline + technology

Disciplined (competent in domain) performers

Supported with needed for a discipline tools and work products.

28

University, school

(education)

Industry, professional training

Components/alpha – how it is working

Modules/work products – how it makeable

29

Domain and endeavor: KNOWLEDGE is an information that you can use in different projects (economy of thinking!)

• Domain/discipline = thinking (operations with abstract typed objects). Changing every 30 years. Studied in schools and universities.

• Technologies/way of working = tools and work products (thinking with an exocortex). Changing in every 5 years. Trained in workplace.

• Link between discipline and technology, discipline and real life should be trained with a help of a teacher.

There is no one word from a textbook in real life

There is no one work from real life in a textbook

=Components, functional elements,Alphas

=Modules, constructive elements, work products

30

Project Essence Diagram: complexity of organization counts!

Engineering management

Engineering

Technology management

Using system

Technology management and entrepreneurship

System of interest

Enabling system

31

System life cycle practices drive alphas

http://arxiv.org/abs/1502.00121

Systems Engineering Essence

System and project life cycle (OMG Essence for systems engineering)

32

satisfied in use

represented

recognized

benefit accrued

Solution needed

viable

identified

used for retirement

consisted

used for operation

conceived

retired

parts

demonstrable

operationalclosed

prepared

under control

concluded

initiated

formed

collaborating

seeded

foundation established

in place

working well

principle established

stakeholders opportunity system definition

system realization work team way of

working

inception

development

deployment

испытания

manufacturing

retiredadjourned

readyused for

verification

involved

satisfied for deployment adressed

started

performingused for production

raw materialsIn agreement

in usevalue established

http://arxiv.org/abs/1502.00121

Case management: issue, ticket, bug

33

issue/reque

st

task/order

notice

How to fight development flow complexity?

Ideas sources:• сomputer operating systems• control engineering• data communications networks• finance and economics• information theory• maneuver warfare• Manufacturing• operations research• probability and statistics• queueing theory

According to Donald Reinertsen34

35

Connectionism

• World is not symbolic! We need means to sense and process raw world complexity!

• Non-symbolic models: distributed representations.• Connectionism (e.g. deep learning): deal with informal implicit knowledge

processing.• Since 2012 (GPU enabled)

NVIDIA® Jetson™ TX1

http://www.nvidia.com/object/embedded-systems.html

36

Avatarization of engineering software• Learning of CAD and/or programming/configuration• Natural language and/or programming language• Human-computer dialog for justification of intents and constraints• Joint human-computer idea generation and/or editing of ideas by

human• Convenient dialog with software: avatar with name and image,

emotion recognition and usage

Company Virtual intelligent assistant Google Google Apple Siri Microsoft Cortana Facebook M Amazon Alexa Autodesk ???????????

37

Thank you!

Anatoly Levenchuk,TechInvestLab, presidentINCOSE Russian chapter, research directorhttp://ailev.ruailev@asmp.msk.su

Book «Systems engineering thinking» (in Russian: http://techinvestlab.ru/systems_engineering_thinking)