organisation-oriented chemical programming peter dittrich bio systems analysis group dept. of...

Organisation-Oriented Chemical Programming

Peter Dittrich

Bio Systems Analysis GroupDept. of Mathematics and Computer Science

Friedrich Schiller University Jena

Friedrich-Schiller-Universität Jena Jena Centre for Bioinformatics

Motivation

How to program IT systems using chemical-like systems?

26.08.2010 Jena Peter Dittrich - FSU & JCB Jena 2

Overview

1. Why using chemical-like systems?

2. How to find the right chemical program?

3. Example: Maximal independent set problem.

4. Chemical Organization Theory

5. Organization-oriented chemical programming

6. Evolved vs. manual design

7. Messy Chemistries

8. Outlook: Three Open Problems

26.08.2010 Jena 3Peter Dittrich - FSU & JCB Jena

Artificial chemical computing

Chemistry Helps Computing

Real chemical

computing

[J. S. Astor, C. Adami:., Artificial Life 6(3), 189-218, 2000]


COG (MIT, Brooks et al.)


PSI (D. Dörner)


Growing Artificial NNs

[J. S. Astor, Christophs Adami: A Developmental Model for the Evolution of Artificial Neural Networks., Artificial Life 6(3), 189-218, 2000 http://norgev.alife.org/]

[Astor/Adami]


Morpho-Genetic Systems

[Source: Espinosa-Soto, C., P. Padilla-Longoria, E. R. Alvarez-Buylla; The Plant Cell, 16:2923-2939 (2004)]

Arabidopsis wild type and ap3 mutant flower

genetic network(local rules)


Amorphous Computing


Formation of Artificial Organs

cf:. Uwe Brinkschulte et al.26.08.2010 Jena 12Peter Dittrich - FSU & JCB Jena

Formation of Artificial Organs

cf:. Uwe Brinkschulte et al.26.08.2010 Jena 13Peter Dittrich - FSU & JCB Jena

Organic Middleware OCmu

See T. Ungerer, Univ. Augsburg


Characteristics of Applications

• Low-level control in a distributed, dynamic, unpredictable, and unreliable IT system.


Why chemistry?

Compare with conventional and connectionistic computing.

Peter Dittrich - FSU & JCB Jena 1726.08.2010 Jena

“Invisible Networks”


Structure-Function-DualismSelf-Modification / Strange Loop

• Dualism of – structure and function – data and program– Tape and machine

• Self-modification(s. higher-order & generative programming)

• Strange loop

Examples:Pi-calculus

FRAGLETS (Tschudin et al.)

Overview










Programming Chemical Systems

MICRO(reaction rules)

MACRO(desired behavior)

Abstraction Instantiation


Approaches

1. Optimization (e.g. EA)

2. Engineering (Design)

3. Compiling (e.g., DNA sticker model)

4. Copying (e.g., bionics)

5. Analytic/Proof

Approaches

• Optimization (e.g. EA or „Trial and Error“)

Evolving self-organizing systems is difficult.E.g.: (J. Ziegler / W. Banzhaf)

Approx. 10 functional nodes evolvable



Approaches

1. Optimization (e.g. EA)

2. Engineering (Design)

3. Compiling (e.g., DNA sticker model)

4. Copying (e.g., bionics)

5. Analytic/Proof


Programming by human design requires predictability



Abstraction InstantiationUnderstandCausality





Abstraction Instantiation“A Theory of Emergence”

can only partially

explain the micro-

macro-link

(cf. halting problem)





Abstraction Instantiationmany “partial” theories

Overview











Sketch of an Example Application

1. Inject molecules

2. Molecules distribute

3. Cells differentiate (self-organize)

4. A cell is removed

5. Reorganize


Example: Chemical Program

[1] N. Matsumaru, T. Hinze, and P. Dittrich. Organization-oriented chemical programming for Distributed ArtifactsInternational Journal of Nanotechnology and Molecular Computation (submitted)

Reactions within a membrane

Transport between

membranes i and j


Example: MIS chemistry


Overview











„Chemical Organization“

Organization := a set of molecular species that is(algebraically) closed andself-maintaining

Reaction inside the organization produce only species of that

organization.

Within a self-maintaining set, all species consumed by a reaction

can be produced by a reaction within the self-maintzaining set while no species concentration in the set decreases.

[Speroni di Fenizio/Dittrich (2005/7, Bull. Math. Biol. 2007) inspired by Fontana, Buss, Rössler, Eigen, Kauffman, Maturana, Varela, Uribe]


Practical View

1

32

4

Chemical Organization

Theory

OrganizationsReaction network

1

32

4

Organization

[P. Dittrich, P. Speroni di Fenizi, Chemical Organization Theory, Bull. Math. Biol., 2007]


Practical View

{1}

{2, 3}

{1,2,3,4}

{ }

Hasse diagram of organizations



Theory1

32

4



Practical View

1

32

4

Thoerie chemischer

Organization{1}

{2, 3}

{1,2,3,4}

{ }

Dynamics

[2]

[3]

[4][1]


Theory





1. Example: MIS chemistry




Organization := a set of molecular species that is(algebraically) closed andself-maintaining

Reaction inside the organization produce only species of that

organization.

Within a self-maintaining set, all species consumed by a reaction

can be produced by a reaction within the self-maintzaining set while no species concentration in the set decreases.

[Speroni di Fenizio/Dittrich (2005/7, Bull. Math. Biol. 2007) inspired by Fontana, Buss, Rössler, Eigen, Kauffman, Maturana, Varela, Uribe]




The empty organization


Example 1

d

c

a

b

e

-> a

2

a + b -> 2 b

2

a + c -> 2 c

b -> d

c -> d

b + c -> e

a ->b ->c ->d ->e ->


Example 1

d

c

a

b

e

2

2

Organization {a, b, d}


Checking for Closure

d

c

a

b

e

2

2



Checking for Self-Maintenance

d

c

a

b

e

2

2




d

c

a

b

e

2

2


1. Find flux vector

0 0

00

0

outside of org. = 0



d

c

a

b

e

2

2


0 0

00

0

1. Find flux vector

outside of org. = 0

inside of org. > 01

1

1

10

9 8



d

c

a

b

e

2

2


1. Find flux vector

0 0

00

0

outside of org. = 0

inside of org. > 01

1

1

10

9 8

2. Check production rates

outside of org. = 0 (closure)

0

0



d

c

a

b

e

2

2


1. Find flux vector

0 0

00

0

outside of org. = 0

inside of org. > 01

1

1

10

9 8

2. Check production rates

outside of org. = 0 (closure)

0

0

7

inside of org. 0 (self-maint.)

0

0


All Organizations

d

c

a

b

e

2

2

All Organizations


All Organizations

d

c

a

b

e

2

2

All Organizationen

a

a


All Organizations

All Organizations

d

c

a

b

e

2

2a b d

a


All Organizations

All Organizations

a b d

d

c

a

b

e

2

2a c d

a


All Organizations

a b d a c d

d

c

a

b

e

2

2

a b c d e

a



„Overlapping Hierachy“

a b d a c d

d

c

a

b

e

2

2

a b c d e

a


Generate Organization GO(A) from a set A

• GO(A) = GSM(GCl(A))


Union and Intersection of Organizations

• GO(O1 U O2)



„Overlapping Hierachy“

a b d a c d

d

c

a

b

e

2

2

a b c d e

a



DYNAMICS

Assumption: (Feinberg Condition)

0:)( species allfor 0)(

)(

ir xrLHSiv

Nv

x

xx


Theorem: Fixed points are instances of organizations

a b d a c d

a b c d e

a

)(xx Nv

][

][

][

][

][

e

d

c

b

a

x

)(0 0xNv

0

11

0

7

4

0x

differential equation

fixed point / (stationary solution)


{ a, b, d }

a b d

Dittrich/Speroni d.F., Bull. Math. Biol., 2007


Limit Set Theorem

Given a state x’ and its limit set L,

now take a state x from the limit set

and take the set of species that have strictly positive concentrations

Is this set an organization?

If it is minimal, definitely yes.

If not, we do not know, but no counterexample, yet.[Stephan Peter 2009/10, paper to be submitted]


Organizational Analysis in Space

a. Global Analysis• considers a concrete global topology, as

shown before

b. Local Analysis• all possible local environments are

represented by inflow and outflow reactions.



Global Analysis



1. Example: MIS simulation



Organizational Analysis in Space

a. Global Analysis• considers a concrete global topology, as

shown before

b. Local Analysis• all possible local environments are

represented by inflow and outflow reactions.



Local Analysis

local environment is modeled by inflow and outflow


no neighbors


Local Analysis



no neighbors one “0” neighbors


2.c Local Analysis



Overview











Organization-Oriented Chemical Programming

Computation should be understood as a transition between organizations.


Seven Principles for Organization-Oriented Chemical Programming

P1: There should be one organization for each output behavior class

P2: The result should be in the closure of the input.

P3: The input should generate the organization representing the desired output

P4: Eliminate organizations not representing a desired output

P5: An output organization should have no organization below

P6: Assure, if possible, stoichiometrically the stability of an output organization

P7: Use kinetic laws for fine tuning


Overview











Chemical Flip-Flop: A controllable bi-stable chemical

system

Reaction network

cDA

CdA

CDa

Cda

dCB

DcB

DCb

Dcb

0

0

0

0

Dd

Cc

Bb

Aa

256 possible sets of molecular species

Cf. Matsumaru et. al


Organizations for different inflows

Cf. Matsumaru et. al

26.08.2010 Jena Peter Dittrich - FSU & JCB Jena 75Cf. Matsumaru et. al

26.08.2010 Jena Peter Dittrich - FSU & JCB Jena 79Cf. Matsumaru et. al


Evolution vs. manual design Example: Chemical Flip-Flop

[2] T. Lenser, N. Matsumaru, T. Hinze, P. Dittrich. Tracking the Evolution of Chemical Computing Networks. In S. Bullock, J. Noble, R. Watson, M.A. Bedau (Eds.), Proc. of Artificial Life XI, pp. 343-350, MIT Press, 2008

[3] N. Matsumaru, T. Lenser, F. Centler, P. Speroni di Fenizio, T. Hinze, and P. Dittrich, Common organizational structures within two chemical flip-flop, Proceeding of International Workshop on Natural Computing, 2008

evolved manually designed


3. Evolution vs. manual design


[3] N. Matsumaru, T. Lenser, F. Centler, P. Speroni di Fenizio, T. Hinze, and P. Dittrich, Common organizational structures within two chemical flip-flop, Proceeding of International Workshop on Natural Computing, 2008

evolved manually designed


3. Evolutionary Process: Fitness



3. Evolutionary Process: Number of Organizations

halt input

reset input

set input



3. Evolutionary Process: Average Number of Organizations

halt input

reset inputset input


Overview











Messy Chemistries: Organizational Evolution

current state(concentration vector)


Organizational Evolution

set of species present


Organizational Evolutiongenerate organization

Overview










Outlook: Three Open Problems

1. How to design chemical-like programs?

2. Where do they perform well?

3. What is an ideal chemical language?


Acknowledgements

Pietro Speroni di Fenizio, Naoki Matsumaru,Florian Centler, Christoph Kaleta, Christian Knüpfer, Thorsten Lenser, Thomas Hinze, Dennis Görlich, Gabi

Escuela, Maiko Lohel, Stefan Artmann, Clemens Beckstein, Stephan Diekmann

Funding: BMBF, EU (FP6, FP7), DFG, RLS, DAAD, JSMC (DFG), HIGRADE

Friedrich-Schiller-Universität Jena Jena Centre for Bioinformatics

Commercial: We are looking for Postdocs in

organic & chemical computing and cell cycle modeling


1. Example: Maximum Independent Set Problem (MIS)

• New chemical algorithm– only four species– no distinction of neighbors required



4. Simulator for Quantitative Evaluation of Distributed Chemical

Computing• Specify chemical program by a set of

explicit reaction rules• Specify topology by a graph• Run stochastic simulation• Visualize dynamics



Computing


4. Looking at the dynamics of one node (V2)



Computing


5. Organization Oriented Chemical Computing for Artificial Development

• OO-ChemProg applied to cell differentiation / morphogenesis

• Differentiation can be understood as a transition between organizations

-> might be useful inArtificial Development + Evolution


6. Emergent Control

(manuscript in preparation)


6. Feed back control is everywhere

Source: http://upload.wikimedia.org/wikipedia/en/4/40/Feedback_loop.JPG

s

6. Emergent Control


6. Architecture for Emergent Control

macro-to-microtranslator

system to be controlledfeedforward controller

mic

ro

level

macro

le

vel

macro goals

micro rules or

downward causation

desired macro

behavior(manuscript in preparation)


6. Architecture for Emergent Control



mic

ro

level

macro

le

vel

macro goals

micro rules or

downward causation

desired macro


feedback dynamics


6. Current Situation



mic

ro

level

macro

le

vel

macro goals micro rules

desired macro



6. Strategies for Building a Macro-to-Micro Translator

• Manual “intelligent” design, Policies• Evolution (optimization, playing, etc)• Theory• Mimicking• Compiling• Experiment and numerical inversion



6. In practical applications, emergent control will often be

combined with feedback control.



mic

ro

level

macro

le

vel



6. Emergent Control Examples Studied

a. Balance the number of particles of two types

b. Control the number of clusters in a population of evolving entities.

26.08.2010 Jena 107

6.A Example: Balance the number of particles of two

types

Peter Dittrich - FSU & JCB Jena

26.08.2010 Jena 108

6.A Example: Recovery Time

feedback control

emergent control

Time (arbitrary units)


26.08.2010 Jena 109

6.A Example: Cost

feedback control

emergent control

Time (arbitrary units)

Cost (arbitracy units)



6. Emergent Control Conclusions

• Emergent control is fundamentally different from feedback control.

• In emergent control it is more difficult to consider the user’s demands.– There are various approaches, but no satisfying (i.e. general enough) macro-

to-micro translators yet.

• Emergent control appears to be more costly.• Macro-level models of the dynamics are not

enough for quantitative evaluation.• a powerful abstraction of emergent (self-

organizing) control is needed.

II. Outlook Phase III



1. Controlled emergent control

2. Structured molecules

3. Application scenarios: “chemical” middleware and sensor net


Towards a Chemical Middleware (with Augsburg)



1. Controlled emergent control

2. Structured molecules

3. Application scenarios: “chemical” middleware and sensor net

4. Bringing European community of “chemical-like OC” together. European Mini Workshop (April 2010)

- very focused, in-silico only- How to design/program?- Quantitative evaluation?


Acknowledgement & Jobs

• Thorsten Lenser• Christoph Kaleta• Pietro Speroni di Fenizio• Gabi Escuela Funding: DFG

We are looking for Postdocs and Phd students in in-silico chemical computing (DFG, OC SPP) and in-vivo chemical computing (EU, FP7, CHEM-IT). Contact: Peter Dittrich (di.ttri.ch)


1st phase : primaryRefereed Journal:[1] Naoki Matsumaru, Florian Centler, Pietro Speroni di Fenizio, and Peter Dittrich.Chemical Organization Theory as a Theoretical Base for Chemical Computing.International Journal of Unconventional Computing, 3(4):285{309, 2007

Book Chapter:[2] Naoki Matsumaru, Thorsten Lenser, Thomas Hinze, and Peter Dittrich.Toward Organization-Oriented Chemical Programming: A Case Study with theMaximal Independent Set Problem.In F. Dressler and I. Carreras, editors, Advances in Biologically Inspired InformationSystems, volume 69 of Studies in Computational Intelligence, pages 147{163. Springer,Berlin, 2007

Refereed Proceedings:[3] Peter Dittrich and Naoki Matsumaru.Organization-Oriented Chemical Programming.In 7th International Conference on Hybrid Intelligent Systems (HIS), IEEE ConferenceProceedings, pages 18{23. IEEE, 2007[4] Naoki Matsumaru and Peter Dittrich.Organization-oriented chemical programming for the organic design of distributedcomputing systems.In 1st international conference on bio inspired models of network, information and com-puting systems (BIONETICS), volume 275 of ACM International Conference Proceeding,Cavalese, Italy, 2006. IEEE.also available at http://www.x-cd.com/bionetics06cd/[5] Naoki Matsumaru, Pietro Speroni di Fenizio, Florian Centler, and Peter Dittrich.On the Evolution of Chemical Organizations.In Stefan Artmann and Peter Dittrich, editors, Explorations in the complexity of possiblelife: abstracting and synthesizing the principles of living systems, Proceedings of the 7thGerman Workshop of Articial Life, pages 135{146. Aka, Berlin, 2006[6] Naoki Matsumaru, Florian Centler, and Peter Dittrich.Chemical Organization Theory as a Theoretical Base for Chemical Computing.In Christof Teuscher and Andrew Adamatzky, editors, Proceedings of the 2005 Work-shop on Unconventional Computing: From Cellular Automata to Wetware, pages 75{88.Luniver Press, Beckington, UK, 2005[7] Peter Dittrich.The Bio-Chemical Information Processing Metaphor as a ProgrammingParadigm for Organic Computing.In U. Brinkschulte, J. Becker, C. Hochberger, T. Martinetz, C. Muller-Schloer,�H. Schmeck, T. Ungerer, and R. Wurtz, editors, ARCS '05 - 18th International�Conference on Architecture of Computing Systems 2005, pages 95{99. VDE Verlag,Berlin, 2005[8] Peter Dittrich.Chemical Computing.In Jean-Pierre Ban^atre, Pascal Fradet, Jean-Louis Giavitto, and Olivier Michel, editors,Unconventional Programming Paradigms, International Workshop UPP 2004, Le MontSaint Michel, France, September 15-17, 2004, Revised Selected and Invited Papers,volume 3566 of LNCS, pages 19{32. Springer, Berlin, 2005

Refereed (Extended) Abstract:[9] Naoki Matsumaru, Thorsten Lenser, Thomas Hinze, and Peter Dittrich.Designing a Chemical Program using Chemical Organization Theory.BMC Systems Biology, 1(Suppl 1):P26, 2007.from BioSysBio 2007: Systems Biology, Bioinformatics, and Synthetic Biology, Manchester,UK, 11-13 January 2007


1st phase: secondaryRefereed Journal:[10] Christoph Kaleta, Florian Centler, and Peter Dittrich.Analyzing Molecular Reaction Networks: From Pathways to Chemical Organizations.Mol. Biotechnol., 34(2):117{123, 2006[11] Naoki Matsumaru, Florian Centler, Pietro Speroni di Fenizio, and Peter Dittrich.Chemical organization theory applied to virus dynamics.it - Information Technology, 48(3):154{160, 2006

Refereed Proceedings:[12] Florian Centler, Pietro Speroni di Fenizio, Naoki Matsumaru, and Peter Dittrich.Chemical organizations in the central sugar metabolism of Escherichia Coli.In Mathematical Modeling of Biological Systems, Volume I. A Birkhauser book, 2007�[13] Naoki Matsumaru, Pietro Speroni di Fenizio, Florian Centler, and Peter Dittrich.A Case Study of Chemical Organization Theory Applied to Virus Dynamics.In Jan T. Kim, editor, Systems Biology Workshop at ECAL 2005, Workshop ProceedingsCD-ROM, Kent, UK, 2005[14] Thomas Hinze, Raael Faler, Thorsten Lenser, Naoki Matsumaru, and PeterDittrich.Ezient chemisch rechnen durch deterministische Reaktionssysteme mitRegelpriorisierung.In M. Droste and M. Lohrey, editors, Proceedings of 17. Theorietag Automaten undFormale Sprachen, pages 68{73. Universitat Leipzig, 2007�[15] Thomas Hinze, Sikander Hayat, Thorsten Lenser, Naoki Matsumaru, and PeterDittrich.Hill Kinetics Meets P Systems: A Case Study on Gene Regulatory Networksas Computing Agents in silico and in vivo.In G. Eleftherakis, P. Kefalas, and G. Paun, editors, Proceedings of the Eight Workshopon Membrane Computing (WMC8), pages 363{381. SEERC Publishers, 2007[16] Thomas Hinze, Sikander Hayat, Thorsten Lenser, Naoki Matsumaru, and Peter Dittrich.Hill Kinetics Meets P Systems.In G. Eleftherakis, P. Kefalas, G. Paun, G. Rozenberg, and A. Salomaa, editors, Mem-brane Computing, volume 4860 of LNCS, pages 320{335. Springer Verlag, 2007Refereed (Extended) Abstract[17] Peter Dittrich, Thomas Hinze, Bashar Ibrahim, Thorsten Lenser, and Naoki Matsumaru.Hierarchically Evolvable Components for Complex Systems: Biologically InspiredAlgorithmic Design.In J. Jost, D. Helbing, H. Kantz, and A. Deutsch, editors, Proceedings of the EuropeanConference on Complex Systems (ECCS2007), page 85. TU Dresden, 2007


2nd phase: primaryRefereed Journal:[1] N. Matsumaru, T. Hinze, and P. Dittrich.Organization-oriented chemical programming for MIS problem.International Journal of Nanotechnology and Molecular Computation (submitted)

Refereed Proceedings:[2] T. Lenser, N. Matsumaru, T. Hinze, P. Dittrich. Tracking the Evolution of Chemical Computing Networks. In S. Bullock, J. Noble, R. Watson, M.A. Bedau (Eds.), Proceedings of the Eleventh International Conference on the Simulation and Synthesis of Living Systems (Artificial Life XI), ISBN 978-0-262-75017-2, pp. 343-350, MIT Press, 2008

[3] N. Matsumaru, T. Lenser, F. Centler, P. Speroni di Fenizio, T. Hinze, and P. Dittrich,Common organizational structures within two chemical flip-flop, Proceeding of International Workshop on Natural Computing, 2008

Dissertation:[4] N. Matsumaru

Chemical Programming to Exploit Chemical Reaction Systems for Computation, Friedrich-Schiller-University Jena (submitted on June)


2nd phase: SecondaryRefereed Journal:[5]. C. Kaleta, F. Centler, P Speroni di Fenizio, P. Dittrich : Phenotype prediction in regulated metabolic networks, BMC Systems Biology 2008, 2:37 (25 April 2008)[6] B. Ibrahim , S. Diekmann, E. Schmidt, P. Dittrich : In--Silico Modling of the Mitotic Spindle Assembly Check point, PLoS ONE 3(2):e 1555, 2008[7]. F. Centler, C. Kaleta, P. Speroni di Fenizio, P. Dittrich : Computing Chemical Organizations in Biological Networks Bioinformatics, 24: 1611-1618, 2008

Refereed Proceedings:[8] T. Hinze, R. Fassler, T. Lenser, N. Matsumaru, P. Dittrich. Event-Driven Metamorphoses of P Systems. In P. Frisco, D.W. Corne, G. Paun (Eds.), Prel. Proceedings Ninth International Workshop on Membrane Computing (WMC9), pp. 209-225, Heriot-Watt University, accepted for publication in Series Lecture Notes in Computer Science, Springer Verlag, 2008[9] T. Hinze, S. Hayat, T. Lenser, N. Matsumaru, P. Dittrich : Biosignal--Based Computing by AHHL Induced Synthetic Gene Regulatory Networks. In Proc. of the FirstInternational Conference on Bio-Inspired Systems and Signal Processing (BIOSIGNALS2008), Vol. 1, pp. 162-169 , IEEE Engineering in Medicine and Biology Society, Institute for Systems and Technologies of Information Control and Communication, INSTICC press, 2008


Motivation

(1) Every biological life form processes

information on a chemical level.

(2) Bio-chemical information processing

posses a series of valuable self-x properties

CHemical Abstract Machine

(3) Chemical programming appears to

be difficult.


Aim

How to program chemical-like systems?


Challenge



Understand

Causality

[12] Peter Dittrich. Chemical Computing. In J.-P. Banatre, J.-L. Giavitto, P. Fradet, and O. Michel (Eds.), Unconventional Programming Paradigms (UPP 2004), LNCS, 3566: 19-32. Springer, Berlin, 2005


Results

1. Chemical Programming Paradigm

2. Programming Environment

3. Case Studies and Evaluation


1. Chemical Programming Paradigm

“Organization Oriented Programming”

Computation should appear as a movement within the set of organizations.

[5] P. Dittrich, N. Matsumaru,Organization-Oriented Chemical Programming, In: Proc. of 7th International Conference on Hybrid Intelligent Systems (HIS 2007), IEEE DL, 6 pages, 2007, (in print)

[2] N. Matsumaru, T. Lenser, T. Hinze, and P. Dittrich.Designing a Chemical Program using Chemical Organization Theory.BMC Systems Biology, 1(Suppl 1):P26, 2007, (extended abstract)

[11] Peter Dittrich.The Bio-Chemical Information Processing Metaphor as a ProgrammingParadigm for Organic Computing. In U. Brinkschulte, J. Becker, C. Hochberger, T. Martinetz, C. Mueller-Schloer, H. Schmeck, T. Ungerer, and R. Wuertz, editors, ARCS '05 - 18th International Conference on Architecture of Computing Systems 2005, pages 96-100. VDE Verlag, Berlin, 2005



Practical View

1

32

4


Theory


1

32

4

Organization



Practical View

{1}

{2, 3}

{1,2,3,4}

{ }




Theory1

32

4



Practical View

1

32

4

Thoerie chemischer

Organization{1}

{2, 3}

{1,2,3,4}

{ }

Dynamics

[2]

[3]

[4][1]


Theory





Organization Oriented Design Principles

1. For each solution there must be (at least) one organization containing the desired output molecules.

2. Eliminate as many other organizations as possible.

3. There must be a pathway from the input set to the desired organization.

4. The set of input molecules must generate an organization containing the target organization.

5. Make sure that the target organization is stable from a stoichiometric point of view.



2. Programming Environment

• Analysis– Static, structural (OrgAnalyszer) (FluxAnalyzer)– Dynamical (ODESolver) (Copasi)

• Programming– List of reaction rules (Text editor)– Network structure (CellDesigner.org)

• Protocols

(SBW)– Data format (SBML)

N. Matsumaru, P. Dittrich,CHEMORG I, Project Report, FSU Jena, 2007


3. Evaluation and Case Studies

a. Logic gates(e.g., chemical xor)

b. Boolean networks (e.g., chemical flip-flop)

c. Maximum independent set

[1] N. Matsumaru, F. Centler, P. Speroni di Fenizio, and P. Dittrich.Chemical Organization Theory as a Theoretical Base for Chemical Computing.International Journal on Unconventional Computing, 28 pages, 2007, (in print)

[4] N. Matsumaru, T. Lenser, T. Hinze, P. Dittrich,Toward Organization-Oriented Chemical Programming: a case study with the maximal independent set problem. In F. Dressler and I. Carreras (Eds.), Advances in Biologically Inspired Information Systems, SCI, 69: 147-163, Springer, Berlin, 2007

[6] N. Matsumaru and P. Dittrich.Organization-oriented chemical programming for the organic design of distributed computing systems. In Proc.of Bionetics, Cavalese, Italy, December 11-13, 7 pages, IEEE, 2006.

[9] N. Matsumaru, F. Centler, and P. Dittrich.Chemical Organization Theory as a Theoretical Base for Chemical Computing.In C. Teuscher and A. Adamatzky, editors, Workshop on Unconventional Computing, p. 71-82. Luniver Press, Beckington, 2005


0

0

0

Cc

Bb

Aa

cBA

CbA

CBa

cba

1

ca b0

1

00

0

0

0

1 1

1

1

Chemical XOR

A

a : a == 0

: a == 1




Chemical XOR (two inputs)

Organizations


Theory

cBA

CbA

CBa

cba

64 possible sets of molecular species1 is an organization

0

0

0

Cc

Bb

Aa

a0 B0

{ a, B, C }

[1] Matsumaru, N. et al. Int. J. Unconv. Comp., 2007 (in print)


Chemical XOR (with one input)

Organizations

Reaction network


Theory

cBA

CbA

CBa

cba

64 possible sets of molecular species3 are an organization

0

0

0

Cc

Bb

Aa

a0[1] Matsumaru, N. et al. Int. J. Unconv. Comp., 2007 (in print)


Chemical Flip-Flop (with two inputs)




Maximal Independent Set

• Def. [Independent set]A set of vertices no two of which are adjacent

• Def. [Maximal Independent set]

Given an undirected graph, an independ set is maximal if no vertex can be added to the independent set.

Note: Maximal independent set is different from maximum independent set.

There are two maximal independent sets.

The maximum independent set has the size of 3.


• Under central daemon [Luby 1985]

• Distributed system [Shukla, et al. 1995]

Algorithms for MIS problem

):.().),Neigh(( FalseIndvTrueIndnvn

):.().),Neigh(( TrueIndvFalseIndnvn

end

Neigh

begin

do 0 while

0 ,

(v))V-({v}-V

V{v} | vII

V

IV,EG


1

neighbors of:number

000ii

n

lkj snsss

i

Algebraic Chemistry for MIS problem

binis

Membership

Vertex ID

to the inde-

}),(|{ 01 Evvss jiij

010 ii ss

},,1|,{ 10 Niss ii M

N

i

i

1

RR

pendent set

[4] N. Matsumaru, T. Lenser, T. Hinze, P. Dittrich, SCI , 69:147-163, Springer, Berlin, 2007

[6] N. Matsumaru and P. Dittrich., Proc.of Bionetics, 2006.



12

03

01

02

13

02

11

2sss

ss

ss

11

02

01

12

ss

ss

} s, s, s, s, s, s{ 13

03

12

02

11

01M

13

02

03

12

ss

ss

0

0

0

13

03

12

02

11

01

ss

ss

ss

Undirected GraphReaction Network

Organizational structure





Organization := a set of molecules that is(algebraically) closed andself-maintaining

There is no reaction producingany other molecules

than the member of the set.

Within the set, all moleculesconsumed by a reaction

can be reproduced by a reaction.




12

03

01

02

13

02

11

2sss

ss

ss

11

02

01

12

ss

ss

} s, s, s, s, s, s{ 13

03

12

02

11

01M

13

02

03

12

ss

ss

0

0

0

13

03

12

02

11

01

ss

ss

ss





}s{ 12





12

03

01

02

13

02

11

2sss

ss

ss

11

02

01

12

ss

ss

} s, s, s, s, s, s{ 13

03

12

02

11

01M

13

02

03

12

ss

ss

0

0

0

13

03

12

02

11

01

ss

ss

ss





},s{ 03

12 s








12

03

01

02

13

02

11

2sss

ss

ss

11

02

01

12

ss

ss

} s, s, s, s, s, s{ 13

03

12

02

11

01M

13

02

03

12

ss

ss

0

0

0

13

03

12

02

11

01

ss

ss

ss

Undirected GraphReaction NetworkThere is no reaction producing

any other molecules than the member of the set.








12

03

01

02

13

02

11

2sss

ss

ss

11

02

01

12

ss

ss

} s, s, s, s, s, s{ 13

03

12

02

11

01M

13

02

03

12

ss

ss

0

0

0

13

03

12

02

11

01

ss

ss

ss





} s, s, s, s, s, s{ 13

03

12

02

11

01M


0

0

0

13

03

12

02

11

01

ss

ss

ss

12

03

01

02

13

02

11

2sss

ss

ss

11

03

02

01

13

01

12

2sss

ss

ss

13

02

01

03

12

03

11

2sss

ss

ss





Undirected graph Organizational structure

Reaction network

MR

:12 molecular species

:32 reaction rules

1v2v

3v

5v

6v

4v




Current and Future Work

1. Structured Molecules

2. Quantitative Evaluation

3. Demonstrator (sensor network scenario)

4. Intrinsic vs. Extrinsic Self-Organization


1. Structured Molecules

Example: implicitly defined molecules

M = { 0, 1, 2, 3, …, }

Example: implicitily defined reaction rules

a + b c with c = a + b mod 4711


1. Candidates to be evaluated

• Bit strings and Boolean expressions• Patter matching• Interacting finite state machines• Scheme• String-based P-systems• Fraglets


1. Preliminary Study of 32-bit-Sized Interacting Machines

[8] N. Matsumaru, P. Speroni di Fenizio, F. Centler, and P. Dittrich.On the Evolution of Chemical Organizations. In S. Artmann and P. Dittrich (Eds.), Proc. of the 7th German Workshop of Articial Life, p.135-146, IOS Press, Amstterdam, 2006


1. Structured Molecules: Fraglets


2. Quantitative Evaluation

• Robustnesse.g., probability of failure after perturbation

• Efficiency of self-organizatione.g., transient time until desired result appears

• ScalabilityHow do robustness and efficiency scale with system/problem size?


2. Benchmark Problem

1. Inject molecules




5. Reorganize


Benchmark Problem

1. Inject molecules




5. Reorganize


3. Demonstrator

• Real sensor network


4. Intrinsic vs. Extrinsic Self-Organization

• Focus so far: How to program?

• In this WP: How to control?


Space

[P. Speroni di Fenizi, P. Dittrich, Chemical Organizations at Different Spatial Scales, LNCS, Springer, 2007]


Evolutionary Design

http://www.esignet.net

[T. Lenser, T. Hinze, P. Dittrich, LNCS, Springer, 2007]


Final Remarks

• Mini-Workshop on chemical-like/particle based organic computing approaches.


References & Acknowledgement

Refereed Journal:[1] N. Matsumaru, F. Centler, P. Speroni di Fenizio, and P. Dittrich.Chemical Organization Theory as a Theoretical Base for Chemical Computing.International Journal on Unconventional Computing, 28 pages, 2007, (in print)

[2] N. Matsumaru, T. Lenser, T. Hinze, and P. Dittrich.Designing a Chemical Program using Chemical Organization Theory.BMC Systems Biology, 1(Suppl 1):P26, 2007, (extended abstract)

[3] N. Matsumaru, F. Centler, P. Speroni di Fenizio, and P. Dittrich,Chemical organization theory applied to virus dynamicsit - Information Technology, 48(3):154-160, 2006

Refereed Proceedings:[4] N. Matsumaru, T. Lenser, T. Hinze, P. Dittrich,Toward Organization-Oriented Chemical Programming: a case study with the maximal independent set problem. In F. Dressler and I. Carreras (Eds.), Advances in Biologically Inspired Information Systems, SCI, 69: 147-163, Springer, Berlin, 2007


[6] N. Matsumaru and P. Dittrich.Organization-oriented chemical programming for the organic design of distributed computing systems. In Proc.of Bionetics, Cavalese, Italy, December 11-13, 7 pages, IEEE, 2006.

[7] F. Centler, P. Speronni di Fenizio, N. Matsumaru, and P. Dittrich.Chemical organizations in the central sugar metabolism of Escherichia Coli. In Modeling and Simulation in Science Engineering and Technology, Post-Proceedings of ECMTB 2005, 2007. (in print)

[8] N. Matsumaru, P. Speroni di Fenizio, F. Centler, and P. Dittrich.On the Evolution of Chemical Organizations. In S. Artmann and P. Dittrich (Eds.), Proc. of the 7th German Workshop of Articial Life, p.135-146, IOS Press, Amstterdam, 2006


[10] N. Matsumaru, P. Speroni di Fenizio, F. Centler, and P. Dittrich.A Case Study of Chemical Organization Theory Applied to Virus Dynamics.In Jan T. Kim, editor, Systems Biology Workshop at ECAL 2005, Workshop Proceedings CD-ROM, 7 pages, Kent, UK, 2005

[11] Peter Dittrich.The Bio-Chemical Information Processing Metaphor as a ProgrammingParadigm for Organic Computing. In U. Brinkschulte, J. Becker, C. Hochberger, T. Martinetz, C. Mueller-Schloer, H. Schmeck, T. Ungerer, and R. Wuertz, editors, ARCS '05 - 18th International Conference on Architecture of Computing Systems 2005, pages 96-100. VDE Verlag, Berlin, 2005


Funding: DFG Grant No. Di 852/4-1


Story 1. Introduction

2. Organization Oriented Programming

3. Framework for Chemical Programming

4. Case Studies1. Boolean Logic

2. Boolean Networks flip-flop

3. Maximum independent set problem

5. Structured molecules1. Pre-liminary study with artificial chemistry

2. New methods for representation and analysis needed

3. Case studies for evaluation

4. Candidates for molecular structures (fraglets)

6. Quantitative evaluation1. Build on maximum independent set problem

2. Connect to other projects

7. Demonstrator

8. Some theory: Intrinsically vs. Extrinsically self-organizing systems

9. Evolving networks, ESIGNET, Modularization?

10. Summary of Publications

I. Results of Phase II

organisation-oriented chemical programming peter dittrich bio systems analysis group dept. of...

Documents

bioinformatics slide

real chemical computing

chemical organization

right chemical program

artificial life

artificial nns

amorphous computing

connectionistic computing