q-bio0501016/paper.tex
1: \documentclass[12pt]{article}
2:  
3: % m\usepackage{natbib}
4: 
5: \usepackage{epsfig}
6: \usepackage{amsmath}
7: \usepackage{amssymb}
8: \usepackage{theorem}
9: 
10:  \setlength{\topmargin}{-2cm}
11:  \setlength{\textheight}{22cm}
12: 
13:  \setlength{\textwidth}{14cm}
14:  \setlength{\columnsep}{2pc} 
15:  \setlength{\oddsidemargin}{1cm}
16: 
17: 
18: 
19: \newcommand{\comment}[1]{{\bf \sc #1}} 
20: 
21: \newcommand{\figwidtha}{10cm}
22: \newcommand{\figwidthb}{7.5cm}
23: \newcommand{\figwidthfull}{16cm}
24: 
25: \newcommand{\important}[1]{{\large \em #1}}
26: 
27: \setcounter{MaxMatrixCols}{30}
28: % -----------------------------
29: % Macros
30: % **********************************************************************
31: %        BEGIN OF ORG-THEORY DEFINITIONS
32: % **********************************************************************
33: 
34: \newcommand{\Mol}{{\cal M}} % set of all molecules
35: \newcommand{\Rea}{{\cal R}} % set of all reactions, reaction relation
36: 
37: \newcommand{\reactsto}{ \to}   % arrow used for reaction
38: \newcommand{\catto}{ \Rightarrow}   % arrow used for catalytic reaction
39: % a \catto b     is equivalent to     a \reactsto a + b
40: 
41: \newcommand{\INR}[2]{(#1 \reactsto #2) \in \Rea}   % A -> B
42:      % INR - in reactions
43: 
44: \newcommand{\Pow}[1]{{\cal P}(#1)}   % the power set, set of all subsets
45: \newcommand{\PowM}[1]{{\cal P}_M(#1)}   % the power multi set, set of all multi sets
46: 
47: 
48: \newcommand{\GO}{G} % function: generate organization
49: \newcommand{\GOloc}{G^{loc}} % function: generate local organizations (a set
50:                              % of sets !!)
51: \newcommand{\GOglob}{G^{glob}} % function: generate global organization
52: \newcommand{\GOalg}{G^{alg}} % function: generate algebraic organization
53: 
54: 
55: \newcommand{\GSM}{G_{SM}} % function: generate self-maintaining set
56: 
57: \newcommand{\GSMloc}{G^{loc}_{SM}} % function: generate local self-maintaing set
58: \newcommand{\GSMglob}{G^{glob}_{SM}} % function: generate global self-maintaing set
59: \newcommand{\GSMalg}{G^{alg}_{SM}} % function: generate algebraic self-maintaing set
60: 
61: \newcommand{\GCL}{G_{CL}} % function: generate closed set
62: \newcommand{\GMM}{G_{MM}} % function: generate mass maintaining set
63: \newcommand{\GSO}{G_{SO}} % function: generate semi organization
64: 
65: 
66: \newcommand{\X}{X} % state space of the dynamical system (ODE)
67: 
68: \newcommand{\fmol}{\phi}   % function that gives the set of molecules
69:                         % for a given state, g.g., M(x) = { 1, 3, 5 }    
70: 
71: \newcommand{\thresh}{\Theta} % concentration threshold for fmol
72: 
73: %        some sets
74: 
75: \newcommand{\Xgen}[1]{X_{gen}^{#1}}   % set of all generators
76: \newcommand{\Xextgen}[1]{X_{extgen}^{#1}} % set of all internal generators
77: \newcommand{\Xintgen}[1]{X_{intgen}^{#1}} % set of all extrernal generators
78: 
79: \newcommand{\Xinst}[1]{X_{inst}^{#1}}  % set of all instances
80: \newcommand{\Xcont}[1]{X_{cont}^{#1}}  % set of all states contained in O
81: 
82: 
83: 
84: % ------------------------------------------------------------
85: %        closed sets
86: 
87: \newcommand{\allCL}{{\cal O}_{CL}}  % set of all closed sets
88: \newcommand{\unionCL}{\sqcup_{CL}} % closed set union
89: \newcommand{\intersecCL}{\sqcap_{CL}} % closed set intersection
90: 
91: % ------------------------------------------------------------
92: % ------------------------------------------------------------
93: %        self maintaining sets
94: 
95: \newcommand{\allSM}{{\cal O}_{SM}}  % set of all self maintaining sets
96: \newcommand{\unionSM}{\sqcup_{SM}} % self maintaining set union
97: \newcommand{\intersecSM}{\sqcap_{SM}} % self maintaining intersection
98: 
99: % ------------------------------------------------------------
100: % ------------------------------------------------------------
101: %        semi organization
102: 
103: \newcommand{\allSO}{{\cal O}_{SO}}  % set of all semi organizations
104: \newcommand{\unionSO}{\sqcup_{SO}} % semi organization union
105: \newcommand{\intersecSO}{\sqcap_{SO}} % semi organization intersection
106: 
107: % ------------------------------------------------------------
108: % ------------------------------------------------------------
109: %        mass maintaining sets
110: 
111: \newcommand{\allMM}{{\cal O}_{MM}}  % set of all mass maintaining sets
112: \newcommand{\unionMM}{\sqcup_{MM}} % mass maintaining union
113: \newcommand{\intersecMM}{\sqcap_{MM}} % mass maintaining intersection
114: 
115: % ------------------------------------------------------------
116: % ------------------------------------------------------------
117: %        organization
118: 
119: \newcommand{\allO} {{\cal O}} % set of all organizations
120: \newcommand{\unionO}{\sqcup} % organization union
121: \newcommand{\intersecO}{\sqcap} % organization intersection
122: 
123: 
124: % ------------------------------------------------------------
125: %        local organization
126: 
127: \newcommand{\allOlocal} {{\cal O}_{\mathrm{local}}} % set of all local organizations
128: 
129: 
130: 
131: % ------------------------------------------------------------
132: %        global organization
133: 
134: \newcommand{\allOglob} {{\cal O}_{\mathrm{global}}} % set of all global organizations
135: \newcommand{\unionglob}{\sqcup_{global}} % global organization union
136: \newcommand{\intersecglob}{\sqcap_{global}} % global organization intersection
137: 
138: 
139: 
140: 
141: 
142: % ------------------------------------------------------------
143: 
144: 
145: \newcommand{\numberof}[2]{\#(#1 \in #2)} 
146: % \newcommand{\numberof}[2]{| \{ x \in #2 |  x = #1 \} |}
147: 
148: \newcommand{\usedup}{\mathrm{usedup}}      % molecules that are used up
149: \newcommand{\produced}{\mathrm{produced}}  % molecules that are produced
150: 
151: 
152: % ------------------------------------------------------------
153: 
154: % ------------------------------------------------------------
155: %        organization
156: 
157: \newcommand{\allOglobal} {{\cal O}_{\mathrm{global}}} % set of all organizations
158: \newcommand{\unionOglobal}{\sqcup_{\mathrm{global}}} % organization union
159: \newcommand{\intersecOglobal}{\sqcap_{\mathrm{global}}} % organization intersection
160: 
161: 
162: \newcommand{\setofO}{\mathbf O} % a set of organizations
163: 
164: 
165: 
166: \newcommand{\fixedset}{fixed set}
167: 
168: \newcommand{\Input}{M_I} % set of input molecules
169: \newcommand{\Output}{M_O} % set of output molecules
170: 
171: %     vectors 
172: 
173: %     matrices
174: 
175: \newcommand{\stochmat}{\mathbf{M}}      % stoichiometric matrix of a reaction system
176: 
177: 
178: 
179: 
180: 
181: 
182: 
183: % ----------------------------------------------------------------------
184: %  Types of Systems we Study  (the levels)
185: % ----------------------------------------------------------------------
186: 
187: 
188: \newcommand{\systema}{{\it general reaction system} }
189: \newcommand{\systemas}{{\it general reaction systems} }
190: \newcommand{\systemb}{{\it consistent reaction system} }
191: \newcommand{\systembs}{{\it consistent reaction systems} }
192: \newcommand{\systemc}{{\it reactive flow system} }
193: \newcommand{\systemd}{{\it catalytic flow system} }
194: \newcommand{\systemds}{{\it catalytic flow systems} }
195: 
196: 
197: \newcommand{\Systema}{General reaction system }
198: \newcommand{\Systemb}{Consistent reaction system }
199: \newcommand{\Systemc}{Reactive flow system }
200: \newcommand{\Systemd}{Catalytic flow system }
201: 
202: \newcommand{\SystemA}{General Reaction System }
203: \newcommand{\SystemB}{Consistent Reaction System }
204: \newcommand{\SystemC}{Reactive Flow System }
205: \newcommand{\SystemD}{Catalytic Flow System }
206: 
207: 
208: % How to call the theory
209: 
210: 
211: % **********************************************************************
212: %        END OF ORG-THEORY DEFINITIONS
213: % **********************************************************************
214: 
215: 
216: % ----------------------------------------------------------------------
217: % GENERAL MATH DEFINITIONS
218: % ----------------------------------------------------------------------
219: 
220: \newcommand{\RR}{{\mathbb  R}} % real numbers
221: \newcommand{\Rel}{\RR} % real numbers
222: 
223: \newcommand{\NN}{{\mathbb  N}} % natural numbers
224: 
225: 
226: 
227: 
228: 
229: 
230: 
231: \theorembodyfont{\rmfamily}
232: \theoremheaderfont{\itshape}
233: 
234: \newtheorem{defi}{Definition} % [section]
235: \newtheorem{theorem}{Theorem}
236: \newtheorem{lemma}{Lemma} % [section]
237: \newtheorem{alg}{Algorithm} % [section]
238: \newtheorem{example}{Example}
239: 
240: 
241: \newcommand{\Def}[2]{\vspace{-.4cm}\begin{defi}  #2 \end{defi}\vspace{-0.2cm}}
242: \newcommand{\Lem}[1]{\vspace{-.4cm}\begin{lemma}  #1 \end{lemma}\vspace{-0.2cm}}
243: \newcommand{\Example}[2]{\begin{example}  #2 \end{example}}
244: 
245: % ------------------------------------------------------------
246: \newcommand{\aidx}[1]{{}}
247: \newcommand{\sidx}[1]{{}}
248: 
249: \newcommand{\apaket}[1]{\begin{center}{\em #1}\end{center}}
250: 
251: \newcommand{\ca}{\em}
252: \newcommand{\calabel}[1]{(Label: #1)}
253: 
254: 
255: \newcommand{\df}[1]{{\em #1}}
256: \newcommand{\au}[1]{{\sc #1}}
257: % \newcommand{\Def}[2]{{Definition {\bf #1}}: #2}
258: \newcommand{\com}[1]{{\tiny \it #1}}
259: \newcommand{\Lra}{\Longrightarrow}
260: 
261: \newcommand{\derivt}[1]{\frac{d}{dt} #1}
262: 
263: % --------------------------------------------------
264: 
265: \newcommand{\beq}{\begin{equation}}
266: \newcommand{\eeq}{\end{equation}}
267: 
268: \newcommand{\bsy}{\begin{center}\begin{tabular}{|l|l|p{12cm}|} \hline }
269: \newcommand{\esy}{\hline \end{tabular}\end{center}}
270: 
271: \newcommand{\ben}{\begin{enumerate}}
272: \newcommand{\een}{\end{enumerate}}
273: \newcommand{\mitem}[1]{\item {\bf #1} \\}
274: 
275: 
276: 
277: \newcommand{\mol}{\mathrm{mol}}
278: \newcommand{\round}{\mathrm{round}}
279: 
280: 
281: \renewcommand{\vec}[1]{\mathbf{#1}}
282: 
283: 
284: 
285: \newcommand{\newpagelevel}{\clearpage}
286: 
287: \begin{document}
288: 
289: 
290: \noindent
291: 
292: {\bf \large Chemical organization theory: towards a theory of constructive dynamical systems}
293: \vspace{0.3cm} \\
294: \noindent
295: {\bf Peter Dittrich, and Pietro Speroni di Fenizio}\footnote{Both authors
296: contributed equally.}
297: \vspace{0.3cm} \\
298: \noindent
299: Bio Systems Analysis Group, Jena Centre for Bioinformatics \& 
300: Department of Mathematics and Computer Science, Friedrich-Schiller-University Jena,
301: D-07737 Jena
302: \vspace{0.3cm} \\
303: \today  
304: %
305: \vspace{0.3cm} \\
306: 
307: 
308: {
309: \bf
310: Complex dynamical networks consisting of many components
311: that interact and produce each other are difficult 
312: to understand, especially, when new components may appear. 
313: In this paper we outline a theory to deal with such systems.
314: The theory consists of two parts. The first part introduces
315: the concept of a chemical organization as a closed
316: and mass-maintaining set of components. This concept allows
317: to map a complex (reaction) network to the set of 
318: organizations,  providing a new view on the system's structure.
319: The second part connects dynamics with the set of
320: organizations, which allows to map a movement of
321: the system in state space to a movement in the set of organizations.
322: }
323: \vspace{0.5cm}
324: 
325: Our world is changing, qualitatively and quantitatively.
326: The characteristics of its dynamics can be as simple as in the
327: case of a friction-less swinging pendulum, or as complex
328: as the dynamical process that results in the creative apparition of novel
329: ideas or entities. We might characterize the nature of a  dynamical process
330: according to its level of novelty production.
331: For example, the friction-less swinging pendulum implies
332: a process where the novelty is only quantitative.
333: Whereas the process of biological evolution is
334: highly creative and generates qualitative novelties, which then spread in a
335: quantitative way.
336: 
337: Fontana and Buss \cite{ac:FB94arrival} called processes and systems that
338: display the production of novelty, constructive (dynamical)
339: processes and constructive (dynamical) systems, respectively.
340: Constructive systems can be found on all levels of scientific
341: abstraction: in nuclear physics, where the collision of
342: atoms or subatomic particles leads to the creation of 
343: new particles; in molecular chemistry, where molecules can
344: react to form new molecules; or in social systems, where 
345: communication can lead to new communication \cite{soc:Luh84}.
346: As a result of a combinatorial explosion, it is easy to create
347: something that is new, e.g., a molecule or a poem that is unique in the
348: whole known universe.
349: Nevertheless, it should be noted that novelty is relative. 
350: Whether something is considered
351: to be new or not, depends on what is already there.
352: Thence it follows that whether a system appears as a
353: constructive dynamical system  is usually context dependent.
354: 
355: Despite the fact that a large amount of interesting
356: dynamical processes are constructive, classical
357: systems theory does not, conveniently, take novelty into account.
358: Classical systems theory assumes a given static set 
359: of components. For example, a classical systems analysis 
360: would first identify all components and their relations; then
361: would identify a state space with fixed dimensionality, e.g.,
362: the state of our pendulum could be described
363: by a two-dimensional vector specifying angle and angular velocity.
364: 
365: 
366: The lack of a theory for constructive dynamical systems
367: has been identified and discussed in detail by Fontana and Buss \cite{ac:FB94arrival}
368: in the context of a theory for biological organization.
369: As a solution, they suggested the important concept 
370: of a (biological) organization as
371: an operationally  closed and dynamically self-maintaining system.
372: Taking this idea further, we suggested to divide the theoretical
373: approach in a static and a dynamic analysis~\cite{ped:SD2002}.
374: Here, we present the results of our study for general reactions systems, 
375: integrating stoichiometry and the concept of mass-maintenance in the definition of
376: organizations, which is a prerequisite for a broad applicability.
377: 
378: 
379: 
380: 
381: \paragraph{Reaction Systems.} 
382: The theory described herein aims at understanding \df{reaction systems}.
383: A reaction system consists of reactants, which we refer to as \df{molecules}.
384: An interaction among molecules that lead to the 
385: appearance or disappearance of molecules is called a \df{reaction}.
386: In particular, reaction systems are used to model chemical processes,
387: but in general, they can be applied to virtually any domain where elements
388: interact to influence the production of other elements, e.g., 
389: population dynamics, evolution of language, economy, or social dynamics.
390: 
391: 
392: 
393: 
394: We have to distinguish between a reaction system as an abstract description
395: of all possible molecules (and their reactions), and a reaction vessel,
396: which contains concrete instances of molecules from the set of all possible molecules.
397: In general, the description of a reaction system can be
398: subdivided into three parts: 
399: (1) the set of all possible molecules $\Mol$, 
400: (2) the set of all possible reactions among all the possible molecules $\Rea$,
401: and (3) the dynamics, which describes how the reactions are applied
402: to a collection of
403: molecules inside a reaction vessel. 
404: 
405: 
406: 
407: 
408: 
409: \subsection*{Static Analysis}
410: 
411: \vspace{0.3cm}
412: \noindent
413: {\bf Algebraic Chemistry.} In the first part of the paper, we are only concerned with
414: the static structure of a reaction system, that is, the molecules
415: and the reactions. And instead of considering a state (e.g., a concentration vector), we
416: limit ourself to the analysis of the set of molecules present in that state.
417: We introduce the concept of an algebraic
418: chemistry, which is a reaction network, including stoichiometric information,
419: from which we will derive the organizational structure of the system.
420:  
421:  
422: 
423: \Def{algebraic chemistry}{
424: Given a set $\Mol$  of elements (called molecules)
425: and a set of reaction rules given
426: by the relation $\Rea: \PowM \Mol \times \PowM \Mol$.
427: We call the pair $\langle \Mol, \Rea \rangle$ an \df{algebraic chemistry}.
428: 
429: }
430: 
431: $\PowM C$ is the set of all multisets with elements from $C$.
432: A multiset differs from a set in the fact that the same element can appear
433: more than one time. 
434: The frequency of occurrence of an element $a$ in a multiset
435: $A$ is denoted by $\numberof a A$.
436: For simplicity, we adopt a notion from chemistry to write 
437: reaction rules.
438: Instead of writing $(\{ s_1, s_2, \dots, s_n \} ,  
439: \{ s'_1, s'_2, \dots, s'_{n'} \} ) \in \Rea$
440: we write:
441: $  s_1 + s_2 + \dots + s_n \reactsto s'_1 + s'_2 + \dots + s'_{n'}$ .
442: Given the left hand side molecules $A = \{ s_1, s_2, \dots, s_n \}$ and 
443: the right hand side molecules 
444: $B = \{ s'_1, s'_2, \dots, s'_{n'} \}$, we write
445: $  \INR A B$ instead of $(A, B) \in \Rea$.
446: $A \reactsto B$ represents a chemical reaction equation
447: where $A$ is the
448: multiset of molecules on the left hand side (also called \df{reactants})
449: and $B$ the multiset of molecules on the right hand side (also called
450: \df{products}). 
451: 
452: \vspace{0.3cm}
453: \noindent
454: {\bf Input and Output.} 
455: There are many processes that give
456: rise to an inflow and outflow, such as, incident sunlight, decaying molecules,
457: or a general dilution flow. 
458: In this paper we handle input simply by 
459: adding a
460: reaction rule $\emptyset \to a$ to the reaction rules $\Rea$
461: for every molecule $a$ that is in the inflow.
462: $\emptyset$ denotes the empty set.
463: Equivalently, for an output molecule  $a$ 
464: (e.g., a molecule that is decaying) 
465: we add the rule $a \to \emptyset$ to the reaction rules $\Rea$.
466: 
467: 
468: 
469: 
470: \paragraph{Semi-Organization.} In classical analysis, we study the movement
471: of the system in state space. Instead, here, we consider the movement from one set
472: of molecules to another. As in the classical analysis of the dynamic of
473: the system, where fixed points and attractors are considered more important
474: than other states, some sets of molecules are more important than others.
475: In order to find those sets, we introduce some
476: properties 
477: that define them, namely: closure, self-maintenance,
478: semi-organization, mass-maintenance, and finally the organization.
479: 
480: All definitions herein refer to an algebraic chemistry $\langle \Mol,
481:   \Rea \rangle$. The first requirement, called closure, assures
482: that a set of molecules contains all molecules that can be
483: produced by reactions among those molecules.
484: 
485: \Def{closed set}{
486: %
487:   A set $C \subseteq \Mol$ is \df{closed}, 
488:   if for all reactions $\INR A B$, with 
489:   $A$ a multiset of elements in $C$  ($A \in \PowM C$),
490:                                 %
491:   $B$ is also a multiset of elements in $C$ ($B \in \PowM C$). 
492: }
493: 
494: Given a set $S \subseteq \Mol$, we can always generate
495: its closure $\GCL(S)$ according to the following definition:
496: 
497: \Def{generate closed set}{
498: %
499: Given a set of molecules $S \subseteq \Mol$,
500: we define $\GCL(S)$ as the smallest closed set
501: $C$ containing $S$. We say that $S$ \df{generates the closed set} 
502: $C = \GCL(S)$ and we call $C$ the \df{closure} of $S$. 
503: }
504: 
505: The closure implies a union and an intersection operator on
506: the closed sets $\allCL$ of an algebraic chemistry.
507: Given two closed sets $U$ and $V$,  the
508: closed set generated by their union ($U \unionCL V$) and intersection
509:  ($U \intersecCL V$) is defined as:
510:  $U \unionCL V      \equiv  \GCL(U \cup V)$, and  
511:  $U \intersecCL V   \equiv  \GCL(U \cap V)$, respectively.
512: %
513: Trivially, closed sets form a lattice
514: $\langle \allCL, \unionCL, \intersecCL \rangle$,
515: which is a common algebraic structure
516: (a poset in which any two elements have a greatest lower bound
517: and a least upper bound, see supl. mat.).
518: The property of closure is important, because the 
519: closed set represents the largest possible set that
520: can be reached from a given set of molecules.
521: Furthermore a set that is closed cannot generate 
522: new molecules and is in that sense more stable than
523: a set that is not closed.
524: 
525: %
526: 
527: The next important property, called self-maintenance,
528: assures that every molecule that is used-up within a set, is
529: produced within that set.
530: \Def{self-maintaining set}{
531:                                 %
532:   A set of molecules  $S \subseteq \Mol$
533:   is called \df{self-maintaining}, if all 
534:   molecules $s \in S$ that are used-up within $S$ 
535:   are also produced within that set $S$.
536: } 
537: More precisely, we say that a molecule $k \in \Mol$ is \df{produced}
538: within a set $C \subseteq \Mol$, if there exists a reaction $(A \to B)$, with 
539: %
540: $A \in \PowM C$, 
541: and $\numberof k A < \numberof k B$.
542: In the same way, we say that a molecule $k \in C$ is \df{used-up} 
543: within the set $C$, if, within the set $C$,  
544: there is a reaction  $(A \to B)$ with 
545: %
546: $A \in \PowM C$, 
547: and $\numberof k A > \numberof k B$.
548: 
549: 
550: %
551: 
552: Taking closure and self-maintenance together, we arrive at
553: the important concept of a semi-organization.
554: 
555: \Def{semi-organization}{
556: %
557: A \d{semi-organization} $O \subseteq \Mol$
558: is a set of molecules that is closed and self-maintaining.
559: }
560: 
561: %
562: 
563: \paragraph{Organization.}
564: In a semi-organization, all molecules that are used-up are produced;
565: but, this does not guarantee that the total amount of mass can be maintained.
566: A small, but important example is the reversible reaction in a flow reactor:
567: $\Mol = \{a, b \}, 
568: \Rea = \{ a \reactsto b, 
569:           b \reactsto a , 
570:           a \reactsto \emptyset,
571:           b \reactsto \emptyset
572: \}$, where both molecules also decay. $O = \{ a, b \}$ is a semi-organization,
573: because the set is closed, $a$ is produced by the reaction $b \reactsto a$,
574: and $b$ is produced by the reaction $a \reactsto b$. But, obviously, the
575: system $\{ a, b\}$ is not stable, because both molecules decay and are not sufficiently
576: reproduced, so that they will finally vanish.
577: %
578: The solution to this problem is to consider the overall ability 
579: of a set to maintain its total mass. We call such sets mass-maintaining:
580: 
581: 
582: %
583: 
584: \Def{mass-maintaining}{
585: Given an algebraic chemistry $\langle \Mol, \Rea \rangle$
586: with $m = |\Mol|$ molecules and $n = |\Rea|$ reactions, and let 
587: $\stochmat = (m_{i,j})$ be the ($m \times n$) stoichiometric matrix implied
588: by the reaction rules $\Rea$, where $m_{i,j}$ denotes the number of molecules
589: of type $i$ produced in reaction $j$. 
590:   A set of molecules $C \subseteq \Mol$ is called \df{mass-maintaining}, 
591:   if there exists a flux vector  $\vec v \in \RR^n$ such that the three following
592:   condition apply: (1) for all  reactions $(A \to B)$ with $A \in \PowM C$ the
593:   flux $v_{(A \to B)} > 0$; (2) for all reactions  $(A \to B)$ with 
594:   $A \notin \PowM C$, $v_{(A \to B)}=0$; and 
595:   (3) for all molecules $i \in C$, 
596:   $f_i \geq 0$ with $ (f_1, \dots,  f_m) = \stochmat \vec v$.
597: }
598: $v_{(A \to B)}$ denotes the element of $\vec v$ describing
599: the flux of reaction $A \reactsto B$. 
600: For the example above, the stoichiometric matrix becomes 
601: $\stochmat = ((-1, 1)$, $(1, -1)$, $(-1, 0)$, $(0, -1))$, and we can see
602: that there is no positive flux vector $\vec v \in \RR^4$, such that
603: $\stochmat \vec v \geq \vec 0$. In fact, only the empty 
604: semi-organization $\{ \}$ is mass-maintaining.
605: In case $a$ and $b$ would not decay, $\Rea = \{ a \reactsto b, b \reactsto a
606: \}$, the set $\{a , b\}$ would be (as desired) mass-mainatining, because threre is
607: a flux vector, e.g., $\vec v = (1.0, 1.0)$, such that $\stochmat \vec v = \vec
608: 0 \geq \vec 0$ with  $\stochmat = ((-1, 1), (1, -1))$.
609: Now, closure and mass-maintenance together lead to the central definition
610: of this work:
611: 
612: \Def{organization}{
613:   A set of molecules $O \subseteq \Mol$ that is closed and
614:   mass-maintaining is called an \df{organization}.
615: }
616: 
617: %
618: An organization represents an important combination of molecular species,
619: which are likely to be observed in a reaction vessel on the long run.
620: A set of molecules that is not closed or not mass-maintaining would not
621: exists for long, because new molecules can appear or some molecules
622: would vanish, respectively.
623: The condition ``mass-maintaining'' is stronger and more difficult to
624: compute than the condition ``self-maintaining''. In fact, the former implies
625: the latter, because in a mass-maintaing set, a molecule $i$ that is used-up within
626: that set must also be produced within that set in order to achieve a 
627: non-negative production rate $f_i$. Thus we can say: 
628: \Lem{
629: Every organization is a semi-organization.
630: }%
631:  
632: 
633: 
634: \Example{Four Species}{In order to illustrate the new concepts, we consider
635: a small example where there are just four molecular species 
636: $\Mol = \{a, b,c,d \}$, which react according to the following reaction
637: rules $\Rea = \{ 
638: a + b \reactsto a + 2b, 
639: a + d \reactsto a + 2d,
640: b + c \reactsto 2c,
641: c \reactsto b,
642: b + d \reactsto c,
643: b \reactsto \emptyset, c \reactsto \emptyset, d \reactsto \emptyset 
644: \}$, where $b,c,d$ decay spontaneously and $a$ is a permanent catalyst.
645: Although the reaction system is small, its organizational structure
646: is already difficult to see when looking at the rules or
647: their graphical representation (Fig.~1).
648: In Fig.~1, all 16 possible sets of molecules are shown as a lattice.
649: There are 9 closed sets, 8 self-maintaining sets, 7 mass-maintaining sets, and
650: 7 semi-organizations, 6 of which are organizations. The organizations
651: are the only combination of molecules that can reside in a
652: reaction vessel for a long time. 
653: We can also see immediately that a reactor containing 
654: $\{b,c,d\}$ must have a transient dynamics ``down''
655: where molecular species are lost, and that a reactor containing
656: $\{ a, b, d \}$ will have a transient where a new molecular species
657: will appear, and so on... 
658: %
659: }
660: 
661: 
662: % ----------------------------------------------------------------------
663: \subsection*{Different reaction systems}
664: % ----------------------------------------------------------------------
665: 
666: 
667: Finding all organizations of a general reaction system
668: appears to be computationally difficult. 
669: One approach is to  find the
670: semi-organizations first, and then check, which of them are also mass-maintaining.
671: The property of the set of organizations and  semi-organizations 
672: depends strongly on the type of system studied. 
673: We will present here four types of systems and analyze their
674: properties with respect to the newly introduced concepts. 
675: %
676: 
677: 
678: 
679: %
680: 
681: \paragraph{\SystemD.} 
682: 
683: In a \systemd all molecules are used-up by first-order reactions
684: of the form $\{ k \} \reactsto \emptyset$ (dilution) and there
685: is no molecule used-up by any other reaction. 
686: So, each molecule $k$ decays spontaneously, 
687: or equivalently, is removed by a dilution flow.
688: Apart from this, each molecule can appear only as a catalyst (without being used-up).
689: Examples of \systemd are the replicator equation\cite{ac:SS83}, 
690: the hypercycle \cite{ac:ES77}, the more general 
691: {\it catalytic network equation} \cite{ac:SFM93}, or AlChemy \cite{ac:Fon92}. 
692: Furthermore some
693: models of genetic regulatory networks %
694: and social system \cite{ped:DKB2002} are \systemds.
695:  
696: \Lem{
697: In a \systemd, every semin-organization is an organization.
698: }%
699: 
700: In a \systemd we can easily check, whether a set $O$ is an organization by
701: just checking whether it is closed and whether each molecule in that
702: set is produced by that set.
703: Furthermore,  given a set $A$, we can always generate an organization by
704: adding all molecules produced by $A$ until $A$ is closed and than
705: removing molecules that are not produced until $A$ is self-maintaining.
706: With respect to the intersection and union of (semi-)organizations
707: the set of all (semi-) organizations of a \systemd forms an 
708: algebraic lattice (see below), which has already been noted
709: by Fontana and Buss \cite{ac:FB94arrival}.
710: 
711: \paragraph{\SystemC.} 
712: 
713: In a  \systemc   all molecules are used-up by first-order reactions
714: of the form $\{ k \} \reactsto \emptyset$ (dilution).
715: But as opposed to the previous system, we allow arbitrary additional reactions
716: in $\Rea$. 
717: This is a typical situation for
718: chemical flow reactors or bacteria that grow \cite{bio:PK2004}. %
719: In a \systemc,  semi-organizations are not necessarily organizations, 
720: as in the reversible reaction example shown before.
721: Nevertheless, both, the semi-organizations and the organizations form a
722: lattice  $\langle \allO, \unionO, \intersecO \rangle$,
723: Moreover, the union ($\unionO$) and intersection  ($\intersecO$)
724: of any two organizations is an organization
725: (see below). 
726: %
727: 
728: 
729: \paragraph{\SystemB.} 
730: In a \systemb there are two types of molecules:  persistent molecules $P$ and
731: non-persistent molecules. All non-persistent molecules 
732: $ k \in \Mol \setminus P$ are 
733: used-up (as in the two systems before) by first-order reactions
734: of the form $\{ k \} \reactsto B$ with $k \notin B$; whereas 
735: a persistent molecule $p \in P$ is not used-up by any reaction at all.
736: %
737: An example of a \systemb is Example~2, where $a$ is a persistent molecule.
738: The \systemb is the most general of the four systems where
739: the semi-organizations and organizations always form a lattice,
740: and where the generate organization operator can  properly be defined (see below). 
741: As in a \systemc, not all semi-organizations are organizations.
742: 
743: 
744: \paragraph{\SystemA.} 
745: 
746: %
747: A \systema consists of arbitrary reactions. 
748: Examples are planetary atmosphere chemistries \cite{ac:YD99}.
749: %
750: In a \systema, neither the set of  organization nor the set of  semi-organizations 
751: necessarily form a lattice.
752: %
753: Because of the lack of properties there is an actual difficulty in
754: studying those systems. The analysis gets much easier, if we are able
755: to transform the system into something else, e.g., a \systemb.
756: This could be done by introducing a small outflow for each molecule.
757: 
758: %
759: \paragraph{Common Properties of \SystemB}
760: %
761: 
762: Consistent reaction systems (including \systemc and \systemd) possess
763: some comfortable properties that allow us to present a series of
764: useful definitions and lemmas. 
765: In a \systemb, given a set of molecules $C$, 
766: we can uniquely {\em generate} a self-maintaining set, a
767: semi-organization, a mass-maintaining set, and an organization
768: in a similar way as we have generated a closed set.
769: And like for closed sets, we can define the union and intersection
770: on self-maintaining sets, semi-organizations, mass-maintaining sets,
771: and organizations, respectively.  
772: Furthermore, each, the self-maintaining sets, semi-organizations, 
773: mass-maintaining sets,
774: and organizations form a lattice together 
775: with their respective union and intersection operators.
776: This does not generalize to \systema, because
777: for a \systema we cannot uniquely generate a self-maintaining set, a
778: semi-organization, a mass-maintaining set, nor an organization
779: as in the case of a \systemb.
780: 
781: %
782: %
783: 
784: %
785: %
786: %
787: 
788: \Def{generate self maintaining set}{
789: Given a set of molecules $C \subseteq \Mol$,
790: we define $\GSM(C)$ as the biggest self-maintaining set
791: $S$ contained in $C$. We say that $C$ 
792: \df{generates the self-maintaining set} $S = \GSM(C)$. 
793: }
794: In order to calculate the self-maintaining set generated by $C$,
795: we remove those molecules 
796: that are used-up and not produced within $C$, 
797: until all molecules used-up are also produced, and thus
798: reaching a self-maintaining set.
799: The operator $\GSM$ (generate self-maintaining set) implies
800: the union $\unionSM$ and intersection $\intersecSM$ on self-maintaining sets:
801: Given two self maintaining sets $S_1$ and $S_2$,  the
802: self-maintaining sets generated by their union ($S_1 \unionCL S_2$) and intersection
803: ($S_1 \intersecCL S_2$) are defined as:
804:  $S_1 \unionSM S_2      \equiv  \GSM(S_1 \cup S_2)$, and  
805:  $S_1 \intersecSM S_2   \equiv  \GSM(S_1 \cap S_2)$, respectively.
806: And as already mentioned, in a \systemb, 
807: $\langle \allSM, \unionSM, \intersecSM \rangle$ forms a lattice,
808: where $\allSM$ is the set of all self-maintaining sets of an algebraic chemistry.
809: Finally note that, if $S$ is self-maintaining, its closure
810: $\GCL(S)$ is self-maintaining, too (in \systembs).
811: 
812: There are many ways in which we can generate a semi-organization from a
813: set. We will present here the simplest one, which implicitly
814: assumes that molecules are produced quickly and vanish slowly.
815: This assumption leads to the largest possible semi-organization
816: generated by a set:
817: 
818: \Def{generate semi-organization}{
819: Given a set of molecules $C \subseteq \Mol$,
820: we define $\GSO(C)$ as $\GSM(\GCL(S))$.
821: We say that $C$ 
822: \df{generates the semi-organization} $O = \GSO(C)$. 
823: }
824: 
825: In the same way as before, the generate semi-organization operator  $\GSO(C)$ implies 
826: the union  $\unionSO$ and intersection $\intersecSO$ on semi-organizations,
827: namely 
828: $O_1 \unionSO O_2      \equiv  \GSO(O_1 \cup O_2)$, and  
829: $O_1 \intersecSO O_2   \equiv  \GSO(O_1 \cap O_2)$, respectively,
830: which implies the lattice of semi-organizations.
831: %
832: 
833: 
834: \Def{generate mass-maintaining set}{
835: Given a set of molecules $C \subseteq \Mol$,
836: we define $\GMM(C)$ as the biggest mass-maintaining set
837: $S$ contained in $C$. We say that $C$ 
838: \df{generates the mass-maintaining set} $S = \GMM(C)$. 
839: }
840: For \systembs,  $\GMM(C)$ is always defined, 
841: because the union ($\cup$)
842: of two mass-maintaining sets is mass-maintaining; and further,
843: every set is either mass-maintaining, or it contains a unique biggest
844: mass-maintaining set. Thus from every set we can generate a mass-maintaining
845: set.
846: Note that mass-maintaining sets are also self-maintaining,
847: $\GMM(\GSM(S)) \equiv \GMM(S)$, which is a useful property, 
848: because  $\GSM(S)$  is easier to compute.
849: As usual, the union $\unionMM$ and intersection $\intersecMM$ of mass-maintaining sets
850: $S_1, S_2$ are defined as
851: $S_1 \unionMM S_2      \equiv  \GMM(S_1 \cup S_2)$,
852: $S_1 \intersecMM S_2  \equiv  \GMM(S_1 \cap S_2)$, respectively.
853: Thus also the set of all mass-maintaining sets  $\allMM$ forms a lattice
854: $\langle \allMM, \unionMM, \intersecMM \rangle$.
855: If $S$ is mass-maintaining, its closure $\GCL(S)$ is mass-maintaining, too
856: (again, not valid for \systemas).
857: 
858: Finally, in \systembs, we can also generate uniquely an organization,
859: (here, again, the largest organization that can be generated from a set) 
860: according to the following definition:
861: 
862: \Def{generate  organization}{
863: Given a set of molecules $C \subseteq \Mol$,
864: we define $\GO(C)$ as $\GMM(\GCL(C))$.
865: We say that $C$ 
866: \df{generates the organization} $O = \GO(C)$. 
867: }
868: 
869: Equivalently $\GO(C) = \GMM(\GSM(\GCL(C)))$, which allows to
870: compute the organization generated by a set more easily in three steps.
871: %
872: Following the same scheme as before,
873: the union $\unionO$ and intersection $\intersecO$ of two organizations
874: $U$ and $V$ is defined as the
875: organization generated by their set-union  and set-intersection:
876: $U \unionO V      \equiv  \GO(U \cup V)$, 
877: $U \intersecO V   \equiv  \GO(U \cap V)$,
878: respectively. 
879: Thus, for \systemb, also the set of all organizations  $\allO$ forms a lattice
880: $\langle \allO, \unionO, \intersecO \rangle$. This important fact should be
881: emphasized by the following lemma:
882: 
883: \Lem{
884:   Given an algebraic chemistry $\langle \Mol, \Rea
885:   \rangle$ of a \systemb  and all its organizations $\allO$,
886:   then  $\langle \allO, \unionO, \intersecO \rangle$ is a
887:   lattice.
888: }%
889: 
890: Knowing that the semi-organizations and organizations form a lattice, and that we
891: can uniquely generate an organization for every set, 
892: is a useful information. In order to find the
893: whole set of organizations, it is impractical just to check all the possible sets
894: of molecules.
895: Instead,  we can start by computing the lattice of semi-organizations,
896: and then test only those sets for mass-maintenance. 
897: Furthermore, if the semi-organizations form a lattice, we can start with
898: small sets of molecules and generate their semi-organizations,
899: while the $\unionSO$ operator can lead us to the more
900: complex semi-organizations.
901: %
902: %
903: 
904: 
905: As a summary, from a practical point of view, we 
906: calculate first the set of semi-organizations.
907: If our system is a \systemd, we automatically obtain the lattice
908: of organizations. Otherwise we have to check for each semi-organization whether
909: it is mass-maintaining or not. If we have a \systemb, then
910: we are assured to obtain a lattice, where there is a unique smallest and 
911: largest organization, and where we can easily obtain the intersection
912: and union of two organizations from the graphical representation of
913: the lattice (see examples).
914: For a \systema, the set of organizations does not 
915: necessarily form a lattice. 
916: Nevertheless, this set of organizations represent
917: the \df{organizational structure} of the reaction network, which can
918: be visualized and which can provide a new view
919: on the dynamics of the system by mapping  the
920: movement of the system in state space to 
921: a movement in the set of organizations, as will be
922: shown in the following section.
923: 
924: 
925: 
926:   
927: 
928: 
929: %
930: %
931: 
932: %
933: 
934: 
935: 
936: % ----------------------------------------------------------------------
937: 
938: \subsection*{Dynamic Analysis}
939: 
940: The static theory  
941: deals with  molecules $\Mol$ and their reaction rules $\Rea$, but not
942: with the evolution of the system in time.
943: To add dynamics to the theory, we have
944: to formalize the dynamics of a system.
945: In a very general approach, 
946: the \df{dynamics} is given  by a \df{state space} $X$ and 
947: a formal definition (mathematical or algorithmical) that
948: describes all possible  movements in $X$.
949: Given an initial state $\vec x_0 \in X$,   
950: the formal definition describes how the state changes over time.
951: For simplicity, we assume a deterministic dynamical process, which
952: can be formalized  by a phase flow 
953: $(X, (T_t)_{t \in \RR})$ where 
954:    $(T_t)_{t \in \RR}$ is a one-parametric group
955:    of transformations from $X$. $T_t(\vec x_0)$ denotes the state at time $t$ of
956:    a system that has been in state $\vec x_0$ at $t=0$.
957: %
958: %
959: 
960: 
961: 
962: %
963: 
964: %
965: \paragraph{Connecting to the Static Theory}
966: %
967: \label{sec:conn-stat-theory}
968: 
969: A state $\vec x \in \X$ represents the state of a reaction vessel 
970: that contains molecules from $\Mol$.
971: In the static part of the theory we consider just the set
972: of molecular species present in the reaction vessel, but not
973: their concentrations, spatial distributions, velocities, and so on.
974: %
975: %
976: %
977: 
978: Now, given the state $\vec x$ of the reaction vessel, we need a function
979: that maps uniquely this state to the set of molecules present.
980: Vice versa, given a set of molecules $A \subseteq \Mol$,
981: we need to know, which states from $\X$ correspond to this set of molecules.
982: For this reason we introduce a mapping $\fmol$ called \df{abstraction},
983: from $\X$ to $\Mol$, which maps a state of the system to the set of molecules
984: that are present in the system being in that state.
985: We require this mapping to be monotonically increasing on the number of molecules. 
986: In other words, if we have two states
987: $\vec x_1$ and $\vec x_2$, and the concentration of the molecule $m_1$ is higher in
988: $\vec x_2$ than in $\vec x_1$; and if $m_1 \in \fmol(\vec x_1)$ then 
989: $m_1 \in \fmol(\vec x_2)$. The exact mapping can %
990: be defined precisely later, depending on the state space, on the dynamics,
991: and on the actual application.
992: 
993: %
994: 
995: The concept of \df{instance} is the opposite of the concept of abstraction.
996: While $\fmol(\vec x)$
997: denotes the molecules represented by the state $\vec x$, an
998: instance $\vec x$ of a set $A$ is a state where exactly the molecules
999: from $A$ are present according to the function $\fmol$.
1000: 
1001: \Def{instance of $A$}{
1002: We say that a state $\vec x \in X$ is an instance of $A \subseteq \Mol$,
1003: iff $\fmol(\vec x) = A$
1004: }
1005: 
1006: In particular, we can define an instance of an organization $O$ 
1007: (if $\fmol(x) = O$) and an instance of a generator of $O$ 
1008: (if $\GO(\fmol(x)) = O$).
1009: Loosely speaking we can
1010: say that $\vec x$ \df{generates organization} $O$.
1011: Note that a state $\vec x$ of a  \systemb, \systemc, and \systemd 
1012: is {\em always} an instance of a generator 
1013: of one and only one organization $O$. 
1014: This leads to the important observation that
1015: a lattice of organizations partitions the state space $\X$,
1016: where a partition $\X_O$ implied by organization $O$ is
1017: defined as the set of all instance of all generators of $O$:
1018: $\X_O = \{\vec x \in \X |  \GO(\fmol(\vec x)) = O \}$.
1019: Note that as the system state evolves over time,
1020: the organization $G(\fmol(\vec x(t)))$ 
1021: generated by $\vec x(t)$ might change (see below, Fig.~2~and~3).
1022: 
1023: 
1024: %
1025:  
1026: 
1027: % ----------------------------------------------------------------------
1028: \paragraph{Fixed Points are Instances of Organizations.}
1029: % ----------------------------------------------------------------------
1030: 
1031: Now we describe a theorem that relates
1032: fixed points to organizations, and by doing so, underlining
1033: the relevancy of organizations.
1034: We will show that, 
1035: given an ODE of a form that is commonly used to describe 
1036: the dynamics of reaction systems, 
1037: every fixed point of this ODE is an instance of an organization.
1038: We therefore assume in this section
1039: that $\vec x$ is a concentration vector
1040: $\vec x = (x_1, x_2, \dots, x_{|\Mol|})$, $\X = \RR^{|\Mol|}, x_i \geq 0$ 
1041: where $x_i$ denotes 
1042: the concentration of molecular species $i$ in the reaction vessel, 
1043: and $\Mol$ is finite.
1044: %
1045: %
1046: %
1047: The dynamics is given by an ODE of the form 
1048: $\dot {\vec x} = \stochmat \vec v(\vec x)$ where 
1049: $\stochmat$ is the stoichiometric matrix implied by the 
1050: algebraic chemistry $\langle \Mol, \Rea \rangle$ (reaction rules).
1051: $\vec v(\vec x) = (v_1(\vec x), \dots, v_n(\vec x)) \in \RR^{|\Rea|}$ is a flux vector
1052: depending on the current concentration $\vec x$, where $|\Rea|$ denotes
1053: the number of reaction rules.
1054: A flux $v_j(\vec x)$ describes the rate of a particular reaction $j$.
1055: For the function $v_j$ we require only that $v_j(\vec x)$ is positive,
1056: if and only if the molecules on the left hand side of the reaction $j$ are
1057: present in the state $\vec x$, and otherwise it must be zero. 
1058: Often it is also assumed that  $v_j(\vec x)$ increases monotonously,
1059: but this is not required here.  %
1060:                                 %
1061: Given the dynamical system as $\dot {\vec x} = \stochmat \vec v(\vec x)$, 
1062: we can define the abstraction of a state
1063: $\vec x$ formally  by using
1064: a (small) threshold $\thresh \geq 0$ 
1065: such that all fixed points have
1066: positive coordinates greater than $\thresh$.
1067: 
1068: \Def{abstraction}{
1069:   \label{def:abstraction}
1070:   Given a dynamical system $\vec {\dot x} = f(\vec x)$ and
1071:   let $\vec x$ be a state in $\X$, then the abstraction $\fmol(\vec x)$ is
1072:   defined by
1073:   \begin{equation}
1074:     \label{eq:fmol}
1075:     \fmol ( \vec x ) = \{ i | x_i > \thresh, i \in \Mol \}, \quad 
1076:     \fmol: \X \to \Pow{\Mol}, \quad \thresh \geq 0     
1077:   \end{equation}
1078:   where $x_i$ is the concentration of molecular species $i$ in state $\vec x$,
1079:   and $\thresh$ is a threshold chosen such that it is smaller than any positive
1080:   coordinate of any fixed point of  $\vec {\dot x} = f(\vec x), x_i \geq 0$.
1081: }
1082: 
1083: Setting $\thresh=0$ is a safe choice, because in this case $\fmol$
1084:  always meets the definition above. But for practical reasons,
1085: it makes often sense to apply a positive threshold greater zero,
1086: e.g., when we take into consideration that the number of molecules
1087: in a reaction vessel is finite. 
1088: %
1089: 
1090: \begin{theorem}
1091: {\bf Hypothesis:} Let us consider a \systema whose
1092: reaction network is given by the algebraic chemistry  $\langle \Mol, \Rea
1093: \rangle$ and whose dynamics is 
1094: given by $\dot {\vec x} = \stochmat \vec v(\vec x) = f(\vec x)$ as defined before.
1095: Let $\vec x' \in \X$ be a fixed point, that is, $f(\vec x')= \vec 0$, 
1096: and let us consider a mapping $\fmol$ as given by Def.~\ref{def:abstraction},
1097: which assigns a set of molecules to each state $\vec x$.
1098: {\bf Thesis:}
1099: $\fmol(\vec x')$ is an organization. 
1100: %
1101: \end{theorem} 
1102: 
1103: 
1104:   {\bf Proof:}
1105:    We need to prove that $\fmol(\vec x')$ is closed and
1106:   mass-maintaining: 
1107:   (a) Closure: Let us assume that $\fmol(\vec x')$ is not closed, then
1108:   there exist a molecule $k$ such that $k \notin \fmol(\vec x')$ and
1109:   $k$ is generated by molecules contained in $\fmol(\vec x')$.
1110:   Since, $\vec x'$ is a fixed point, 
1111:     $f(\vec x')= \stochmat \vec v(\vec x') = 0$.
1112:   Now we decompose the stoichiometric matrix $\stochmat$ into
1113:   two matrices $\stochmat^+$ and $\stochmat^-$ separating  all
1114:   positive from the negative coefficients, respectively, such that
1115:   $\stochmat = \stochmat^+ + \stochmat^-$ and  
1116:   $\stochmat^+ \vec v(\vec x') \geq \vec 0$, and
1117:   $\stochmat^- \vec v(\vec x') \leq \vec 0$ (note that 
1118:   $\vec v(\vec x')$ is always non-negative by definition).
1119:   Let $\dot x^+_k$ and  $\dot x^-_k$ be the $k$-th row of 
1120:   $\stochmat^+ \vec v'(\vec x)$ and
1121:   $\stochmat^- \vec v'(\vec x)$, respectively, which represent
1122:   the inflow (production) and outflow (destruction) of molecules
1123:   of type $k$. Note that  
1124:   $\dot x^+_k + \dot x^-_k = \dot x'_k = 0$ (fixed point condition).
1125:   Since we assumed that $k$ is produced by molecules from $\fmol(\vec x')$,
1126:   $\dot x^+_k$ must be positive, $\dot x^+_k > 0$ and thus
1127:   $\dot x^-_k < 0$.
1128:   But this leads to a contradiction. In order to show this
1129:   we have to differentiate the following two cases:
1130:   (i) Assume that $x'_k=0$, then $\dot x^-_k$ must be zero, 
1131:   too (by definition; and intuitively, because a molecule not present cannot
1132:    vanish).
1133:   (ii) Assume that $0 < x'_k \leq \thresh$ ($x'_k> \thresh$ needs not to 
1134:   be considered, because in that case 
1135:   $k$ would be contained in $\fmol(\vec x)$.).
1136:   Because we assumed that $x'_k$ is a coordinate of a fixed point, 
1137:   $\thresh$ must be smaller than $x'_k$, which is again a contradiction.
1138: 
1139:   (b) Mass-maintaining:  We have to show that $\fmol(\vec x')$ 
1140:   is mass-maintaining.
1141:   Since $\vec x'$ is a fixed point $\stochmat \vec v(\vec x') = \vec 0$, 
1142:   which fulfills condition (3) of the definition of mass-maintaining (Def.~6).
1143:   From the requirements for the flux vector $\vec v$,
1144:   it follows directly that $v_{(A \reactsto B)}(\vec x')>0$ for all 
1145:   $A \in \PowM{\fmol(x')}$, which fulfills condition (1) of Def.~6. 
1146:   Following the same contradictory argument as before in (ii), 
1147:   $x'_k$ must be zero for $k \notin  \fmol(\vec x)$, and therefore
1148:   $v_{(A \reactsto B)}(\vec x')=0$ for $A \notin \PowM{\fmol(\vec x)}$,
1149:   which fulfills the remaining condition (2) of mass-maintaining.
1150:   q.e.d.
1151: %
1152: 
1153: 
1154: From this theorem it follows immediately that a fixed point is an
1155: instance of a closed set, a self-maintaining set, and of a semi-organization.
1156: Let us finally mention that
1157: even if each fixed point is an instance of an organization,
1158: an organization does not necessarily possess a fixed point.
1159: Further note that given an attractor $A \subseteq \X$, 
1160: there exists an organization $O$
1161: such that all points of $A$  are instances of
1162: a generator of $O$.
1163: In fact, it might be natural to suppose 
1164: that all points of an attractor are actually instances of $O$,
1165:  yet it is not clear if this is true for all systems or just for some.
1166: 
1167: 
1168: %
1169: 
1170: 
1171: 
1172: % ----------------------------------------------------------------------
1173: \subsubsection*{Movement from Organization to Organization}
1174: % ----------------------------------------------------------------------
1175: 
1176: %
1177: % ----------------------------------------------------------------------
1178: \paragraph{ODEs and Movement in the Set of Organizations}
1179: % ----------------------------------------------------------------------
1180: 
1181:  Not all system can be studied using ODEs. In particular a discrete system 
1182: is usually  only {\it approximated} by an ODE.
1183: In a discrete dynamical system, the molecular species that are present in the
1184: reaction vessel can change in time, e.g,  
1185: as the last molecule of a certain type
1186: vanishes. In an ODE instead, this does
1187: not generally happen, where molecules can {\it tend to} zero as
1188:   time {\it tends to} infinity. 
1189: So, even if in reality a molecule disappears, 
1190: in an ODE model it might still be present in a tiny quantity. 
1191: The fact that every molecule ends up being present in (at least) a
1192:   tiny quantity, generally precludes us to notice that the system is actually
1193:   moving from a state where some molecules are present, 
1194: to another state where a
1195:   different set of molecules is present. Yet this is what happens in reality,
1196:   and in this respect, an ODE is a poor approximation of reality. 
1197: A common approach to overcome this problem is to introduce a concentration 
1198: threshold $\thresh$, below which a molecular species is considered not to
1199: be present. We use this threshold in order to define the abstraction 
1200: $\fmol$, which just returns the set of molecules present in a certain state.
1201: Additionally, we might use the threshold to manipulate the numerical
1202: integration of an ODE by setting a concentration to zero, when it falls
1203: below the threshold. In this case, a constructive perturbation has 
1204: to be greater than this threshold. 
1205: 
1206:  
1207:  
1208: 
1209:  
1210: %
1211: 
1212: % ------------------------------------------------------------
1213: \paragraph{Downward Movement}
1214: % ------------------------------------------------------------
1215: \label{sec:movement-down-down}
1216: 
1217: Not all organizations are stable. %
1218: %
1219: The fact that there exits a flux vector, such
1220: that no molecule of that organization vanishes,
1221: does not imply that this flux vector can be realized 
1222: when taking dynamics into account.
1223: As a result a molecular species can disappear.
1224: Each molecular species that disappears simplifies the
1225: system. Some molecules can be generated back. But eventually the system can
1226: move from a state that generates organization $O_1$ into a  state that generates
1227: organization $O_2$, with $O_2$ always below $O_1$ ($O_2 \subset O_1$). 
1228: We call this spontaneous movement a \df{downward movement}. 
1229: 
1230: 
1231: 
1232: 
1233: 
1234: % ------------------------------------------------------------
1235: \paragraph{Upward Movement}
1236: % ------------------------------------------------------------
1237: \label{sec:movement-up-up}
1238: 
1239: 
1240: Moving up to an organization above requires that
1241: a new molecular species appears in the system. 
1242: This new molecular species cannot be produced by a reaction among
1243: present molecules (condition of closure).  
1244: Thus moving to an organization above is more complicated then the
1245: movement down and requires a couple of specifications
1246: that describe how new molecular species enter the system.
1247: Here we assume that new molecular species appear by some sort
1248: of random perturbations or purposeful interference. 
1249: We assume that a small quantity of molecules of that 
1250: new molecular species (or a set of molecular species) suddenly appears.
1251: Often, in practice, the perturbation (appearance of new molecular species) has a
1252: much  slower time scale than the internal dynamics (e.g., chemical reaction kinetics)
1253: of the system. 
1254: 
1255: \Example{Upward and downward Movement}{
1256: \label{exa:movement-up}
1257: Assume a system with two molecular species $\Mol = \{ a, b \}$ and
1258: the reactions $\Rea = \{ a  \reactsto  2a, b \reactsto 2b, a + b \reactsto a, 
1259: a \reactsto, b \reactsto \}$. All combinations of molecules are organizations,
1260: thus there are four organizations (Fig.~2).
1261: Assume further that the dynamics is governed by
1262: the ODE $\dot x_1 =  x_1 - x_1^2, \dot x_2 = x_2 - x_2^2 - x_1 x_2$ where
1263: $x_1$ and $x_2$ denote the concentration of species $a$ and $b$, respectively.
1264: We map a state to a set of molecules by
1265: using a small, positive threshold $\thresh = 0.1 > 0$ (Def.~\ref{def:abstraction}).
1266: Now, assume that the system is in state $\vec x_0 = (0,1)$ thus in
1267: organization $\{ b \}$.
1268: If a small quantity of $a$ appears (constructive perturbation), 
1269: the amount of $a$ will grow and $b$ will tend to zero.
1270: The system will move upward to organization $\{ a, b \}$ in a transient
1271: phase, while finally moving down and converging to a fixed point in
1272: organization $\{ a \}$. This movement can now be visualized in the lattice
1273: of organizations as shown in Fig.~2.
1274: }
1275: 
1276: %
1277: 
1278: 
1279: % ----------------------------------------------------------------------
1280: \paragraph{Visualizing Possible Movements in the Set of Organizations}
1281: % ----------------------------------------------------------------------
1282: 
1283: %
1284: 
1285: 
1286: In order to display potential movements in the lattice or set of
1287: organizations, we can draw links between
1288: organizations. As exemplified in Fig.~2 and Fig.~3, these links can
1289: indicate possible downward movements (down-link, blue) or upward movements (up-link,
1290: red). A neutral link (black line) denotes that neither the system can 
1291: move spontaneously down, nor can a constructive perturbation move the system
1292: up. Whether the latter is true depends on the definition of ``constructive
1293: perturbation'' applied. For the example of Fig.~3 we defined a constructive
1294: perturbation as inserting a small quantity of {\it one} new molecular species.
1295: 
1296: The dynamics in between organizations is more complex than this intuitive
1297: presentation might suggest, for example in some cases it is possible to 
1298: move from one organization $O_1$ to an organization $O_2$, with $O_2$ above
1299: (or below) $O_1$
1300: without passing through the organizations in between $O_1$ and $O_2$.
1301: In Fig.~3 this is the case for an upward movement from organization
1302: $\{ a \}$ to organization $\{a,b,c\}$ caused by a constructive perturbation
1303: where a small quantity of $c$ has been inserted. 
1304: 
1305: %
1306: 
1307: % ----------------------------------------------------------------------
1308: \paragraph{Organizations in Real Systems}
1309: % ----------------------------------------------------------------------
1310: 
1311: In preliminary studies \cite{ped:SDZB2000} we have shown
1312: that artificial chemical reaction networks that are based on a 
1313: structure-to-function mapping (e.g., ref.~\cite{ac:FB94arrival,ac:Ban93})
1314: possess a more complex lattice
1315: of organization than networks created randomly. 
1316: From this observation we can already expect that natural
1317: networks possess non-trivial organization structures.
1318: Our investigation of planetary photo-chemistries \cite{ac:YD99} and
1319: bacterial metabolism (which will be published elsewhere), 
1320: revealed lattices of organizations that vanish when the networks are
1321: randomized, indicating a non-trivial structure.
1322: Here, in order to give an impression, we show a lattice
1323: of organizations obtained from a model of the central sugar
1324: metabolism of {\it E.coli} by Puchalka and Kierzek \cite{bio:PK2004}.
1325: The model consists of 92 species and 197 reactions, including gene expression,
1326: signal transduction, transport, and enzymatic activities.
1327: Figure~4 shows the lattice of organizations resulting from the 
1328: original model, ignoring inhibiting interactions.
1329: The smallest organization, $O_1$, contains 76 molecules including
1330: the glucose metabolism and all input molecules.
1331: The input molecules, chosen according to ref.~\cite{bio:PK2004},
1332: include the external food set (Glcex, Glyex, Lacex) and
1333: all promoters (see suppl. material).
1334: Two other organizations, $O_3$ and $O_4$, contain the Lactose and Glycerol
1335: metabolism, respectively. Their union results in the largest
1336: organization $O_5$ that contains all molecules.
1337: In summary, we can conclude, that the organizations found are
1338: biological meaningful, indicating a promising potential for
1339: future applications of this theory.
1340:   
1341:  
1342: %
1343: 
1344: % ----------------------------------------------------------------------
1345: \paragraph{Conclusion}
1346: % ----------------------------------------------------------------------
1347: We presented a general way to define organizations in reaction systems, and
1348: proved a theorem that showed that fixed points are instances of those
1349: organizations. We also investigated the relative structure of the
1350: organizations, opening up the door to visualize the dynamical movement of the
1351: system through organizations. Much work still needs to be done, and we believe
1352: that this line of research is just in its infancy. Reaction systems
1353: with inhibiting interactions need to be investigated, as well as the structure
1354: inside organizations (i.e. the relation between attractive states and
1355: organizations).  In fact the whole issue of movement among sets of molecules
1356: could only be introduced here. The concepts of attractive and stable
1357: organizations, which play
1358: a fundamental role in dynamical systems, has still to be formally defined.
1359: Yet we believe that our work represents a
1360: step forward towards a formal study of constructive dynamical system.
1361: \vspace{0.2cm}
1362: 
1363: \noindent
1364: {\footnotesize
1365: We thank Florian Centler for providing the {\it E. Coli.} analysis. 
1366: %
1367: This work was supported by Federal Ministry of Education and Research (BMBF)
1368: Grant~0312704A to Friedrich-Schiller-University Jena.
1369: }
1370: 
1371: 
1372: %
1373: 
1374: \begin{thebibliography}{10}
1375: 
1376: \bibitem{ac:FB94arrival}
1377: Fontana\aidx{Fontana, Walter}, W \& Buss\aidx{Buss, Leo W.}, L.~W.
1378: \newblock (1994) {\em Bull. Math. Biol.} {\bf 56}, 1--64.
1379: 
1380: \bibitem{soc:Luh84}
1381: Luhmann, N.
1382: \newblock (1984) {\em Soziale Systeme}.
1383: \newblock (Suhrkamp, Frankfurt a.M.).
1384: 
1385: \bibitem{ped:SD2002}
1386: Speroni~di Fenizio, P \& Dittrich, P.
1387: \newblock (2002) {\em J. of Three Dimensional Images} {\bf 16},
1388:   160--163.
1389: 
1390: \bibitem{ac:SS83}
1391: Schuster\aidx{Schuster, Peter}, P \& Sigmund\aidx{Sigmund, Karl}, K.
1392: \newblock (1983) {\em J. Theor. Biol.} {\bf 100}, 533--8.
1393: 
1394: \bibitem{ac:ES77}
1395: Eigen\aidx{Eigen, Manfred}, M \& Schuster\aidx{Schuster, Peter}, P.
1396: \newblock (1977) {\em Naturwissenschaften} {\bf 64}, 541--565.
1397: 
1398: \bibitem{ac:SFM93}
1399: Stadler\aidx{Stadler, Peter F.}, P.~F, Fontana\aidx{Fontana, Walter}, W,  \&
1400:   Miller, J.~H.
1401: \newblock (1993) {\em Physica D} {\bf 63}, 378--392.
1402: 
1403: \bibitem{ac:Fon92}
1404: Fontana\aidx{Fontana, Walter}, W.
1405: \newblock (1992) in {\em Artificial Life II} eds.{} Langton, C.~G, Taylor, C,
1406:   Farmer, J.~D,  \& Rasmussen, S.
1407: \newblock (Addison-Wesley, Redwood City, CA), pp. 159--210.
1408: 
1409: \bibitem{ped:DKB2002}
1410: Dittrich, P, Kron, T,  \& Banzhaf, W.
1411: \newblock (2003) {\em JASSS}
1412:   {\bf 6}.
1413: 
1414: \bibitem{bio:PK2004}
1415: Puchalka, J \& Kierzek, A.
1416: \newblock (2004) {\em Biophys. J.} {\bf 86}, 1357--1372.
1417: 
1418: \bibitem{ac:YD99}
1419: Yung, Y.~L \& DeMore, W.~B.
1420: \newblock (1999) {\em Photochemistry of Planetary Athmospheres}.
1421: \newblock (Oxford University Press, New York).
1422: 
1423: \bibitem{ped:SDZB2000}
1424: Speroni~di Fenizio, P, Dittrich\aidx{Dittrich, Peter}, P, Ziegler, J,  \&
1425:   Banzhaf\aidx{Banzhaf, Wolfgang}, W.
1426: \newblock (2000) {\em Towards a Theory of Organizations}.
1427: \newblock (GWAL 4, Bayreuth, 5.-7. April, 2000).
1428: 
1429: \bibitem{ac:Ban93}
1430: Banzhaf\aidx{Banzhaf, Wolfgang}, W.
1431: \newblock (1993) {\em Biol. Cybern.} {\bf 69}, 269--281.
1432: 
1433: %
1434: 
1435: \end{thebibliography}
1436: 
1437: 
1438: %
1439: 
1440: 
1441: 
1442: % ----------------------------------------------------------------------
1443: \newpage
1444: {\bf FIGURE CAPTION}
1445: 
1446: \vspace{2cm}
1447: Figure~1: Example with four species. Reaction network (a ``2'' means
1448: that two molecules are produced) and graphical
1449: representation of the lattice of all
1450: sets. The vertical position is determined by the set size.
1451: The solid lines depict the lattice of organizations.
1452: 
1453: \vspace{2cm}
1454: 
1455: Figure~2: Example of an up-movement caused by a constructive perturbation,
1456: followed by a down-movement. (a) reaction network, 
1457: (b) concentration vs. time plot of a trajectory,
1458: (c) lattice of organizations including trajectory.  
1459: 
1460: \vspace{2cm}
1461: 
1462: Figure~3: Lattice of organizations of Example 1, including up-links and down-links.
1463: Furthermore a trajectory is shown starting in organization $\{a,b,c,d\}$
1464: moving down to organization $\{a , b\}$.
1465: 
1466: \vspace{2cm}
1467: 
1468: Figure~4: Lattice of organizations of a model of the central sugar
1469: metabolism of {\it E.coli.} \cite{bio:PK2004}. 
1470: In an organization, only names of new molecular species are printed that
1471: are not present in an organization below.
1472: 
1473: 
1474: % ----------------------------------------------------------------------
1475: \newpage
1476: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1477: %%%   FIGURE 1
1478: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1479:  \vspace{2cm}
1480:   \noindent
1481:   \begin{tabular}{p{\linewidth}}
1482: %  
1483:   \centerline{\epsfig{figure=fig1.eps,width=10cm}} \\
1484:   \vspace{3cm}
1485: 
1486:  {Figure~1: Example with four species. Reaction network (a ``2'' means
1487:     that two molecules are produced) and graphical
1488:     representation of the lattice of all
1489:     sets. The vertical position is determined by the set size.
1490:      The solid lines depict the lattice of organizations.
1491:   }
1492:   \end{tabular}
1493: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1494: \newpage
1495: 
1496: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1497: %%%  FIGURE 2
1498: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1499: \vspace{2cm}
1500:  \noindent
1501:  \centerline{\epsfig{figure=fig2.eps,width=10cm}} \\
1502:  \vspace{3cm}
1503: 
1504:   \noindent
1505: {Figure~2: Example of an up-movement caused by a constructive perturbation,
1506:  followed by a down-movement. (a) reaction network, 
1507:    (b) concentration vs. time plot of a trajectory,
1508:  (c) lattice of organizations including trajectory.  
1509:  }
1510: 
1511: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1512: \newpage
1513: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1514: %%%  FIGURE 3
1515: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1516: \vspace{2cm}
1517:  \hspace{0.3cm}
1518:  \noindent
1519:  \centerline{\epsfig{figure=fig3.eps,width=10cm}}\\
1520: % 
1521: \vspace{3cm}
1522: 
1523:   \noindent
1524:  {Figure 3: Lattice of organizations of Example 1, including up-links and down-links.
1525:  Furthermore a trajectory is shown starting in organization $\{a,b,c,d\}$
1526:  moving down to organization $\{a , b\}$.}
1527:  %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1528: 
1529: \newpage
1530: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1531: %%% FIGURE 4
1532: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1533:  \vspace{2cm}
1534:  \hspace{0.3cm}
1535:  \noindent
1536: % 
1537:  \centerline{\epsfig{figure=fig4.eps,width=10cm}}\\
1538:  \vspace{3cm}
1539: 
1540:   \noindent
1541:  {Figure 4: Lattice of organizations of a model of the central sugar
1542:  metabolism of {\it E.coli.} \cite{bio:PK2004}. 
1543:  In an organization, only names of new molecular species are printed that
1544:  are not present in an organization below.}
1545:  
1546: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1547: \end{document}
1548: 
1549: 
1550: 
1551: 
1552: 
1553: 
1554: