Untangling complex syste.., p.98

Untangling Complex Systems, page 98

 

Untangling Complex Systems
Select Voice:
Brian (uk)
Emma (uk)  
Amy (uk)
Eric (us)
Ivy (us)
Joey (us)
Salli (us)  
Justin (us)
Jennifer (us)  
Kimberly (us)  
Kendra (us)
Russell (au)
Nicole (au)



Larger Font   Reset Font Size   Smaller Font  



  The best evaporation-rate is 5. When it is 0, ants have three broad paths to the three

  sources of food to follow. When the evaporation-rate is 20, ants can trust just in shreds of

  tracks towards the food, and most of them waste time by roving randomly throughout the

  entire space available.

  13.5. By playing with the number of strategies and the memory size, different dynamics are

  observable. A list of results is reported in Table 13.4.

  According to the results presented in Table 13.4, a good solution is when both memory

  size and number of strategies are fixed to ten. There are overcrowded outcomes, but the num-

  ber of people does not exceed too much above sixty, and when there is no overcrowding, the

  number of people is close to sixty. The bar is never “desert,” and this is good for the owner.

  13.6. The first principle of thermodynamics allows us to write dU = dq + dw. Introducing the second law of thermodynamics, we have dU = TdS − PdV. Since the ideal gases are at constant temperature, their internal energies do not change. Therefore, TdS = PdV. Introducing

  500

  Untangling Complex Systems

  TABLE 13.3

  Times Taken by the Ants to Eat the Food as a Function of the Evaporation Rate of Pheromone

  Evaporation Rate

  Time (ticks)

  Average Time

  Picture

  0

  600

  790

  729

  624

  775

  703.6

  5

  470

  561

  493

  504

  447

  495

  20

  1031

  1538

  1625

  1816

  1598

  1521.6

  TABLE 13.4

  Possible Dynamics of Attendance to the El Farol Bar as Functions of the Memory-size and

  the Number of Strategies

  Memory-size

  N° strategies

  Dynamics

  Attendance

  1

  1

  Fixed point

  >70

  1

  10

  Oscillations

  ] ~85, ~25[

  1

  20

  Oscillations

  ] ~90, ~10[

  5

  1

  Fixed point or oscillations with small amplitudes

  ≥70

  5

  10

  Chaotic

  ] ~70, ~45[

  5

  20

  Chaotic

  ] ~90, ~20[

  10

  1

  Chaotic (small amplitudes)

  ≥65

  10

  10

  Chaotic

  ] ~67, ~50[

  10

  20

  Chaotic

  ] ~70, ~40[

  the equation of state for ideal gases, we obtain that dS = ( nR V ) dV . For the mixing process (I), the total variation of the thermodynamic entropy is:

  

  

  

  

  ∆

  Vbf

  V

  S = n

  gf

  b Rln 

   + ngRln

   = ( nb + ng) Rln 2

   Vbi 

  V

   gi 

  where the subscripts b and g refer to the black and gray gases, respectively.

  How to Untangle Complex Systems?

  501

  If we consider the information entropy, the uncertainty (H) that we have about a gas con-

  tained in a volume Vi is Hi = log 2 Vi. Therefore, in the expansion from Vi to Vf , the increment of uncertainty and hence the variation of the information entropy is ∆ H = log 2 V

  (

  )

  f Vi . It is

  evident the strict relation between the two expressions:

  (∆ H )( n

  b + nr ) Rln 2 = ∆ S

  Analogously, for the process of free expansion (∆ H)( nb) Rln 2 = ∆ S.

  Appendix A: Numerical Solutions

  of Differential Equations

  No human investigation can become real science without going through mathematical people.

  Leonardo da Vinci (1452–1519 AD)

  In addressing the dynamics of a Complex System, the typical approach consists of the following

  steps. First, the key variables of the phenomenon are identified, together with the nature of their

  interactions. Second, differential equations describing the time evolution of the system are for-

  mulated. If the system is spatially homogeneous, the differential equations are ordinary (ODEs);

  otherwise they are partial (PDEs). Third, the differential equations are solved either analytically,

  graphically, or numerically (most non-linear differential equations cannot be solved analytically).

  Before the advent of the computer, numerical methods were impractical because they required an

  exhausting work of hand-calculation. With the implementation of always faster computers, all that

  has changed. Finally, after determining the fixed points, the stability properties of the steady state’s

  solutions are probed.

  Differential equations such as [A.1] are rules describing the time evolution of X as a function of X.

  dX

  = f ( X ) [A.1]

  dt

  If only first order derivatives are involved, equation [A.1] is a first order differential equation. If

  the function f ( X ) is continuous and smooth, i.e . , differentiable in every point, then the solution to equation [A.1] exists and is unique.

  In this Appendix A, a few methods of solving numerically differential equations, (such as [A.1]), are presented.

  A.1 EULER’S METHOD

  The most basic explicit method for the numerical integration of ordinary differential equations

  is Euler’s Method. It is named after Leonhard Euler, a Swiss mathematician, and physicist of the

  eighteenth century. Euler’s Method converts the rule [A.1] into values for X. Let us suppose that

  initially, i.e . , for t = t ,

  . The derivative is equal to dX

  ( ) =

  0 X = X 0

  f ( X ). Euler’s method assumes

  dt t

  0

  0

  that the derivative dX

  ( )

  dt

  is constant over a certain time interval Δ t. Therefore, the next value for X

  t 0

  will be given by

  dX 

  X ( t 0 + t

  ∆ = t 1) = X 1 = X 0 + f ( X 0) t

  ∆ = X 0 + 

  

  t

  ∆ [A.2]

   dt  t 0

  503

  504

  Appendix A

  X

  X′1

  X 1

  X 0

  t

  t 0

  t 1

  t′1

  FIGURE A.1 Description of the function X( t) by the Euler’s method. The solutions shown as black circles are obtained by selecting a longer ∆ t than that chosen for the solutions represented as black squares.

  Actually, the derivative f ( X ) is constantly changing, but the assumption that it is constant over the time interval ∆ t becomes less and less wrong, as ∆ t becomes closer and closer to zero, i.e.,

  infinitesimal. At time t , in location, the derivative is now equal to f ( X . We step forward to 1

  X 1

  1)

  X 2 = X ( t 1 + t

  ∆ ) = X 1 + f ( X 1) t

  ∆ , and then we iterate the procedure:

  1. We calculate the derivative in Xn

  2. We use the rate of change to determine the next value Xn+1:

  Xn+1 = X ( tn + t

  ∆ ) = Xn + f ( Xn) t

  ∆ [A.3]

  until we have a satisfactory number of solutions.

  A visualization of the possible solutions achievable by Euler’s Method is shown in Figure A.1 where a non-linear function X( t) is plotted as a continuous black curve. The squared and the circle dots

  represent the solutions of the differential equation [A.1] obtained by applying Euler’s Method with

  two different time intervals. The black circles are obtained with a longer ∆ t . It is evident that the

  shorter the ∆ t , the more accurate the description of the rule.

  A.2 ALTERNATIVE METHODS

  An improvement on the Euler’s Method was proposed by the German mathematicians Carl Runge

  and Martin Kutta at the beginning of the twentieth century. The Runge-Kutta Method samples the

  rate of change at several points along the interval Δ t and averages them. In particular, the fourth-

  order Runge-Kutta Method, implemented in MATLAB through the command ode45, requires cal-

  culating the following four amounts

  k 1 = f ( Xn) t

  ∆

  

  1 

  k 2 = f  Xn + k 1  t

  ∆

  

  2 

  [A.4]

  

  1

  

  k 3 = f  Xn + k 2  t

  ∆

  

  2

  

  k 4 = f ( Xn + k 3)∆ t

  Appendix A

  505

  X

  X 0

  t

  t 0

  FIGURE A.2 The gray squares represent the numerical solution of the function depicted by a continuous

  black curve.1 They describe the curve much better than the numerical results obtained by the Euler’s Method (see Figure A.1).

  to find Xn+1 from Xn. In fact,

  1

  X 1

  ( 1 2 2 2 3 4)

  n+ = X n +

  k + k + k + k [A.5]

  6

  This method gives much more accurate results than Euler’s Method as it can be easily inferred by

  looking at Figure A.2.

  Some differential equations may be more critical, and they may require an adaptive step size.

  In fact, there are methods where Δ t is automatically adjusted on the fly: Δ t is made small where

  the derivative changes rapidly and conversely, it is made large when the derivative changes slowly.

  A.3 HINTS FOR FURTHER READINGS

  More detailed information about differential equations and their numerical solutions can be found

  in many textbooks. See, for example, the book Differential Equations by Blanchard et al. (2006)

  and the book by Morton and Mayers (2005) for the numerical solution of partial differential

  equations.

  1 The syntax for the application of fourth-order Runge-Kutta method in MATLAB is the following one: function dy = heating(t,y)

  dy = zeros(1,1);

  dy = 0.2*(20−y);

  Then, you must paste in the Command Window the following string:

  [t,y]

  = ode45(‘heating’, [0 12], 5);

  Appendix B: The Maximum

  Entropy Method

  The real voyage of discovery consists, not in seeking new landscapes, but in having new eyes.

  Marcel Proust (1871–1922 AD)

  The Galilean scientific method is grounded on the collection of experimental data, their analy-

  sis, and interpretation. In the time-resolved spectroscopies, the time-evolving signal of a sample

  recorded upon pulse-excitation is a convolution of the pulse shape of the excitation (IRF) with the

  response of the molecular system ( N( t)) (see Figure B.1).

  After deconvolution, the experimental data should be expressed as a weighted infinite sum of

  exponentials:

  ∞

  N t

  α τ e− t

  ( )

  /

  =

  ( ) τ d

  ∫

  τ [B.1]

  0

  Our computational task is to determine the spectrum of α( τ) values having inevitably noisy data

  available. Mathematically, α( τ) is the inverse Laplace transform of the measured signal decon-

  volved from the IRF. Although deconvolution is well-conditioned and relatively stable, inverting

  the Laplace transform is ill-conditioned for a noisy signal. This statement means that small errors

  in the measurement of R( t) can lead to huge errors in the reconstruction of α( τ). In other words, the ill-conditioning leads to a multiplicity of allowable solutions. The ensemble of all possible shapes of

  α( τ) is displayed as a rectangle labeled as set A in Figure B.2. Among all the elements of set A, only a subset of them agrees with the noisy data set shown on the left of Figure B.2. It is the set labeled as B, confined inside the dotted boundaries. Some of these spectra of set B are devoid of physical

  meaning (e.g., corresponding to negative concentrations). Therefore, set B is partitioned into two

  subsets: subset B” contains the unphysical spectra α( τ) to be rejected, whereas subset B’ contains the feasible spectra α( τ). All the elements of subset B’ fit the data and are physically allowable. Among them, we need to choose just one element. This operation is possible by maximizing some function

  of the spectra, h[ α( τ)]. The function h[ α( τ)] should be chosen to introduce the fewest artifacts. For this purpose, we use the Maximum Entropy Method (MEM).

  The roots of the Maximum Entropy Method reside in the probability theory. Statisticians are

  divided into two schools of thought regarding the interpretation of probability: the “classical,”

  “ frequentist,” or “objective” school instituted by Fisher in the 1940s, and the “subjective” or

  “Bayesian” school, named after Thomas Bayes who proposed it around 1750 (Brochon 1994). The

  “objective” school of thought regards the probability of an event as an objective property of that

  event. It is possible to measure that probability by determining the frequency of positive events in

  an infinite number of trials. In calculating a probability distribution, the “objectivists” believe that

  they are making predictions which are in principle verifiable, just as they were dealing with deter-

  ministic events of classical mechanics. The question that an “objectivist” raises when he judges a

  probability distribution p( x) is: “Does it correctly describe the observable fluctuations of x?”

  The “subjective” school of thought regards the probability as personal degree of belief: the prob-

  ability of an event is merely a formal expression of our expectation that the event will or did occur,

  507

  508

  Appendix B

  Object of enquiry

  Instrument

  Experimental result

  IRF

  R( t)

  a.u.

  0

  Time

  R( t) = ( IRF ) ⊗ ( N( t))

  Instrumental

  True

  response

  value

  function

  FIGURE B.1 Schematic representation of the steps of a time-resolved measurement. The sample is the object

  of inquiry; the instrument allows to collect a time-resolved signal R( t) that is a convolution of the Instrumental Response Function (IRF) and the molecular kinetics N( t). The analysis of R( t) is carried out by the operator with the help of some computational tools.

  Set A

  Set B

  IRF

  Inverse

  R( t)

  Laplace

  transform

  Subset B’

  a.u.

  Subset B”

  0

  Time

  FIGURE B.2 Partition of all possible spectra α( τ) contained in set A, into two ensembles: set B containing all the spectra agreeing with the experimental data and set not B ( B) containing the remaining elements. Set

  B is further partitioned into two subsets: B’ containing the spectra with physical meaning and B” containing

  all the spectra devoid of physical significance.

  based on whatever information is available. The test of a good subjective probability distribution

  p( x) is: “Does it correctly represent our state of knowledge as to the value of x?” (Jaynes 1957). The Bayesian view of probability is the source of MEM.

  The key aim of scientific data analysis is to determine the probability density of h[ α( τ)] given some data D, i.e., the conditional probability Pr( h D). According to the Bayes’ theorem, the probability of D and h, Pr( h, D) is given by

  Pr( ,

  h D) = Pr( )

  h Pr D

  ( h) = Pr( D)Pr h

  ( D ) [B.2]

  Appendix B

  509

  The Bayes’ theorem can be rearranged in the following form:

  Pr h

  ( ) Pr( D h)

  Pr( h D) =

  [B.3]

  Pr D

  ( )

  The term Pr( D h) is the “likelihood:” it is the probability of obtaining some data D if the true spectrum h is known. In general, the likelihood expresses the instrumental response, its resolution and

  uncertainty. In the case of Gaussian noise statistics, the likelihood is

  −1 2

  χ

  e 2

  Pr( D h) =

  [B.4]

  Zl

  with

  1 2 1

  T

  2

  χ =

  D −

  −

  (

  R )

  h (σ )( D −

  )

  Rh [B.5]

  2

  2

  where:

  Rh is the calculated data from the spectrum h;

  σ−2 is the covariance matrix for the data;

  Z is the normalization factor.

  l

  The term Pr( D) of equation [B.3] is the “plausibility” of the data based on our prior knowledge of

  the system. If the prior knowledge of the system maintains constant, this term is like a normaliza-

  tion factor ( Z ). Otherwise, it can be used to discriminate between models in which the data would

  D

  have a different probability of occurring, for example, between a set of discrete or continuous dis-

 

Add Fast Bookmark
Load Fast Bookmark
Turn Navi On
Turn Navi On
Turn Navi On
Scroll Up
Turn Navi On
Scroll
Turn Navi On
183