1: \documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb,floatfix]{revtex4}
2: %\documentclass[twocolumn,floatfix,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
3: % Some other (several out of many) possibilities
4: %\documentclass[preprint,aps]{revtex4}
5: %\documentclass[preprint,aps,draft]{revtex4}
6: %\documentclass[prb]{revtex4}% Physical Review B
7:
8: \usepackage{graphicx}% Include figure files
9: \usepackage{dcolumn}% Align table columns on decimal point
10: \usepackage{bm}% bold math
11:
12: \begin{document}
13:
14: \preprint{APS/123-QED}
15:
16: \title{Bose-Einstein Condensation in Competitive Processes}
17:
18: \author{Hideaki Shimazaki}
19: %\email{hshimaz2@jhmi.edu}
20: %\altaffiliation[Also at ]{}%Lines break automatically or can be forced
21: %with \\
22: \author{Ernst Niebur}%
23: %\email{Second.Author@institution.edu}
24: \affiliation{%
25: Department of Neuroscience, Zanvyl Krieger Mind/Brain Institute, School of
26: Medicine, Johns Hopkins University, Baltimore, Maryland 21218
27: }%
28:
29: \date{\today}% It is always \today, today,
30:
31: \begin{abstract}
32: We introduce an irreversible discrete multiplicative process that
33: undergoes Bose-Einstein condensation as a generic model of
34: competition. New players with different abilities successively join
35: the game and compete for limited resources. A player's future gain is
36: proportional to its ability and its current gain. The theory provides
37: three principles for this type of competition: competitive exclusion,
38: punctuated equilibria, and a critical condition for the distribution
39: of the players' abilities necessary for the dominance and the
40: evolution. We apply this theory to genetics, ecology and
41: economy.
42: \end{abstract}
43:
44: \pacs{03.75.Nt, 87.23.-n, 87.23.Kg, 89.65.-s}% PACS, the Physics and Astronomy
45: % Classification Scheme.
46: %\keywords{Suggested keywords}%Use showkeys class option if keyword
47: %display desired
48: %
49: \maketitle
50: %
51: Competition occurs when two or more players such as organisms,
52: individuals or companies strive for common but limited resources. It
53: plays a significant role in biological and social activities, and is
54: the basic idea underlying evolution. Recently, Bianconi \&
55: Barab\'{a}si demonstrated that their competitive complex network model
56: behaves analogously to a Bose gas \cite{Bianconi01}. A few nodes
57: recruited many connections and acted like hubs, while most others had
58: sparse links. This phenomenon was described in the mathematical
59: framework of Bose-Einstein condensation (BEC). Here, we analyze a
60: discrete multiplicative process that also exhibits BEC. This generic
61: competition model is defined by the following
62: conditions at each time step: 1. A player's gain is proportional to
63: its ability and to its gain at the previous time step. 2. A new player
64: joins the competition. 3. Players compete for a fixed total amount of
65: resources. The constant addition of players is the basis of the
66: model's open and irreversible characteristics and makes the process
67: suitable for the analysis of natural and social competition. We show
68: that the cumulative gain of each player is related to the Bose
69: distribution, except for that of two players. One is the pioneer. The
70: other is a high-ability player that dominates most resources under a
71: certain statistical condition for the distribution of the players'
72: ability. We investigated the implication of BEC in this mathematical game for
73: different competition models. Applied to the competition among clonal
74: strains of evolving asexual bacteria, BEC in an irreversible process
75: provides theoretical grounds for the principle of competitive
76: exclusion \cite{Gause64} and the hypothesis of punctuated equilibria
77: \cite{Elena96}. Furthermore, the existence of a transition point
78: elucidates the condition necessary for evolutionary development,
79: and explains the observation that a high mutation rate is not
80: necessarily advantageous for adaptive evolution
81: \cite{Sniegowski97,deVisser99}. Finally, we investigate possible
82: applications to ecological and economic activities.
83:
84: %\subsection{\label{sec:level4}Generic Route to Bose statistics}
85: Before we present our competition model, we introduce a novel, general
86: framework for how a function $\varphi(\epsilon)$ formally maps into a Bose
87: system. Define a function $\varphi(\epsilon)$ that satisfies the
88: following conditions for an arbitrary $C^2$ density function $g\left(
89: \epsilon \right) (\geq 0)$ defined on $\epsilon \in
90: [0,\epsilon_{\max}]$,
91: \begin{eqnarray}
92: \int {d\epsilon \, g\left( \epsilon \right)} &=& 1, \label{33} \\
93: \int {d\epsilon \, g\left( \epsilon \right)\varphi \left( \epsilon \right)}
94: &=& m + y_0, \label{34} \\
95: \int {d\epsilon \, g\left( \epsilon \right)e^{ - \beta \epsilon } \varphi
96: \left( \epsilon \right)} &=& M . \label{35}
97: \end{eqnarray}
98: The r.h.s. of these equations and $\beta$ in (\ref{35}) are all
99: positive constants. Then $\varphi \left( \epsilon \right) $ is given by
100: \begin{equation}\label{36}
101: \varphi \left( \epsilon \right) = \frac{y_0}{{1 - e^{ - \beta
102: \epsilon - \alpha } }},
103: \end{equation}
104: where $e^{-\alpha} = {m \mathord{\left/ {\vphantom {m M}} \right. \kern-\nulldelimiterspace} M}$. For, dividing (\ref{34}) by $y_0$, then subtracting (\ref{33}) yields
105: \begin{equation}\label{24}
106: \int {d\epsilon \, g\left( \epsilon \right)\frac{{\varphi \left( \epsilon
107: \right) - y_0 }}{{y_0 }}} = N,
108: \end{equation}
109: where $N = {m \mathord{\left/ {\vphantom {m {y_0 }}} \right. \kern-\nulldelimiterspace} {y_0 }}$. From (\ref{35})$\times {N \mathord{\left/ {\vphantom {N M}} \right. \kern-\nulldelimiterspace} M} - $(\ref{24}) and the fundamental lemma of the calculus of variation, we obtain (\ref{36}). Hence, the so-called occupation number in (\ref{24}),
110: \begin{equation}\label{37}
111: n\left( \epsilon \right) = \frac{{\varphi \left( \epsilon \right)
112: - y_0 }}{{y_0 }} = \frac{1}{{e^{\beta \epsilon + \alpha } - 1}} ,
113: \end{equation}
114: becomes the Bose distribution, or, equivalently, the system $\varphi \left( \epsilon \right)$ maps into a Bose system. From (\ref{36}), such a system may be obtained from the sum of a geometric progression with ratio $e^{- \beta \epsilon - \alpha } < 1$. This motivates analysis of the following competition process.
115:
116: %\subsection{\label{sec:level1} Competition Model}
117: Consider a simple multiplicative process
118: \begin{equation}\label{1}
119: y_i(t+1) = a_i b(t)y_i(t),
120: \end{equation}
121: where $y_i(t)$ is the gain of the $i$th player at time $t$ and $a_i$ is its ability, a positive and time-independent random variable chosen from a distribution $\rho \left( a \right)$. From (\ref{1}), a player's gain is proportional to its ability and the gain of the previous time step. Furthermore, a new player joins the race at every time step with initial value $y_i(i) = y_0 $, which means we consider $t + 1 $ simultaneous equations at time $t $. The only exception is the first player (pioneer), whose initial value is $y_0 + m$. Finally, the population races for a fixed amount of resources, modelled by the normalization $b(t)$ in (\ref{1}) which is defined by $b(t) = {m \mathord{\left/ {\vphantom {m {M_t }}} \right. \kern-\nulldelimiterspace} {M_t }}$, where $M_t$ is given by
122: \begin{equation}\label{2}
123: M_t = \sum\limits_{j = 0}^t {a_j y_j \left( t \right)} .
124: \end{equation}
125: This assures that the total gain distributed among $t$ players at time
126: $t$ is $m $ for $t > 0$. As a new player joins the race with $y_0 $,
127: the total gain of the population is fixed to $m + y_0 $ every time
128: step. We now consider the time evolution of players except for the pioneer. The gain of the $i$th player at time $t$ is given by $y_i(t) =\Omega _i(t) y_0$, where
129: \begin{equation}\label{3}
130: \Omega _i(t) = a_i^{t-i} \left( {\prod\limits_{t' = i}^{t - 1}
131: {b\left( t' \right)} } \right) .
132: \end{equation}
133: Without loss of generality, we can set $a_i \leq 1$ (discussed later),
134: and we let $a_i = e^{ - \beta \epsilon_i } $, thus introducing a
135: formal parametrization of the ability distribution by the inverse
136: temperature $\beta (=1/T) $. The random variable $\epsilon \in
137: [0,\epsilon_{\max}]$ is chosen from $g\left( \epsilon \right)$,
138: a state density function which defines the system.
139: From (\ref{3}), we have
140: \begin{equation}\label{38}
141: \ln \Omega _i = \left( {t - i} \right)\left\{ { - \beta \epsilon_i -
142: \left\langle {\ln b} \right\rangle } \right\} ,
143: \end{equation}
144: using $\ln \prod\nolimits_{t' = i}^{t - 1} {b\left( t' \right)} =
145: \sum\nolimits_{t' = i}^{t - 1} {\ln b\left( t' \right)} \sim \left( t -i
146: \right) \left\langle {\ln b} \right\rangle $. By assuming stationarity of
147: $b(t)$, discussed below, we let its time average be
148: \begin{equation}\label{5}
149: \left\langle {\ln b} \right\rangle = - \alpha .
150: \end{equation}
151: The normalization factor $\alpha$ keeps the total gain constant throughout the dynamical process. Its role is analogous to that of the chemical potential of a quantum gas introduced for the conservation of particle number. From (\ref{38}) and (\ref{5}), we obtain $y_i(t) = e^{( { - \beta \epsilon_i - \alpha }) \left(t-i \right)} y_0 $. Given $e^{-\beta \epsilon_i - \alpha} < 1$, the cumulative gain $\varphi _t \left( \epsilon_i \right) = \sum\nolimits_{t' = i}^t {y_i \left( {t'} \right)} $, converges to $\varphi \left( \epsilon_i \right)$ in (\ref{36}). Hence, the normalization of $\varphi _t \left( \epsilon_i \right)$ by the initial value $y_0 $,
152: \begin{equation}\label{10}
153: n_t \left( \epsilon_i \right) = \frac{{\varphi _t \left( \epsilon_i
154: \right) - y_0 }}{{y_0 }} ,
155: \end{equation}
156: takes the form of the Bose distribution in the thermodynamic limit $\left( {t \to \infty } \right)$. Indeed, in this limit, $M_t $ obeys
157: \begin{equation}\label{11}
158: \mathop {\lim }\limits_{t \to \infty } M_t = \int {d\epsilon \, g\left(
159: \epsilon \right)\sum\limits_{t' = 0}^\infty {e^{ - \beta \epsilon }
160: y\left( {t'} \right)} } .
161: \end{equation}
162: Since the stationarity of $b\left( t \right)$ yields $\mathop {\lim }\limits_{t \to \infty } \ln b\left( t \right) = \left\langle {\ln b} \right\rangle $, we have $\mathop {\lim }\limits_{t \to \infty } {m \mathord{\left/ \right. \kern-\nulldelimiterspace} {M_t }} = e^{ - \alpha } $ from (\ref{5}). By substituting this into (\ref{11}), we obtain an equality,
163: \begin{equation}\label{13}
164: \int {d\epsilon \, g\left( \epsilon \right)\frac{1}{{e^{\beta \epsilon +
165: \alpha } - 1}}} = N ,
166: \end{equation}
167: where $N = {m \mathord{\left/ {\vphantom {m {y_0 }}}
168: \right. \kern-\nulldelimiterspace} {y_0 }}$. The dynamical system
169: (\ref{1}) thus formally maps into a Bose system in the thermodynamic limit.
170:
171: Comparing this system to a quantum gas, adding the gain $m$ at each
172: step corresponds to adding $m$ Bose particles. This gain is then
173: distributed among players with different abilities; these abilities
174: correspond to energy levels. Different from a quantum gas, a new
175: energy level is introduced at every step (with initial occupation of
176: $y_0$), the gains (particles) do not move to different energy levels
177: once they are allocated, and they are not quantized. Despite these
178: differences, the system preserves a basic feature of a
179: Bose gas: gains tend to accumulate where other gains are already
180: present. Thus, condensation analogous to BEC is expected at low
181: temperatures.
182:
183: %\subsection{\label{sec:level2}BEC}
184: To study this prediction, we simulated the multiplicative process
185: (\ref{1}), adopting a standard density function,
186: \begin{equation}\label{39}
187: g\left( \epsilon \right) = C_\sigma \epsilon ^{\sigma - 1} ,
188: \end{equation}
189: where $C_\sigma = {\sigma \mathord{\left/
190: \right. \kern-\nulldelimiterspace} {\epsilon _{\max }^\sigma }}$
191: $\left( \sigma > 1\right)$. We found two distinct phases for the
192: distribution of normalized cumulative gain (\ref{10}). At high
193: temperature, $n_t \left( \epsilon_i \right)$ obeys the Bose
194: distribution (FIG.~\ref{FIG1}-a). Our assumption of stationarity of $b
195: \left( t \right)$ is confirmed numerically since $n_t \left(
196: \epsilon_i \right)$ agrees well with the Bose distribution with
197: $\alpha$ being provided on the assumption of stationarity. At low
198: temperatures, the coordinated distribution breaks down: one player
199: with low energy (but not necessarily the lowest energy because of the
200: normalization) dominates a large fraction of the cumulative
201: resources. Note that the stationarity of $b(t)$ no longer holds in
202: this domain. This Bose-Einstein condensate exists only below the
203: critical point $\alpha = 0$. For, if $\alpha $ is negative, $\varphi
204: \left( \epsilon_i \right)$ of the player with $\epsilon_i < - {\alpha
205: \mathord{\left/ \right. \kern-\nulldelimiterspace} \beta } $ does not
206: converge. Its normalized cumulative gain cannot be incorporated in the
207: integral in (\ref{13}); rather it has to be added as an extra
208: term. The critical temperature for this transition can be predicted in
209: the usual way. Change of variables $y = \beta \epsilon $ and use of
210: $\alpha = 0$ in (\ref{13}) yields
211: \begin{equation}\label{15}
212: N = \frac{\sigma }{{\left( {\beta_c \epsilon _{\max } }
213: \right)^\sigma }}\int_0^{\beta_c \epsilon _{\max } } {\frac{{y^{\sigma -
214: 1} }}{{e^{y } - 1}} \,dy} .
215: \end{equation}
216: By approximating the upper limit of integral in (\ref{15}) by infinity, the critical temperature $T_c({ = 1/\beta _c })$ is given by
217: \begin{equation}\label{17}
218: T_c^{\left( 1 \right)} \sim \epsilon _{\max } \left\{ {N^{ - 1} \Gamma
219: \left( {\sigma + 1} \right)\zeta \left( \sigma \right)} \right\}^{ -
220: {\textstyle{ \frac{1}{\sigma} }}} ,
221: \end{equation}
222: for $N \ll 1$. When $N \gg 1$, this approximation is not valid because of the high critical temperature. For $N \gg 1$, one uses $e^y \simeq 1 + y$ in (\ref{15}) and obtains
223: \begin{equation}\label{32}
224: T_c^{\left( 2 \right)} \sim \epsilon _{\max } \left( {1 - \sigma ^{ - 1} }
225: \right)N.
226: \end{equation}
227: The transition can be easily seen in the occupation by the most
228: capable player, defined by ${\varphi_t \left( {\epsilon _{\min } }
229: \right)} \mathord{\left/ {\sum\nolimits_j {\varphi_t \left( {\epsilon
230: _j } \right)} } \right.} $, where $\epsilon_{\min} = \min \left\{
231: \epsilon_j \right\} $. As the temperature decreases below $T_c $, its
232: occupation dramatically increases, supporting the prediction
233: (FIG.~\ref{FIG1}-c). The time evolution of each player's gain also
234: shows characteristic differences above and below the predicted
235: $T_c$. At $T > T_c$, the gain of each player monotonically
236: decreases. Below $T_c$, time evolution is not monotonic
237: (FIG.~\ref{FIG3}-b and area in FIG.~\ref{FIG2}-c). This is
238: particulary interesting because this BEC occurs in an irreversible
239: process. In contrast to BEC in a quantum gas in equilibrium, the
240: condensation state dose not imply the dominance of resources by a
241: single player at the lowest energy throughout the whole competition
242: process. Instead, there are replacements of dominant player upon the
243: entrance of a player with lower energy level.
244:
245: %%%%%%%%%%%%%%%%%%%%%%%%%%%%FIGURE%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
246: \begin{figure}[t]
247: \includegraphics[width=8.6cm]{FIG1}
248: \caption{\label{FIG1} (a) Normalized cumulative gain (\ref{10}) for $m = 5$ and $T<T_c$ (left), $T \approx T_c$ (center) and $T>T_c$ (right). Solid line is the Bose distribution with $\alpha $ from (\ref{5}). The plus sign indicates a pioneer. Plots of the last ten entrants were excluded. (b) Numerical calculation of $|\alpha|$ with $\alpha$ from (\ref{5}), averaged over 500~trials. Symbols +, o, x indicate $m = 5$, $10$ and $20$. Change in the sign of $\alpha$ indicates the transition. Triangles on the abscissa are the transition temperatures from (\ref{32}); $T_c \approx 3.33 $, $6.67 $ and $13.3 $. Shown in inset is the dependence of $T_c$ on exponent $\sigma $ for $m = 20$ (dots) and the analytical solution from (\ref{32}), solid line. (c) Cumulative occupation by the most capable player was calculated by $ \varphi_t \left( \epsilon _{\min } \right) \left/ \sum\nolimits_j \varphi_t \left( {\epsilon _j } \right) \right. $, where $\epsilon_{\min} = \min \left\{ \epsilon_j \right\} $ and $t = 200$. Symbols as in (c). Dashed line, occupation by pioneer for $m = 20$. In all simulations, $y_0 = 1 $, $\sigma = 3$ and $\epsilon _{\max } = 1$.}
249: \end{figure}
250: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
251: %%%%%%%%%%%%%%%%%%%%%%%%%%%%FIGURE%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
252: \begin{figure}[b]
253: \includegraphics[width=8.6cm]{FIG2}
254: \caption{\label{FIG2} (a) Schematic diagram of propagation of \textit{E. coli}. Symbols, o:~wild type (WT), others: mutants. (b) State density (\ref{39}) (dashed line) with $\sigma = 3$ and $\epsilon_{\max}=1$ used in simulation and corresponding ability distribution (solid line) $\rho \left( a \right) = g\left( \epsilon \right)\left| {{\textstyle{{d\epsilon } \over {da}}}} \right| $ for $T = 1/3, 1$ and 3 from left to right. Note $a \in [e^{ - \beta \epsilon _{\max } } ,1] $. (c) Time evolution at $T = 1$ ($m=200$, $l=2$, $y_0=1$). As $T_c \approx 66.6$, the process is in BEC. Graded areas indicates population of each strains. WT abilities were set $\epsilon = 0.5$, which is slightly lower than $\left< \epsilon \right> \approx 0.76$ for $\sigma = 3$ since WT is likely to fit to environment better than chance. (d) Weighted mean ability shows punctuated equilibria at $T=0.1$ (bottom) and $T=1$ (center). At $T=100>T_c$ (top), stationarity (no evolution) is observed.}
255: \end{figure}
256: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
257: %%%%%%%%%%%%%%%%%%%%%%%%%%%%FIGURE%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
258: \begin{figure}[t]
259: \includegraphics[width=8.6cm]{FIG3}
260: \caption{\label{FIG3} (a) Market share of all nine different razors
261: from The Gillette Company introduced over the last 3.5 decades
262: \cite{Tellis01}. (b) Simulated market share from (\ref{1}), using
263: $y_0 = 1, m = 9$. Rather than assuming a specific state density, we
264: assigned $a = 0.025, 0.04$, $0.043$ ... \textit{etc.} for successive
265: products, with each newly introduced product having higher ability
266: than the previous one. The specific values were selected to best fit
267: the empirical data. In years without a product introduction by
268: Gillette, the simulation introduced ``products'' with randomly drawn lower
269: abilities ($a \ll 0.025$) that are not competitive and disappear
270: (traces adjacent to abscissa). Share of the $j$th product at time
271: $t$ was defined by ${y_j(t)} \left/ (m + y_0) \right. \times 45$ as
272: 45\% is the mean total share in (a). We considered the Carbon razor
273: as the pioneer because it was the dominant product at the beginning of
274: the period for which we have data. The introduction year of ``Prem
275: Double Edge'' was not available, thus we assigned 1963 arbitrarily
276: to fit the plot. }
277: \end{figure}
278: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
279:
280: %\subsection{\label{sec:level4}Applications}
281: We verified the existence of a BEC analogue in a discrete
282: multiplicative process, as was shown in a continuous model
283: \cite{Bianconi01}. To see its applicability to real-world competition,
284: we mention that (\ref{1}) is a natural extension of Malthusian
285: population increase: population, when unchecked, increases in a
286: geometrical ratio, that is $y_i \left( {t + 1} \right) = \tilde a_i
287: y\left( t \right)$, where $\tilde a_i \in \left[ {\tilde a_{\min }
288: ,\tilde a_{\max } } \right]$ and $\tilde a_{\min } > 1$. We
289: introduced normalization to model limited resources, obtaining
290: (\ref{1}) with $a_i$ replaced by $\tilde a_i$. Note that we can
291: substitute $\tilde a_i = \tilde a_{\max} a_i$ (i.e. $a_i \leq 1$)
292: without loss of generality because $\tilde a_{\max}$ is common in the
293: nominator and the denominator in (\ref{1}). This enables us to apply
294: (\ref{1}) to population dynamics and even economic activities.
295:
296: \textit{Application to Genetics:} We consider competition of clonal
297: strains of asexual \textit{Escherichia coli} serially propagated on
298: glucose-limited medium. For this aim, we slightly generalize model
299: (\ref{1}) such that we allow $l \left( \geq 1 \right)$ new players
300: to join the competition with initial value $y_0$
301: at each step. From a similar argument, we obtain (\ref{13}), where $N
302: = m \left/ y_0 l \right.$. Assigning $y_0 = 1$, we have $l$ mutants
303: (with usually different abilities) at each time step
304: \cite{Footnote1}. We chose this procedure because it is unlikely to
305: have $y_0 > 1$ mutants of the same ability. The original $m + y_0 l$
306: wild types propagate $m$ wild types and produce $y_0 l$ mutants of
307: novel abilities. Then the mixed population of $m + y_0 l$ is
308: normalized to $m$, and generates $y_0 l$ mutants (FIG.~\ref{FIG2}-a,
309: c). Eventually all wild types are swept away and replaced by
310: mutants. This is much faster in BEC than at high temperatures (see
311: reduction of pioneer occupation in BEC in FIG.~\ref{FIG1}-c). Also at
312: low temperatures inferior strains are quickly expelled by a superior
313: strain, which is analogous to the competitive exclusion principle
314: \cite{Gause64,Gerrish98}. Subsequent transient replacement of the
315: dominant mutants supports the hypothesis of punctuated equilibria
316: \cite{Elena96}. Weighted mean ability ($M_t$ in (\ref{2}) if $l =1$)
317: shows step-like evolution at low temperatures (FIG.~\ref{FIG2}-d), but
318: this effect diminishes with increasing temperature. Now, new mutants
319: all have similar abilities, competition is crowded and severe, and the
320: sudden upsurges of the weighted mean ability of the population are
321: smeared out. Finally, at sufficiently high temperatures, no mutant
322: becomes an absolute winner; evolution ceases when the population lacks
323: prominent players. Note that the stationarity of weighted mean
324: ability confirms the stationarity of $b(t)$, which in turn allows the
325: argument below (\ref{5}).
326:
327: A striking statement of the theory is the prediction of a singular
328: point on the emergence of evolution. Most mutations are likely to be
329: deleterious \cite{Gerrish98}, which was simulated by the use of an
330: increasing state density (\ref{39}) so that inferior mutants are
331: likely to occur (FIG.~\ref{FIG2}-c). In this case, the critical
332: temperature is predicted by (\ref{32}). Through its dependence on $N$,
333: $T_c$ is found to be dependent on the mutation rate ${{y_0 l}
334: \mathord{\left/ \right. \kern-\nulldelimiterspace} {\left( {m + y_0
335: l} \right) = \left( {N + 1} \right)^{ - 1} }} $. Therefore, at a given
336: temperature, decreasing mutation rate generates evolutionary
337: development. A sufficiently low mutation rate not only prevents
338: deleterious offspring but is necessary for adaptive evolution
339: itself.
340:
341: Although introducing the notion of temperature and the state density
342: of a quantum gas
343: was instructive, our model does not require these
344: constructs. Our conclusions hold as long as a sharply decreasing
345: ability distribution $\rho_\lambda(a)$ exists on $a \in [a_{\min},1]$ where $a_{\min}>0$, with $\lambda$ a parameter, such that the integral (\ref{13}),
346: \begin{equation}\label{41}
347: I\left( {\lambda ,\alpha } \right) = \int_{a_{\min}}^1 {da \, \rho _\lambda \left( a
348: \right)\frac{1}{{a^{ - 1} e^\alpha - 1}}} = N ,
349: \end{equation}
350: is finite. Since $I \left( {\lambda ,\alpha } \right) \leq I\left(
351: {\lambda ,0} \right) \equiv N_c \left( \lambda \right) $, we have a
352: finite $N_c$ and we can determine the corresponding critical mutation
353: rate for a given $\lambda$. Since $N_c$ is finite, the process is in
354: condensation for $N > N_c $, while evolution ceases for $N < N_c$. In
355: those cases in which the critical mutation rate is in the
356: physiological range, this argument explains ``clonal interference''
357: \cite{Gerrish98} which is an empirically observed constraint on the
358: role of mutator alleles that rises to high frequency in association
359: with beneficial mutations \cite{Sniegowski97,deVisser99}. The critical
360: point, however, only determines the occurrence of adaptive evolution
361: at $t \rightarrow \infty$, and a low mutation rate slows the rising of
362: dominant mutants. This implies the existence of an optimal mutation
363: rate for a transiently changing environment.
364:
365: \textit{Application to Ecology and Economy:} In ecology, the
366: Lotka-Volterra equation predicts that coexistence
367: between strongly competing species is impossible, a result confirmed
368: in laboratory experiments on \textit{Paramecium} \cite{Gause64}
369: though applicability in complex ecosystems is uncertain \cite{Colinvaux86}.
370: These results established the \textit{principle of competitive
371: exclusion}; or ``one species, one niche.'' Our process models interspecific competition with completely
372: overlapping niches, and BEC is a natural extension of the competitive
373: exclusion principle in an irreversible process. This principle may
374: also apply to economic activities if a product or a company satisfies
375: the ecological conditions of the exclusion principle
376: \cite{Hardin60}. For instance, products sold for a specialized market
377: may compete within their niche in the marketplace. Figure \ref{FIG3}-a
378: shows the market share of different razors successively introduced by
379: The Gillette Company, illustrating a much-studied case of new products
380: eroding the market share of their own predecessors
381: \cite{Tellis01}. Aspects of the dynamics of product replacement may
382: thus be captured by our competitive process (Fig.~\ref{FIG3}-b).
383:
384: We thank HG Schuster for very helpful suggestions and the NSF for support through a CAREER award to EN.
385: \bibliography{bec}% Produces the bibliography via BibTeX.
386: \end{document}
387:
388: ------=_NextPart_000_003B_01C2FA04.CA1C2310--
389: