Particle decay: Difference between revisions

From formulasearchengine
Jump to navigation Jump to search
en>Maxviwe
m Reverted 2 edits by 217.21.224.165 (talk) identified as vandalism to last revision by Kulmalukko. (TW)
 
en>Darko.veberic
Line 1: Line 1:
The name of the author is Nestor. She functions as a monetary officer and she will not change it anytime quickly. For many years she's been living in Kansas. Playing crochet is something that I've done for many years.<br><br>Here is my website ... [http://dienoobs-css.de/index.php?mod=users&action=view&id=17012 http://dienoobs-css.de]
'''CMA-ES''' stands for Covariance Matrix Adaptation Evolution Strategy. [[Evolution strategies]] (ES) are [[stochastic]], [[derivative]]-free methods for [[numerical optimization]]  of non-[[Linear map|linear]] or non-[[Convex function|convex]] [[continuous optimization]] problems. They belong to the class of [[evolutionary algorithms]] and [[evolutionary computation]].  An [[evolutionary algorithm]] is broadly based on the principle of [[biological evolution]], namely the repeated interplay of variation (via mutation and recombination) and selection: in each generation (iteration) new individuals (candidate solutions, denoted as <math>x</math>) are generated by variation, usually in a stochastic way, and then some individuals are selected for the next generation based on their fitness or [[objective function]] value <math>f(x)</math>. Like this, over the generation sequence, individuals with better and better <math>f</math>-values are generated.
 
In an [[evolution strategy]], new candidate solutions are sampled according to a [[multivariate normal distribution]] in the <math>\mathbb{R}^n</math>. Pairwise dependencies between the variables in this distribution are represented by a [[covariance matrix]]. The covariance matrix adaptation (CMA) is a method to update the [[covariance matrix]] of this distribution. This is particularly useful, if the function <math>f</math> is [[ill-conditioned]].
 
Adaptation of the [[covariance matrix]] amounts to learning a second order model of the underlying [[objective function]] similar to the approximation of the inverse [[Hessian matrix]] in the [[Quasi-Newton method]] in classical [[Optimization (mathematics)|optimization]]. In contrast to most classical methods, fewer assumptions on the nature of the underlying objective function are made. Only the ranking between candidate solutions is exploited for learning the sample distribution and neither derivatives nor even the function values themselves are required by the method.
 
== Principles ==
[[Image:Concept of directional optimization in CMA-ES algorithm.png|thumb|right|400px|Illustration of an actual optimization run with covariance matrix adaptation on a simple two-dimensional problem. The spherical optimization landscape is depicted with solid lines of equal <math>f</math>-values. The population (dots) is much larger than necessary, but clearly shows how the distribution of the population (dotted line) changes during the optimization. On this simple problem, the population concentrates over the global optimum within a few generations.]]
Two main principles for the adaptation of parameters of the search distribution are exploited in the CMA-ES algorithm.
 
First, a [[maximum-likelihood]] principle, based on the idea to increase the probability of successful candidate solutions and search steps. The mean of the distribution is updated such that the [[likelihood]] of previously successful candidate solutions is maximized. The [[covariance matrix]] of the distribution is updated (incrementally) such that the likelihood of previously successful search steps is increased. Both updates can be interpreted as a [[Information geometry#Natural gradient|natural gradient]] descent. Also, in consequence, the CMA conducts an iterated [[principal components analysis]] of successful search steps while retaining ''all'' principal axes. [[Estimation of Distribution Algorithms|Estimation of distribution algorithms]] and the [[Cross-Entropy Method]] are based on very similar ideas, but estimate (non-incrementally) the covariance matrix by maximizing the likelihood of successful solution ''points'' instead of successful search ''steps''.
 
Second, two paths of the time evolution of the distribution mean of the strategy are recorded, called search or evolution paths. These paths contain significant information about the correlation between consecutive steps. Specifically, if consecutive steps are taken in a similar direction, the evolution paths become long. The evolution paths are exploited in two ways. One path is used for the covariance matrix adaptation procedure in place of single successful search steps and facilitates a possibly much faster variance increase of favorable directions. The other path is used to conduct an additional step-size control. This step-size control aims to make consecutive movements of the distribution mean orthogonal in expectation. The step-size control effectively prevents [[premature convergence]] yet allowing fast convergence to an optimum.
 
== Algorithm ==
In the following the most commonly used (μ/μ<sub>w</sub>, λ)-CMA-ES is outlined, where in each iteration step a weighted combination of the μ best out of λ new candidate solutions is used to update the distribution parameters. The main loop consists of three main parts: 1) sampling of new solutions, 2) re-ordering of the sampled solutions based on their fitness, 3) update of the internal state variables based on the re-ordered samples. A [[pseudocode]] of the algorithm looks as follows.
<code>
  '''set''' <math>\lambda</math>  // number of samples per iteration, at least two, generally > 4
  '''initialize''' <math>m</math>, <math>\sigma</math>, <math>C=I</math>, <math>p_\sigma=0</math>, <math>p_c=0</math>  // initialize state variables
  '''while''' ''not terminate''  // iterate
    '''for''' <math>i</math> '''in''' <math>\{1...\lambda\}</math>  // sample <math>\lambda</math> new solutions and evaluate them
        <math>x_i</math> = sample_multivariate_normal(mean=<math>m</math>, covariance_matrix=<math>\sigma^2 C </math>)
        <math>f_i</math> = fitness(<math>x_i</math>)
    <math>x_{1...\lambda}</math> &larr; <math>x_{s(1)...s(\lambda)}</math> with <math>s(i)</math> = argsort(<math>f_{1...\lambda}</math>, <math>i</math>)  // sort solutions
    <math>m'</math> = <math>m</math>  // we need later <math>m - m'</math> and <math>x_i - m'</math>     
    <math>m</math> &larr; update_m<math>(x_1, ... ,</math> <math>x_\lambda)</math>  // move mean to better solutions
    <math>p_\sigma</math> &larr; update_ps<math>(p_\sigma,</math> <math>\sigma^{-1} C^{-1/2} (m - m'))</math>  // update isotropic evolution path
    <math>p_c</math> &larr; update_pc<math>(p_c,</math> <math>\sigma^{-1}(m - m'),</math> <math>||p_\sigma||)</math>  // update anisotropic evolution path
    <math>C</math> &larr; update_C<math>(C,</math> <math>p_c,</math> <math>{(x_1 - m')}/{\sigma},... ,</math> <math>{(x_\lambda - m')}/{\sigma})</math>  // update covariance matrix
    <math>\sigma</math> &larr; update_sigma<math>(\sigma,</math> <math>||p_\sigma||)</math>  // update step-size using isotropic path length
  '''return''' <math>m</math> or <math>x_1</math>
</code>
The order of the five update assignments is relevant. In the following, the update equations for the five state variables are specified.
 
Given are the search space dimension <math>n</math> and the iteration step <math>k</math>. The five state variables are
 
: <math>m_k\in\mathbb{R}^n</math>, the distribution mean and current favorite solution to the optimization problem,
 
: <math>\sigma_k>0</math>, the step-size,
 
: <math> C_k</math>, a symmetric and [[Positive-definite matrix|positive definite]] <math>n\times n</math> [[covariance matrix]] with <math> C_0 = I</math> and
 
: <math> p_\sigma\in\mathbb{R}^n, p_c\in\mathbb{R}^n</math>, two evolution paths, initially set to the zero vector.
 
The iteration starts with sampling <math>\lambda>1</math> candidate solutions <math>x_i\in\mathbb{R}^n </math> from a [[multivariate normal distribution]] <math>\textstyle \mathcal{N}(m_k,\sigma_k^2 C_k)</math>, i.e.
for <math>i=1,...,\lambda</math>
 
:: <math>
\begin{align}
  x_i \ &\sim\ \mathcal{N}(m_k,\sigma_k^2 C_k)     
    \\&\sim\ m_k + \sigma_k\times\mathcal{N}(0,C_k)
\end{align}
  </math>
 
The second line suggests the interpretation as perturbation (mutation) of the current favorite solution vector <math>m_k</math> (the distribution mean vector). The candidate solutions <math> x_i</math> are evaluated on the objective function <math>f:\mathbb{R}^n\to\mathbb{R}</math> to be minimized.  Denoting the <math>f</math>-sorted candidate solutions as
 
: <math>
    \{x_{i:\lambda}\;|\;i=1\dots\lambda\} = \{x_i\;|\;i=1\dots\lambda\} \;\;\text{and}\;\;
    f(x_{1:\lambda})\le\dots\le f(x_{\mu:\lambda})\le f(x_{\mu+1:\lambda}) \dots,     
  </math>
 
the new mean value is computed as
 
:: <math>
\begin{align}
  m_{k+1} &= \sum_{i=1}^{\mu} w_i\, x_{i:\lambda}
  \\ &= m_k + \sum_{i=1}^{\mu} w_i\, (x_{i:\lambda} - m_k)
\end{align}
  </math>
 
where the positive (recombination) weights <math> w_1 \ge w_2 \ge \dots \ge w_\mu > 0</math> sum to one. Typically, <math>\mu \le \lambda/2</math> and the weights are chosen such that <math>\textstyle \mu_w := 1 / \sum_{i=1}^\mu w_i^2 \approx \lambda/4</math>. The only feedback used from the objective function here and in the following is an ordering of the sampled candidate solutions due to the indices <math>i:\lambda</math>.
 
The step-size <math>\sigma_k</math> is updated using ''cumulative step-size adaptation'' (CSA), sometimes also denoted as ''path length control''. The evolution path (or search path) <math>p_\sigma</math> is updated first.
 
:: <math>
  p_\sigma \gets \underbrace{(1-c_\sigma)}_{\!\!\!\!\!\text{discount factor}\!\!\!\!\!}\, p_\sigma
    + \overbrace{\sqrt{1 - (1-c_\sigma)^2}}^{
    \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\text{complements for discounted variance}
    \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!} \underbrace{\sqrt{\mu_w}
    \,C_k^{\;-1/2} \, \frac{\overbrace{m_{k+1} - m_k}^{\!\!\!\text{displacement of}\; m\!\!\!}}{\sigma_k}}_{\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!
                      \text{distributed as}\; \mathcal{N}(0,I)\;\text{under neutral selection}
                      \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!}
  </math>
:: <math>
  \sigma_{k+1} = \sigma_k \times \exp\bigg(\frac{c_\sigma}{d_\sigma}
                          \underbrace{\left(\frac{\|p_\sigma\|}{E\|\mathcal{N}(0,I)\|} - 1\right)}_{\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!
    \text{unbiased about 0 under neutral selection}
    \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!
}\bigg)
  </math>
 
where
 
: <math>c_\sigma^{-1}\approx n/3</math> is the backward time horizon for the evolution path <math>p_\sigma</math> and larger than one,
 
: <math>\mu_w=\left(\sum_{i=1}^\mu w_i^2\right)^{-1}</math> is the variance effective selection mass and <math>1 \le \mu_w \le \mu</math> by definition of <math>w_i</math>,
 
: <math>C_k^{\;-1/2} = \sqrt{C_k}^{\;-1} = \sqrt{C_k^{\;-1}}</math> is the unique symmetric [[Square root of a matrix|square root]] of the [[Invertible matrix|inverse]] of <math>C_k</math>, and
 
: <math>d_\sigma</math> is the damping parameter usually close to one. For <math>d_\sigma=\infty</math> or <math>c_\sigma=0</math> the step-size remains unchanged.
 
The step-size <math>\sigma_k</math> is increased if and only if <math>\|p_\sigma\|</math> is larger than the [[expected value]]
 
: <math>\begin{align}E\|\mathcal{N}(0,I)\| &= \sqrt{2}\,\Gamma((n+1)/2)/\Gamma(n/2)
  \\&\approx \sqrt{n}\,(1-1/(4\,n)+1/(21\,n^2)) \end{align}</math>
 
and decreased if it is smaller. For this reason, the step-size update tends to make consecutive steps [[Conjugate gradient#The conjugate gradient method as a direct method|<math>C_k^{-1}</math>-conjugate]], in that  after the adaptation has been successful <math>\textstyle\left(\frac{m_{k+2}-m_{k+1}}{\sigma_{k+1}}\right)^T\! C_k^{-1} \frac{m_{k+1}-m_{k}}{\sigma_k} \approx 0</math>.<ref>{{Citation
| first = N.
| last = Hansen
| chapter = The CMA evolution strategy: a comparing review
| title = Towards a new evolutionary computation. Advances on estimation of distribution algorithms
| pages = 1769–1776
| publisher = Springer
| year = 2006
| location =
| url =
| accessdate =
| id =
}}</ref>
 
Finally, the [[covariance matrix]] is updated, where again the respective evolution path is updated first.
 
:: <math>
  p_c \gets \underbrace{(1-c_c)}_{\!\!\!\!\!\text{discount factor}\!\!\!\!\!}\,
            p_c +
    \underbrace{\mathbf{1}_{[0,\alpha\sqrt{n}]}(\|p_\sigma\|)}_{\text{indicator function}}
    \overbrace{\sqrt{1 - (1-c_c)^2}}^{
    \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\text{complements for discounted variance}
    \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!}
    \underbrace{\sqrt{\mu_w}
    \, \frac{m_{k+1} - m_k}{\sigma_k}}_{\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!
                      \text{distributed as}\; \mathcal{N}(0,C_k)\;\text{under neutral selection}
                      \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!}
  </math>
 
:: <math>
  C_{k+1} = \underbrace{(1 - c_1 - c_\mu + c_s)}_{\!\!\!\!\!\text{discount factor}\!\!\!\!\!}
              \, C_k + c_1 \underbrace{p_c p_c^T}_{
  \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!
  \text{rank one matrix}
  \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!}
        + \,c_\mu \underbrace{\sum_{i=1}^\mu w_i \frac{x_{i:\lambda} - m_k}{\sigma_k}
            \left( \frac{x_{i:\lambda} - m_k}{\sigma_k} \right)^T}_{
                    \text{rank} \;\min(\mu,n)\; \text{matrix}}
  </math>
 
where <math>T</math> denotes the transpose and
 
: <math>c_c^{-1}\approx n/4</math> is the backward time horizon for the evolution path <math>p_c</math> and larger than one,
 
: <math>\alpha\approx 1.5</math> and the [[indicator function]] <math>\mathbf{1}_{[0,\alpha\sqrt{n}]}(\|p_\sigma\|)</math> evaluates to one [[if and only if|iff]] <math>\|p_\sigma\|\in[0,\alpha\sqrt{n}]</math> or, in other words, <math>\|p_\sigma\|\le\alpha\sqrt{n}</math>, which is usually the case,
 
: <math>c_s = (1 - \mathbf{1}_{[0,\alpha\sqrt{n}]}(\|p_\sigma\|)^2) \,c_1 c_c (2-c_c) </math> makes partly up for the small variance loss in case the indicator is zero,
 
: <math>c_1 \approx 2 / n^2</math> is the learning rate for the rank-one update of the [[covariance matrix]] and
 
: <math>c_\mu \approx \mu_w / n^2 </math> is the learning rate for the rank-<math>\mu</math> update of the [[covariance matrix]] and must not exceed <math>1 - c_1</math>.
 
The [[covariance matrix]] update tends to increase the [[Likelihood function|likelihood]] for <math>p_c</math> and for <math>(x_{i:\lambda} - m_k)/\sigma_k</math> to be sampled from <math>\mathcal{N}(0,C_{k+1})</math>. This completes the iteration step.
 
The number of candidate samples per iteration, <math>\lambda</math>, is not determined a priori and can vary in a wide range. Smaller values, for example <math>\lambda=10</math>, lead to more local search behavior. Larger values, for example <math>\lambda=10n</math> with default value <math>\mu_w \approx \lambda/4</math>, render the search more global. Sometimes the algorithm is repeatedly restarted with increasing <math>\lambda</math> by a factor of two for each restart.<ref>{{cite conference
| first = A.
| last = Auger
| coauthors = N. Hansen
| title = A Restart CMA Evolution Strategy With Increasing Population Size
| booktitle = 2005 IEEE Congress on Evolutionary Computation, Proceedings
| pages = 1769–1776
| publisher = IEEE
| year = 2005
| location =
| url=http://www.lri.fr/~auger/cec-restartcma.pdf
| accessdate =
| id =
}}</ref> Besides of setting <math>\lambda</math> (or possibly <math>\mu</math> instead, if for example <math>\lambda</math> is predetermined by the number of available processors), the above introduced parameters are not specific to the given objective function and therefore not meant to be modified by the user.
 
==Example code in MATLAB/Octave==
<syntaxhighlight lang="matlab">
function xmin=purecmaes  % (mu/mu_w, lambda)-CMA-ES
 
  % --------------------  Initialization -------------------------------- 
  % User defined input parameters (need to be edited)
  strfitnessfct = 'frosenbrock';  % name of objective/fitness function
  N = 20;              % number of objective variables/problem dimension
  xmean = rand(N,1);    % objective variables initial point
  sigma = 0.3;          % coordinate wise standard deviation (step size)
  stopfitness = 1e-10;  % stop if fitness < stopfitness (minimization)
  stopeval = 1e3*N^2;  % stop after stopeval number of function evaluations
 
  % Strategy parameter setting: Selection 
  lambda = 4+floor(3*log(N));  % population size, offspring number
  mu = lambda/2;              % number of parents/points for recombination
  weights = log(mu+1/2)-log(1:mu)'; % muXone array for weighted recombination
  mu = floor(mu);       
  weights = weights/sum(weights);    % normalize recombination weights array
  mueff=sum(weights)^2/sum(weights.^2); % variance-effectiveness of sum w_i x_i
 
  % Strategy parameter setting: Adaptation
  cc = (4+mueff/N) / (N+4 + 2*mueff/N);  % time constant for cumulation for C
  cs = (mueff+2) / (N+mueff+5);  % t-const for cumulation for sigma control
  c1 = 2 / ((N+1.3)^2+mueff);    % learning rate for rank-one update of C
  cmu = min(1-c1, 2 * (mueff-2+1/mueff) / ((N+2)^2+mueff));  % and for rank-mu update
  damps = 1 + 2*max(0, sqrt((mueff-1)/(N+1))-1) + cs; % damping for sigma
                                                      % usually close to 1
  % Initialize dynamic (internal) strategy parameters and constants
  pc = zeros(N,1); ps = zeros(N,1);  % evolution paths for C and sigma
  B = eye(N,N);                      % B defines the coordinate system
  D = ones(N,1);                      % diagonal D defines the scaling
  C = B * diag(D.^2) * B';            % covariance matrix C
  invsqrtC = B * diag(D.^-1) * B';    % C^-1/2
  eigeneval = 0;                      % track update of B and D
  chiN=N^0.5*(1-1/(4*N)+1/(21*N^2));  % expectation of
                                      %  ||N(0,I)|| == norm(randn(N,1))
 
  % -------------------- Generation Loop --------------------------------
  counteval = 0;  % the next 40 lines contain the 20 lines of interesting code
  while counteval < stopeval
   
      % Generate and evaluate lambda offspring
      for k=1:lambda,
          arx(:,k) = xmean + sigma * B * (D .* randn(N,1)); % m + sig * Normal(0,C)
          arfitness(k) = feval(strfitnessfct, arx(:,k)); % objective function call
          counteval = counteval+1;
      end
   
      % Sort by fitness and compute weighted mean into xmean
      [arfitness, arindex] = sort(arfitness); % minimization
      xold = xmean;
      xmean = arx(:,arindex(1:mu))*weights;  % recombination, new mean value
   
      % Cumulation: Update evolution paths
      ps = (1-cs)*ps ...
            + sqrt(cs*(2-cs)*mueff) * invsqrtC * (xmean-xold) / sigma;
      hsig = norm(ps)/sqrt(1-(1-cs)^(2*counteval/lambda))/chiN < 1.4 + 2/(N+1);
      pc = (1-cc)*pc ...
            + hsig * sqrt(cc*(2-cc)*mueff) * (xmean-xold) / sigma;
 
      % Adapt covariance matrix C
      artmp = (1/sigma) * (arx(:,arindex(1:mu))-repmat(xold,1,mu));
      C = (1-c1-cmu) * C ...                  % regard old matrix 
          + c1 * (pc*pc' ...                % plus rank one update
                  + (1-hsig) * cc*(2-cc) * C) ... % minor correction if hsig==0
          + cmu * artmp * diag(weights) * artmp'; % plus rank mu update
 
      % Adapt step size sigma
      sigma = sigma * exp((cs/damps)*(norm(ps)/chiN - 1));
   
      % Decomposition of C into B*diag(D.^2)*B' (diagonalization)
      if counteval - eigeneval > lambda/(c1+cmu)/N/10  % to achieve O(N^2)
          eigeneval = counteval;
          C = triu(C) + triu(C,1)'; % enforce symmetry
          [B,D] = eig(C);          % eigen decomposition, B==normalized eigenvectors
          D = sqrt(diag(D));        % D is a vector of standard deviations now
          invsqrtC = B * diag(D.^-1) * B';
      end
   
      % Break, if fitness is good enough or condition exceeds 1e14, better termination methods are advisable
      if arfitness(1) <= stopfitness || max(D) > 1e7 * min(D)
          break;
      end
 
  end % while, end generation loop
 
  xmin = arx(:, arindex(1)); % Return best point of last iteration.
                            % Notice that xmean is expected to be even
                            % better.
 
% --------------------------------------------------------------- 
function f=frosenbrock(x)
    if size(x,1) < 2 error('dimension must be greater one'); end
    f = 100*sum((x(1:end-1).^2 - x(2:end)).^2) + sum((x(1:end-1)-1).^2);
</syntaxhighlight>
 
== Theoretical Foundations ==
Given the distribution parameters&mdash;mean, variances and covariances&mdash;the [[multivariate normal distribution|normal probability distribution]] for sampling new candidate solutions is the [[maximum entropy probability distribution]] over <math>\mathbb{R}^n</math>, that is, the sample distribution with the minimal amount of prior information built into the distribution. More considerations on the update equations of CMA-ES are made in the following.
 
=== Variable Metric ===
The CMA-ES implements a stochastic [[variable-metric]] method. In the very particular case of a convex-quadratic objective function
 
:: <math> f(x) = {\textstyle\frac{1}{2}}(x-x^*)^T H (x-x^*)</math>
 
the covariance matrix <math>C_k</math> adapts to the inverse of the [[Hessian matrix]] <math>H</math>, [[up to]] a scalar factor and small random fluctuations. More general, also on the function <math>g \circ f</math>, where <math> g </math> is strictly increasing and therefore order preserving and <math>f</math> is convex-quadratic, the covariance matrix <math>C_k</math> adapts to <math>H^{-1}</math>, [[up to]] a scalar factor and small random fluctuations.
 
=== Maximum-Likelihood Updates ===
 
The update equations for mean and covariance matrix maximize a [[likelihood]] while resembling an [[Expectation maximization|expectation-maximization]] algorithm. The update of the mean vector <math>m</math> maximizes a log-likelihood, such that
 
:: <math> m_{k+1} = \arg\max_{m} \sum_{i=1}^\mu w_i \log p_\mathcal{N}(x_{i:\lambda} | m) </math>
 
where
 
:: <math> \log p_\mathcal{N}(x) =
  - \frac{1}{2} \log\det(2\pi C) - \frac{1}{2} (x-m)^T C^{-1} (x-m) </math>
 
denotes the log-likelihood of <math>x</math> from a multivariate normal distribution with mean <math>m</math> and any positive definite covariance matrix <math>C</math>. To see that <math>m_{k+1}</math> is independent of <math>C</math> remark first that this is the case for any diagonal matrix <math>C</math>, because the coordinate-wise maximizer is independent of a scaling factor. Then, rotation of the data points or choosing <math>C</math> non-diagonal are equivalent.
 
The rank-<math>\mu</math> update of the covariance matrix, that is, the right most summand in the update equation of <math>C_k</math>, maximizes a log-likelihood in that
 
:: <math> \sum_{i=1}^\mu w_i \frac{x_{i:\lambda} - m_k}{\sigma_k}
            \left( \frac{x_{i:\lambda} - m_k}{\sigma_k} \right)^T
    = \arg\max_{C} \sum_{i=1}^\mu w_i \log p_\mathcal{N}\left(\left.\frac{x_{i:\lambda} - m_k}{\sigma_k} \right| C\right) </math>
 
for <math>\mu\ge n</math> (otherwise <math>C</math> is singular, but substantially the same result holds for <math>\mu < n</math>). Here, <math> p_\mathcal{N}(x | C) </math> denotes the likelihood of <math>x</math> from a multivariate normal distribution with zero mean and covariance matrix <math>C</math>. Therefore, for <math>c_1=0</math> and <math>c_\mu=1</math>, <math>C_{k+1}</math> is the above [[maximum-likelihood]] estimator. See [[estimation of covariance matrices]] for details on the derivation.
 
=== Natural Gradient Descent in the Space of Sample Distributions ===
 
Akimoto ''et al.''<ref name=akimoto2010>{{cite conference
| first = Y.
| last = Akimoto
| coauthors = Y. Nagata and I. Ono and S. Kobayashi
| title = Bidirectional Relation between CMA Evolution Strategies and Natural Evolution Strategies
| booktitle = Parallel Problem Solving from Nature, PPSN XI
| pages = 154–163
| publisher = Springer
| year = 2010
| location =
| url =
| accessdate =
| id =
}}</ref> recently found that the update of the distribution parameters resembles the descend in direction of a sampled [[Information geometry#Natural gradient|natural gradient]] of the expected objective function value {{math|E <var>f</var> (<var>x</var>)</var>}} (to be minimized), where the expectation is taken under the sample distribution. With the parameter setting of <math>c_\sigma=0</math> and <math>c_1=0</math>, i.e. without step-size control and rank-one update, CMA-ES can thus be viewed as an instantiation of [[Natural Evolution Strategies]] (NES).<ref name=akimoto2010/><ref name=glasmachers2010>{{cite conference
| first = T.
| last = Glasmachers
| coauthors = T. Schaul, Y. Sun, D. Wierstra and J. Schmidhuber
| title = Exponential Natural Evolution Strategies
| booktitle = Genetic and Evolutionary Computation Conference GECCO
| year = 2010
| location = Portland, OR
| url = http://www.idsia.ch/~tom/publications/xnes.pdf
}}</ref>
The natural gradient is independent of the parameterization of the distribution. Taken with respect to the parameters {{math|<var>&theta;</var>}} of the sample distribution {{math|<var>p</var>}}, the gradient of {{math|E <var>f</var> (<var>x</var>)</var>}} can be expressed as
 
:: <math> \begin{align}
  {\nabla}_{\!\theta} E(f(x) | \theta)
    &= \nabla_{\!\theta} \int_{\mathbb R^n}f(x) p(x) \mathrm{d}x
  \\ &= \int_{\mathbb R^n}f(x) \nabla_{\!\theta} p(x) \mathrm{d}x
  \\ &= \int_{\mathbb R^n}f(x) p(x) \nabla_{\!\theta} \ln p(x) \mathrm{d}x
  \\ &= E(f(x) \nabla_{\!\theta} \ln p(x|\theta))
\end{align}</math>
 
where <math>p(x)=p(x|\theta)</math> depends on the parameter vector <math>\theta</math>, the so-called [[Score (statistics)|score function]], <math>\nabla_{\!\theta} \ln p(x|\theta) = \frac{\nabla_{\!\theta} p(x)}{p(x)} </math>, indicates the relative sensitivity of {{math|<var>p</var>}} w.r.t. {{math|<var>&theta;</var>}}, and the expectation is taken with respect to the distribution {{math|<var>p</var>}}. The [[Information geometry#Natural gradient|''natural'' gradient]] of {{math|E <var>f</var> (<var>x</var>)</var>}}, complying with the [[Fisher information metric]] (an informational distance measure between probability distributions and the curvature of the [[relative entropy]]), now reads
 
:: <math> \begin{align}
  \tilde{\nabla} E(f(x) | \theta)
  &= F^{-1}_\theta \nabla_{\!\theta} E(f(x) | \theta)
\end{align}</math>
 
where the [[Fisher information]] matrix <math> F_{\theta} </math> is the expectation of the [[Hessian matrix|Hessian]] of {{math|-ln<var>p</var>}} and renders the expression independent of the chosen parameterization. Combining the previous equalities we get
 
:: <math> \begin{align}
  \tilde{\nabla} E(f(x) | \theta)
  &= F^{-1}_\theta E(f(x) \nabla_{\!\theta} \ln p(x|\theta))
  \\ &= E(f(x) F^{-1}_\theta \nabla_{\!\theta} \ln p(x|\theta))
\end{align}</math>
 
A Monte Carlo approximation of the latter expectation takes the average over {{math|<var>&lambda;</var>}} samples from {{math|<var>p</var>}}
 
:: <math> \tilde{\nabla} \widehat{E}_\theta(f) := -\sum_{i=1}^\lambda \overbrace{w_i}^{\!\!\!\!\text{preference weight}\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!} \underbrace{F^{-1}_\theta \nabla_{\!\theta} \ln p(x_{i:\lambda}|\theta)}_{\!\!\!\!\!\text{candidate direction from }x_{i:\lambda}\!\!\!\!\!}
\quad\mathrm{with~}w_i = -f(x_{i:\lambda})/\lambda</math>
 
where the notation <math>i:\lambda</math> from above is used and therefore <math>w_i</math> are monotonously decreasing in <math>i</math>.
We might use, for a more robust approximation, rather <math>w_i</math> as defined in the CMA-ES and zero for {{math|<var>i</var> > <var>&mu;</var>}} (implementing a consistent estimator for the [[Cumulative distribution function|CDF]] of <math>f(X), X\sim p(.|\theta)</math> at the point <math>f(x_{i:\lambda})</math> composed with a fixed monotonous decreased transformation) and let
 
:: <math>\theta = [m_k^T \mathrm{vec}(C_k)^T \sigma_k]^T \in \mathbb{R}^{n+n^2+1}</math>
 
such that <math> p(.|\theta) </math> is the density of the [[multivariate normal distribution]] <math>\mathcal N(m_k,\sigma_k^2 C_k)</math>. Then, we have an explicit expression for the inverse of the Fisher information matrix where <math>\sigma_k</math> is fixed
 
:: <math>F^{-1}_{\theta | \sigma_k} = \left[\begin{array}{cc}\sigma_k^2 C_k&0\\ 0&2 C_k\otimes C_k\end{array}\right]</math>
 
and for
 
:: <math>\ln p(x|\theta) = \ln p(x|m_k,\sigma_k^2 C_k) = -\frac{1}{2}(x-m_k)^T \sigma_k^{-2} C_k^{-1} (x-m_k)
        \,-\, \frac{1}{2}\ln\det(2\pi\sigma_k^2 C_k)</math>
 
and, after some calculations, the updates in the CMA-ES turn out as<ref name=akimoto2010/>
 
<span id="update_in_gradient_formulation">
:: <math> \begin{align}
  m_{k+1}
    &= m_k - \underbrace{[\tilde{\nabla} \widehat{E}_\theta(f)]_{1,\dots, n}}_{
    \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!
    \text{natural gradient for mean}
    \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!
    }
  \\
    &= m_k + \sum_{i=1}^\lambda w_i (x_{i:\lambda} - m_k)
  \end{align} </math>
 
and
 
:: <math> \begin{align}
  C_{k+1}
    &= C_k + c_1(p_c p_c^T - C_k) 
        - c_\mu\,\mathrm{mat}(\overbrace{[\tilde{\nabla} \widehat{E}_\theta(f)]_{n+1,\dots,n+n^2}}^{
    \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!
        \text{natural gradient for covariance matrix}
    \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!
    })\\
    &= C_k + c_1(p_c p_c^T - C_k)
        + c_\mu \sum_{i=1}^\lambda w_i \left(\frac{x_{i:\lambda} - m_k}{\sigma_k} \left(\frac{x_{i:\lambda} - m_k}{\sigma_k}\right)^T - C_k\right)
  \end{align}</math>
</span>
 
where mat forms the proper matrix from the respective natural gradient sub-vector. That means, setting <math>c_1=c_\sigma=0</math>, the CMA-ES updates descend in direction of the approximation <math> \tilde{\nabla} \widehat{E}_\theta(f)</math> of the natural gradient while using different step-sizes (learning rates) for the [[Fisher information#Orthogonal parameters|orthogonal parameters]] <math>m</math> and <math>C</math> respectively.
 
=== Stationarity or Unbiasedness ===
It is comparatively easy to see that the update equations of CMA-ES satisfy some stationarity conditions, in that they are essentially unbiased. Under neutral selection, where <math>x_{i:\lambda} \sim \mathcal N(m_k,\sigma_k^2 C_k)</math>,  we find that
 
:: <math> E(m_{k+1}\,|\, m_k) = m_k </math>
 
and under some mild additional assumptions on the initial conditions
 
:: <math> E(\log \sigma_{k+1} \,|\, \sigma_k) = \log \sigma_k </math>
 
and with an additional minor correction in the covariance matrix update for the case where the indicator function evaluates to zero, we find
 
:: <math> E(C_{k+1} \,|\, C_k) = C_k </math>
 
=== Invariance ===
[[Invariant (mathematics)|Invariance properties]] imply uniform performance on a class of objective functions. They have been argued to be an advantage, because they allow to generalize and predict the behavior of the algorithm and therefore strengthen the meaning of empirical results obtained on single functions. The following invariance properties have been established for CMA-ES.
 
* Invariance under order-preserving transformations of the objective function value <math>f</math>, in that for any <math>h:\mathbb{R}^n\to\mathbb{R}</math> the behavior is identical on <math>f:x\mapsto g(h(x))</math> for all strictly increasing <math>g:\mathbb{R}\to\mathbb{R}</math>. This invariance is easy to verify, because only the <math>f</math>-ranking is used in the algorithm, which is invariant under the choice of <math>g</math>.
 
* [[Scale-invariance]], in that for any <math>h:\mathbb{R}^n\to \mathbb{R}</math> the behavior is independent of <math>\alpha>0</math> for the objective function <math>f:x\mapsto h(\alpha x)</math> given <math>\sigma_0\propto1/\alpha</math> and <math>m_0\propto1/\alpha</math>.
 
* Invariance under rotation of the search space in that for any <math>h:\mathbb{R}^n\to \mathbb{R}</math> and any <math>z\in\mathbb{R}^n</math> the behavior on <math>f:x\mapsto h(R x)</math>  is independent of the [[orthogonal matrix]] <math>R</math>, given <math>m_0=R^{-1} z</math>. More general, the algorithm is also invariant under general linear transformations <math>R</math> when additionally the initial covariance matrix is chosen as <math>R^{-1}{R^{-1}}^T</math>.
 
Any serious parameter optimization method should be translation invariant, but most methods do not exhibit all the above described invariance properties. A prominent example with the same invariance properties is the [[Nelder–Mead method]], where the initial simplex must be chosen respectively.
 
=== Convergence ===
 
Conceptual considerations like the scale-invariance property of the algorithm, the analysis of simpler [[evolution strategies]], and overwhelming empirical evidence suggest that the algorithm converges on a large class of functions fast to the global optimum, denoted as <math>x^*</math>. On some functions, convergence occurs independently of the initial conditions with probability one. On some functions the probability is smaller than one and typically depends on the initial <math>m_0</math> and <math>\sigma_0</math>. Empirically, the fastest possible convergence rate in <math>k</math> for rank-based direct search methods can often be observed (depending on the context denoted as ''[[Rate of convergence#linear convergence|linear]]'' or ''log-linear'' or ''exponential'' convergence). Informally, we can write
 
:: <math>\|m_k - x^*\| \;\approx\; \|m_0 - x^*\| \times e^{-ck}
  </math>
 
for some <math>c>0</math>, and more rigorously
 
:: <math>\frac{1}{k}\sum_{i=1}^k\log\frac{\|m_i - x^*\|}{\|m_{i-1} - x^*\|}
  \;=\; \frac{1}{k}\log\frac{\|m_k - x^*\|}{\|m_{0} - x^*\|}
    \;\to\; -c < 0 \quad\text{for}\; k\to\infty\;,
  </math>
 
or similarly,
 
:: <math>E\log\frac{\|m_k - x^*\|}{\|m_{k-1} - x^*\|}
    \;\to\; -c < 0 \quad\text{for}\; k\to\infty\;.
  </math>
 
This means that on average the distance to the optimum decreases in each iteration by a "constant" factor, namely by <math>\exp(-c)</math>. The convergence rate <math>c</math> is roughly <math>0.1\lambda/n</math>, given <math>\lambda</math> is not much larger than the dimension <math>n</math>. Even with optimal <math>\sigma</math> and <math>C</math>, the convergence rate <math>c</math>  cannot largely exceed <math>0.25\lambda/n</math>, given the above recombination weights <math>w_i</math> are all non-negative. The actual linear dependencies in <math>\lambda</math> and <math>n</math> are remarkable and they are in both cases the best one can hope for in this kind of algorithm. Yet, a rigorous proof of convergence is missing.
 
===Interpretation as Coordinate System Transformation===
Using a non-identity covariance matrix for the [[multivariate normal distribution]] in [[evolution strategies]] is equivalent to a coordinate system transformation of the solution vectors,<ref name=hansen2008 /> mainly because the sampling equation
 
:<math>
\begin{align}
  x_i &\sim\ m_k + \sigma_k\times\mathcal{N}(0,C_k)
    \\
  &\sim\ m_k + \sigma_k \times C_k^{1/2}\mathcal{N}(0,I)
\end{align}
</math>
 
can be equivalently expressed in an "encoded space" as
:<math>
  \underbrace{C_k^{-1/2}x_i}_{\text{represented in the encode space}
    \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!}
    \sim\ \underbrace{C_k^{-1/2} m_k} {} + \sigma_k \times\mathcal{N}(0,I)
</math>
 
The covariance matrix defines a [[bijective]] transformation (encoding) for all solution vectors into a space, where the sampling takes place with identity covariance matrix. Because the update equations in the CMA-ES are invariant under linear coordinate system transformations, the CMA-ES can be re-written as an adaptive encoding procedure applied to a simple [[evolution strategy]] with identity covariance matrix.<ref name=hansen2008>{{cite conference
| first = N.
| last = Hansen
| title = Adpative Encoding: How to Render Search Coordinate System Invariant
| booktitle = Parallel Problem Solving from Nature, PPSN X
| pages = 205–214
| publisher = Springer
| year = 2008
| location =
| url = http://hal.archives-ouvertes.fr/inria-00287351/en/
| accessdate =
| id =
}}</ref>
This adaptive encoding procedure is not confined to algorithms that sample from a multivariate normal distribution (like evolution strategies), but can in principle be applied to any iterative search method.
 
== Performance in Practice ==
 
In contrast to most other [[evolutionary algorithms]], the CMA-ES is, from the users perspective, quasi parameter-free. However, the number of candidate samples λ (population size) can be adjusted by the user in order to change the characteristic search behavior (see above). CMA-ES has been empirically successful in hundreds of applications and is considered to be useful in particular on non-convex, non-separable, ill-conditioned, multi-modal or noisy objective functions. The search space dimension ranges typically between two and a few hundred. Assuming a black-box optimization scenario, where gradients are not available (or not useful) and function evaluations are the only considered cost of search, the CMA-ES method is likely to be outperformed by other methods in the following conditions:
 
* on low-dimensional functions, say <math>n < 5 </math>, for example by the [[Nelder-Mead method|downhill simplex method]] or surrogate-based methods (like [[kriging]] with expected improvement);
 
* on separable functions without or with only negligible dependencies between the design variables in particular in the case of multi-modality or large dimension, for example by [[differential evolution]];
 
* on (nearly) [[Convex function|convex]]-quadratic functions with low or moderate [[condition number]] of the [[Hessian matrix]], where [[BFGS method|BFGS]] or [[NEWUOA]] are typically ten times faster;
 
* on functions that can already be solved with a comparatively small number of function evaluations, say no more than <math>10 n</math>, where CMA-ES is often slower than, for example, [[NEWUOA]] or [[MCS algorithm|Multilevel Coordinate Search]] (MCS ).
 
On separable functions, the performance disadvantage is likely to be most significant in that CMA-ES might not be able to find at all comparable solutions. On the other hand, on non-separable functions that are ill-conditioned or rugged or can only be solved with more than <math>100 n</math> function evaluations, the CMA-ES shows most often superior performance.
 
== Variations and Extensions ==
The (1+1)-CMA-ES<ref>{{cite conference
| first = C.
| last = Igel
| coauthors = T. Suttorp and N. Hansen
| title = A Computational Efficient Covariance Matrix Update and a (1+1)-CMA for Evolution Strategies
| booktitle = Proceedings of the Genetic and Evolutionary Computation Conference (GECCO)
| pages = 453–460
| publisher = ACM Press
| year = 2006
| location =
| url = http://www.cs.york.ac.uk/rts/docs/GECCO_2006/docs/p453.pdf
| accessdate =
| id =
}}</ref> generates only one candidate solution per iteration step which becomes the new distribution mean if it is better than the current mean. For <math>c_c=1</math> the (1+1)-CMA-ES is a close variant of [[Gaussian adaptation]]. Some [[Natural Evolution Strategies]] are close variants of the CMA-ES with specific parameter settings. Natural Evolution Strategies do not utilize evolution paths (that means in CMA-ES setting <math>c_c=c_\sigma=1</math>) and they formalize the update of variances and covariances on a [[Cholesky decomposition|Cholesky factor]] instead of a covariance matrix. The CMA-ES has also been extended to [[multiobjective optimization]] as MO-CMA-ES.<ref>{{cite journal
| doi = 10.1162/evco.2007.15.1.1
| first = C.
| last = Igel
| coauthors = N. Hansen and S. Roth
| title = Covariance Matrix Adaptation for Multi-objective Optimization
| journal = Evolutionary Computation
| volume = 15
| issue = 1
| pages = 1–28
| publisher = MIT press
| year = 2007
| location =
| url = http://www.mitpressjournals.org/doi/pdfplus/10.1162/evco.2007.15.1.1
| pmid = 17388777
| accessdate =
| id =
}}</ref> Another remarkable extension has been the addition of a negative update of the covariance matrix with the so-called active CMA.<ref>{{cite conference
| first = G.A.
| last = Jastrebski
| coauthors = D.V. Arnold
| title = Improving Evolution Strategies through Active Covariance Matrix Adaptation
| booktitle = 2006 IEEE World Congress on Computational Intelligence, Proceedings
| pages = 9719–9726
| publisher = IEEE
| year = 2006
| location =
| doi = 10.1109/CEC.2006.1688662
| accessdate =
| id =
}}</ref>
 
With the advent of niching methods in evolutionary strategies, the question of an optimal niche radius arises. An "adaptive individual niche radius" is introduced in <ref>
{{cite book
| first = Ofer M.
| last = Shir
| coauthors = Bäck, Thomas
| booktitle = Parallel Problem Solving from Nature-PPSN IX
| pages = 142–151
| year = 2006
| location =
| publisher= Springer
}}</ref>
 
== See also ==
* [[Global optimization]]
* [[Stochastic optimization]]
 
== References ==
{{Reflist}}
 
==Bibliography==
*Hansen N, Ostermeier A (2001). Completely derandomized self-adaptation in evolution strategies. [http://www.mitpressjournals.org/toc/evco/9/2 ''Evolutionary Computation'', '''9'''(2)] pp.&nbsp;159–195. [http://www.lri.fr/~hansen/cmaartic.pdf]
*Hansen N, Müller SD, Koumoutsakos P (2003). Reducing the time complexity of the derandomized evolution strategy with covariance matrix adaptation (CMA-ES). [http://www.mitpressjournals.org/toc/evco/11/1 ''Evolutionary Computation'', '''11'''(1)] pp.&nbsp;1–18.  [http://mitpress.mit.edu/journals/pdf/evco_11_1_1_0.pdf]
*Hansen N, Kern S (2004). Evaluating the CMA evolution strategy on multimodal test functions. In Xin Yao et al., editors, ''Parallel Problem Solving from Nature - PPSN VIII'', pp.&nbsp;282–291, Springer. [http://www.lri.fr/~hansen/ppsn2004hansenkern.pdf]
*Igel C, Hansen N, Roth S (2007). Covariance Matrix Adaptation for Multi-objective Optimization. [http://www.mitpressjournals.org/toc/evco/15/1 ''Evolutionary Computation'', '''15'''(1)] pp.&nbsp;1–28. [http://www.mitpressjournals.org/doi/pdfplus/10.1162/evco.2007.15.1.1]
 
==External links==
* [http://www.lri.fr/~hansen/cmaesintro.html A short introduction to CMA-ES by N. Hansen]
* [http://www.lri.fr/~hansen/cmatutorial.pdf The CMA Evolution Strategy: A Tutorial]
* [http://www.lri.fr/~hansen/cmaes_inmatlab.html CMA-ES source code page]
 
{{Major subfields of optimization}}
 
{{DEFAULTSORT:Cma-Es}}
[[Category:Evolutionary algorithms]]
[[Category:Stochastic optimization]]
[[Category:Optimization algorithms and methods]]
 
[[fr:Stratégie d'évolution#CMA-ES]]

Revision as of 12:48, 27 July 2013

CMA-ES stands for Covariance Matrix Adaptation Evolution Strategy. Evolution strategies (ES) are stochastic, derivative-free methods for numerical optimization of non-linear or non-convex continuous optimization problems. They belong to the class of evolutionary algorithms and evolutionary computation. An evolutionary algorithm is broadly based on the principle of biological evolution, namely the repeated interplay of variation (via mutation and recombination) and selection: in each generation (iteration) new individuals (candidate solutions, denoted as ) are generated by variation, usually in a stochastic way, and then some individuals are selected for the next generation based on their fitness or objective function value . Like this, over the generation sequence, individuals with better and better -values are generated.

In an evolution strategy, new candidate solutions are sampled according to a multivariate normal distribution in the . Pairwise dependencies between the variables in this distribution are represented by a covariance matrix. The covariance matrix adaptation (CMA) is a method to update the covariance matrix of this distribution. This is particularly useful, if the function is ill-conditioned.

Adaptation of the covariance matrix amounts to learning a second order model of the underlying objective function similar to the approximation of the inverse Hessian matrix in the Quasi-Newton method in classical optimization. In contrast to most classical methods, fewer assumptions on the nature of the underlying objective function are made. Only the ranking between candidate solutions is exploited for learning the sample distribution and neither derivatives nor even the function values themselves are required by the method.

Principles

Illustration of an actual optimization run with covariance matrix adaptation on a simple two-dimensional problem. The spherical optimization landscape is depicted with solid lines of equal -values. The population (dots) is much larger than necessary, but clearly shows how the distribution of the population (dotted line) changes during the optimization. On this simple problem, the population concentrates over the global optimum within a few generations.

Two main principles for the adaptation of parameters of the search distribution are exploited in the CMA-ES algorithm.

First, a maximum-likelihood principle, based on the idea to increase the probability of successful candidate solutions and search steps. The mean of the distribution is updated such that the likelihood of previously successful candidate solutions is maximized. The covariance matrix of the distribution is updated (incrementally) such that the likelihood of previously successful search steps is increased. Both updates can be interpreted as a natural gradient descent. Also, in consequence, the CMA conducts an iterated principal components analysis of successful search steps while retaining all principal axes. Estimation of distribution algorithms and the Cross-Entropy Method are based on very similar ideas, but estimate (non-incrementally) the covariance matrix by maximizing the likelihood of successful solution points instead of successful search steps.

Second, two paths of the time evolution of the distribution mean of the strategy are recorded, called search or evolution paths. These paths contain significant information about the correlation between consecutive steps. Specifically, if consecutive steps are taken in a similar direction, the evolution paths become long. The evolution paths are exploited in two ways. One path is used for the covariance matrix adaptation procedure in place of single successful search steps and facilitates a possibly much faster variance increase of favorable directions. The other path is used to conduct an additional step-size control. This step-size control aims to make consecutive movements of the distribution mean orthogonal in expectation. The step-size control effectively prevents premature convergence yet allowing fast convergence to an optimum.

Algorithm

In the following the most commonly used (μ/μw, λ)-CMA-ES is outlined, where in each iteration step a weighted combination of the μ best out of λ new candidate solutions is used to update the distribution parameters. The main loop consists of three main parts: 1) sampling of new solutions, 2) re-ordering of the sampled solutions based on their fitness, 3) update of the internal state variables based on the re-ordered samples. A pseudocode of the algorithm looks as follows.

 set   // number of samples per iteration, at least two, generally > 4
 initialize , , , ,   // initialize state variables
 while not terminate  // iterate
    for  in   // sample  new solutions and evaluate them
        = sample_multivariate_normal(mean=, covariance_matrix=)
        = fitness()
     with  = argsort(, )  // sort solutions
     =   // we need later  and        
     ← update_m   // move mean to better solutions 
     ← update_ps   // update isotropic evolution path
     ← update_pc    // update anisotropic evolution path
     ← update_C     // update covariance matrix
     ← update_sigma   // update step-size using isotropic path length
 return  or  

The order of the five update assignments is relevant. In the following, the update equations for the five state variables are specified.

Given are the search space dimension and the iteration step . The five state variables are

, the distribution mean and current favorite solution to the optimization problem,
, the step-size,
, a symmetric and positive definite covariance matrix with and
, two evolution paths, initially set to the zero vector.

The iteration starts with sampling candidate solutions from a multivariate normal distribution , i.e. for

The second line suggests the interpretation as perturbation (mutation) of the current favorite solution vector (the distribution mean vector). The candidate solutions are evaluated on the objective function to be minimized. Denoting the -sorted candidate solutions as

the new mean value is computed as

where the positive (recombination) weights sum to one. Typically, and the weights are chosen such that . The only feedback used from the objective function here and in the following is an ordering of the sampled candidate solutions due to the indices .

The step-size is updated using cumulative step-size adaptation (CSA), sometimes also denoted as path length control. The evolution path (or search path) is updated first.

where

is the backward time horizon for the evolution path and larger than one,
is the variance effective selection mass and by definition of ,
is the unique symmetric square root of the inverse of , and
is the damping parameter usually close to one. For or the step-size remains unchanged.

The step-size is increased if and only if is larger than the expected value

and decreased if it is smaller. For this reason, the step-size update tends to make consecutive steps -conjugate, in that after the adaptation has been successful .[1]

Finally, the covariance matrix is updated, where again the respective evolution path is updated first.

where denotes the transpose and

is the backward time horizon for the evolution path and larger than one,
and the indicator function evaluates to one iff or, in other words, , which is usually the case,
makes partly up for the small variance loss in case the indicator is zero,
is the learning rate for the rank-one update of the covariance matrix and
is the learning rate for the rank- update of the covariance matrix and must not exceed .

The covariance matrix update tends to increase the likelihood for and for to be sampled from . This completes the iteration step.

The number of candidate samples per iteration, , is not determined a priori and can vary in a wide range. Smaller values, for example , lead to more local search behavior. Larger values, for example with default value , render the search more global. Sometimes the algorithm is repeatedly restarted with increasing by a factor of two for each restart.[2] Besides of setting (or possibly instead, if for example is predetermined by the number of available processors), the above introduced parameters are not specific to the given objective function and therefore not meant to be modified by the user.

Example code in MATLAB/Octave

function xmin=purecmaes   % (mu/mu_w, lambda)-CMA-ES

  % --------------------  Initialization --------------------------------  
  % User defined input parameters (need to be edited)
  strfitnessfct = 'frosenbrock';  % name of objective/fitness function
  N = 20;               % number of objective variables/problem dimension
  xmean = rand(N,1);    % objective variables initial point
  sigma = 0.3;          % coordinate wise standard deviation (step size)
  stopfitness = 1e-10;  % stop if fitness < stopfitness (minimization)
  stopeval = 1e3*N^2;   % stop after stopeval number of function evaluations
  
  % Strategy parameter setting: Selection  
  lambda = 4+floor(3*log(N));  % population size, offspring number
  mu = lambda/2;               % number of parents/points for recombination
  weights = log(mu+1/2)-log(1:mu)'; % muXone array for weighted recombination
  mu = floor(mu);        
  weights = weights/sum(weights);     % normalize recombination weights array
  mueff=sum(weights)^2/sum(weights.^2); % variance-effectiveness of sum w_i x_i

  % Strategy parameter setting: Adaptation
  cc = (4+mueff/N) / (N+4 + 2*mueff/N);  % time constant for cumulation for C
  cs = (mueff+2) / (N+mueff+5);  % t-const for cumulation for sigma control
  c1 = 2 / ((N+1.3)^2+mueff);    % learning rate for rank-one update of C
  cmu = min(1-c1, 2 * (mueff-2+1/mueff) / ((N+2)^2+mueff));  % and for rank-mu update
  damps = 1 + 2*max(0, sqrt((mueff-1)/(N+1))-1) + cs; % damping for sigma 
                                                      % usually close to 1
  % Initialize dynamic (internal) strategy parameters and constants
  pc = zeros(N,1); ps = zeros(N,1);   % evolution paths for C and sigma
  B = eye(N,N);                       % B defines the coordinate system
  D = ones(N,1);                      % diagonal D defines the scaling
  C = B * diag(D.^2) * B';            % covariance matrix C
  invsqrtC = B * diag(D.^-1) * B';    % C^-1/2 
  eigeneval = 0;                      % track update of B and D
  chiN=N^0.5*(1-1/(4*N)+1/(21*N^2));  % expectation of 
                                      %   ||N(0,I)|| == norm(randn(N,1))
  
  % -------------------- Generation Loop --------------------------------
  counteval = 0;  % the next 40 lines contain the 20 lines of interesting code 
  while counteval < stopeval
    
      % Generate and evaluate lambda offspring
      for k=1:lambda,
          arx(:,k) = xmean + sigma * B * (D .* randn(N,1)); % m + sig * Normal(0,C) 
          arfitness(k) = feval(strfitnessfct, arx(:,k)); % objective function call
          counteval = counteval+1;
      end
    
      % Sort by fitness and compute weighted mean into xmean
      [arfitness, arindex] = sort(arfitness); % minimization
      xold = xmean;
      xmean = arx(:,arindex(1:mu))*weights;   % recombination, new mean value
    
      % Cumulation: Update evolution paths
      ps = (1-cs)*ps ... 
            + sqrt(cs*(2-cs)*mueff) * invsqrtC * (xmean-xold) / sigma; 
      hsig = norm(ps)/sqrt(1-(1-cs)^(2*counteval/lambda))/chiN < 1.4 + 2/(N+1);
      pc = (1-cc)*pc ...
            + hsig * sqrt(cc*(2-cc)*mueff) * (xmean-xold) / sigma;

      % Adapt covariance matrix C
      artmp = (1/sigma) * (arx(:,arindex(1:mu))-repmat(xold,1,mu));
      C = (1-c1-cmu) * C ...                  % regard old matrix  
           + c1 * (pc*pc' ...                 % plus rank one update
                   + (1-hsig) * cc*(2-cc) * C) ... % minor correction if hsig==0
           + cmu * artmp * diag(weights) * artmp'; % plus rank mu update

      % Adapt step size sigma
      sigma = sigma * exp((cs/damps)*(norm(ps)/chiN - 1)); 
    
      % Decomposition of C into B*diag(D.^2)*B' (diagonalization)
      if counteval - eigeneval > lambda/(c1+cmu)/N/10  % to achieve O(N^2)
          eigeneval = counteval;
          C = triu(C) + triu(C,1)'; % enforce symmetry
          [B,D] = eig(C);           % eigen decomposition, B==normalized eigenvectors
          D = sqrt(diag(D));        % D is a vector of standard deviations now
          invsqrtC = B * diag(D.^-1) * B';
      end
    
      % Break, if fitness is good enough or condition exceeds 1e14, better termination methods are advisable 
      if arfitness(1) <= stopfitness || max(D) > 1e7 * min(D)
          break;
      end

  end % while, end generation loop

  xmin = arx(:, arindex(1)); % Return best point of last iteration.
                             % Notice that xmean is expected to be even
                             % better.
  
% ---------------------------------------------------------------  
function f=frosenbrock(x)
    if size(x,1) < 2 error('dimension must be greater one'); end
    f = 100*sum((x(1:end-1).^2 - x(2:end)).^2) + sum((x(1:end-1)-1).^2);

Theoretical Foundations

Given the distribution parameters—mean, variances and covariances—the normal probability distribution for sampling new candidate solutions is the maximum entropy probability distribution over , that is, the sample distribution with the minimal amount of prior information built into the distribution. More considerations on the update equations of CMA-ES are made in the following.

Variable Metric

The CMA-ES implements a stochastic variable-metric method. In the very particular case of a convex-quadratic objective function

the covariance matrix adapts to the inverse of the Hessian matrix , up to a scalar factor and small random fluctuations. More general, also on the function , where is strictly increasing and therefore order preserving and is convex-quadratic, the covariance matrix adapts to , up to a scalar factor and small random fluctuations.

Maximum-Likelihood Updates

The update equations for mean and covariance matrix maximize a likelihood while resembling an expectation-maximization algorithm. The update of the mean vector maximizes a log-likelihood, such that

where

denotes the log-likelihood of from a multivariate normal distribution with mean and any positive definite covariance matrix . To see that is independent of remark first that this is the case for any diagonal matrix , because the coordinate-wise maximizer is independent of a scaling factor. Then, rotation of the data points or choosing non-diagonal are equivalent.

The rank- update of the covariance matrix, that is, the right most summand in the update equation of , maximizes a log-likelihood in that

for (otherwise is singular, but substantially the same result holds for ). Here, denotes the likelihood of from a multivariate normal distribution with zero mean and covariance matrix . Therefore, for and , is the above maximum-likelihood estimator. See estimation of covariance matrices for details on the derivation.

Natural Gradient Descent in the Space of Sample Distributions

Akimoto et al.[3] recently found that the update of the distribution parameters resembles the descend in direction of a sampled natural gradient of the expected objective function value Buying, selling and renting HDB and personal residential properties in Singapore are simple and transparent transactions. Although you are not required to engage a real property salesperson (generally often known as a "public listed property developers In singapore agent") to complete these property transactions, chances are you'll think about partaking one if you are not accustomed to the processes concerned.

Professional agents are readily available once you need to discover an condominium for hire in singapore In some cases, landlords will take into account you more favourably in case your agent comes to them than for those who tried to method them by yourself. You need to be careful, nevertheless, as you resolve in your agent. Ensure that the agent you are contemplating working with is registered with the IEA – Institute of Estate Brokers. Whereas it might sound a hassle to you, will probably be worth it in the end. The IEA works by an ordinary algorithm and regulations, so you'll protect yourself in opposition to probably going with a rogue agent who prices you more than they should for his or her service in finding you an residence for lease in singapore.

There isn't any deal too small. Property agents who are keen to find time for any deal even if the commission is small are the ones you want on your aspect. Additionally they present humbleness and might relate with the typical Singaporean higher. Relentlessly pursuing any deal, calling prospects even without being prompted. Even if they get rejected a hundred times, they still come again for more. These are the property brokers who will find consumers what they need eventually, and who would be the most successful in what they do. 4. Honesty and Integrity

This feature is suitable for you who need to get the tax deductions out of your PIC scheme to your property agency firm. It's endorsed that you visit the correct site for filling this tax return software. This utility must be submitted at the very least yearly to report your whole tax and tax return that you're going to receive in the current accounting 12 months. There may be an official website for this tax filling procedure. Filling this tax return software shouldn't be a tough thing to do for all business homeowners in Singapore.

A wholly owned subsidiary of SLP Worldwide, SLP Realty houses 900 associates to service SLP's fast rising portfolio of residential tasks. Real estate is a human-centric trade. Apart from offering comprehensive coaching applications for our associates, SLP Realty puts equal emphasis on creating human capabilities and creating sturdy teamwork throughout all ranges of our organisational hierarchy. Worldwide Presence At SLP International, our staff of execs is pushed to make sure our shoppers meet their enterprise and investment targets. Under is an inventory of some notable shoppers from completely different industries and markets, who've entrusted their real estate must the expertise of SLP Worldwide.

If you're looking for a real estate or Singapore property agent online, you merely need to belief your instinct. It is because you don't know which agent is sweet and which agent will not be. Carry out research on a number of brokers by looking out the internet. As soon as if you find yourself certain that a selected agent is dependable and trustworthy, you'll be able to choose to utilize his partnerise find you a house in Singapore. More often than not, a property agent is considered to be good if she or he places the contact data on his web site. This is able to imply that the agent does not thoughts you calling them and asking them any questions regarding properties in Singapore. After chatting with them you too can see them of their office after taking an appointment.

Another method by way of which you could find out whether the agent is sweet is by checking the feedback, of the shoppers, on the website. There are various individuals would publish their comments on the web site of the Singapore property agent. You can take a look at these feedback and the see whether it will be clever to hire that specific Singapore property agent. You may even get in contact with the developer immediately. Many Singapore property brokers know the developers and you may confirm the goodwill of the agent by asking the developer. (to be minimized), where the expectation is taken under the sample distribution. With the parameter setting of and , i.e. without step-size control and rank-one update, CMA-ES can thus be viewed as an instantiation of Natural Evolution Strategies (NES).[3][4] The natural gradient is independent of the parameterization of the distribution. Taken with respect to the parameters Buying, selling and renting HDB and personal residential properties in Singapore are simple and transparent transactions. Although you are not required to engage a real property salesperson (generally often known as a "public listed property developers In singapore agent") to complete these property transactions, chances are you'll think about partaking one if you are not accustomed to the processes concerned.

Professional agents are readily available once you need to discover an condominium for hire in singapore In some cases, landlords will take into account you more favourably in case your agent comes to them than for those who tried to method them by yourself. You need to be careful, nevertheless, as you resolve in your agent. Ensure that the agent you are contemplating working with is registered with the IEA – Institute of Estate Brokers. Whereas it might sound a hassle to you, will probably be worth it in the end. The IEA works by an ordinary algorithm and regulations, so you'll protect yourself in opposition to probably going with a rogue agent who prices you more than they should for his or her service in finding you an residence for lease in singapore.

There isn't any deal too small. Property agents who are keen to find time for any deal even if the commission is small are the ones you want on your aspect. Additionally they present humbleness and might relate with the typical Singaporean higher. Relentlessly pursuing any deal, calling prospects even without being prompted. Even if they get rejected a hundred times, they still come again for more. These are the property brokers who will find consumers what they need eventually, and who would be the most successful in what they do. 4. Honesty and Integrity

This feature is suitable for you who need to get the tax deductions out of your PIC scheme to your property agency firm. It's endorsed that you visit the correct site for filling this tax return software. This utility must be submitted at the very least yearly to report your whole tax and tax return that you're going to receive in the current accounting 12 months. There may be an official website for this tax filling procedure. Filling this tax return software shouldn't be a tough thing to do for all business homeowners in Singapore.

A wholly owned subsidiary of SLP Worldwide, SLP Realty houses 900 associates to service SLP's fast rising portfolio of residential tasks. Real estate is a human-centric trade. Apart from offering comprehensive coaching applications for our associates, SLP Realty puts equal emphasis on creating human capabilities and creating sturdy teamwork throughout all ranges of our organisational hierarchy. Worldwide Presence At SLP International, our staff of execs is pushed to make sure our shoppers meet their enterprise and investment targets. Under is an inventory of some notable shoppers from completely different industries and markets, who've entrusted their real estate must the expertise of SLP Worldwide.

If you're looking for a real estate or Singapore property agent online, you merely need to belief your instinct. It is because you don't know which agent is sweet and which agent will not be. Carry out research on a number of brokers by looking out the internet. As soon as if you find yourself certain that a selected agent is dependable and trustworthy, you'll be able to choose to utilize his partnerise find you a house in Singapore. More often than not, a property agent is considered to be good if she or he places the contact data on his web site. This is able to imply that the agent does not thoughts you calling them and asking them any questions regarding properties in Singapore. After chatting with them you too can see them of their office after taking an appointment.

Another method by way of which you could find out whether the agent is sweet is by checking the feedback, of the shoppers, on the website. There are various individuals would publish their comments on the web site of the Singapore property agent. You can take a look at these feedback and the see whether it will be clever to hire that specific Singapore property agent. You may even get in contact with the developer immediately. Many Singapore property brokers know the developers and you may confirm the goodwill of the agent by asking the developer. of the sample distribution Buying, selling and renting HDB and personal residential properties in Singapore are simple and transparent transactions. Although you are not required to engage a real property salesperson (generally often known as a "public listed property developers In singapore agent") to complete these property transactions, chances are you'll think about partaking one if you are not accustomed to the processes concerned.

Professional agents are readily available once you need to discover an condominium for hire in singapore In some cases, landlords will take into account you more favourably in case your agent comes to them than for those who tried to method them by yourself. You need to be careful, nevertheless, as you resolve in your agent. Ensure that the agent you are contemplating working with is registered with the IEA – Institute of Estate Brokers. Whereas it might sound a hassle to you, will probably be worth it in the end. The IEA works by an ordinary algorithm and regulations, so you'll protect yourself in opposition to probably going with a rogue agent who prices you more than they should for his or her service in finding you an residence for lease in singapore.

There isn't any deal too small. Property agents who are keen to find time for any deal even if the commission is small are the ones you want on your aspect. Additionally they present humbleness and might relate with the typical Singaporean higher. Relentlessly pursuing any deal, calling prospects even without being prompted. Even if they get rejected a hundred times, they still come again for more. These are the property brokers who will find consumers what they need eventually, and who would be the most successful in what they do. 4. Honesty and Integrity

This feature is suitable for you who need to get the tax deductions out of your PIC scheme to your property agency firm. It's endorsed that you visit the correct site for filling this tax return software. This utility must be submitted at the very least yearly to report your whole tax and tax return that you're going to receive in the current accounting 12 months. There may be an official website for this tax filling procedure. Filling this tax return software shouldn't be a tough thing to do for all business homeowners in Singapore.

A wholly owned subsidiary of SLP Worldwide, SLP Realty houses 900 associates to service SLP's fast rising portfolio of residential tasks. Real estate is a human-centric trade. Apart from offering comprehensive coaching applications for our associates, SLP Realty puts equal emphasis on creating human capabilities and creating sturdy teamwork throughout all ranges of our organisational hierarchy. Worldwide Presence At SLP International, our staff of execs is pushed to make sure our shoppers meet their enterprise and investment targets. Under is an inventory of some notable shoppers from completely different industries and markets, who've entrusted their real estate must the expertise of SLP Worldwide.

If you're looking for a real estate or Singapore property agent online, you merely need to belief your instinct. It is because you don't know which agent is sweet and which agent will not be. Carry out research on a number of brokers by looking out the internet. As soon as if you find yourself certain that a selected agent is dependable and trustworthy, you'll be able to choose to utilize his partnerise find you a house in Singapore. More often than not, a property agent is considered to be good if she or he places the contact data on his web site. This is able to imply that the agent does not thoughts you calling them and asking them any questions regarding properties in Singapore. After chatting with them you too can see them of their office after taking an appointment.

Another method by way of which you could find out whether the agent is sweet is by checking the feedback, of the shoppers, on the website. There are various individuals would publish their comments on the web site of the Singapore property agent. You can take a look at these feedback and the see whether it will be clever to hire that specific Singapore property agent. You may even get in contact with the developer immediately. Many Singapore property brokers know the developers and you may confirm the goodwill of the agent by asking the developer., the gradient of Buying, selling and renting HDB and personal residential properties in Singapore are simple and transparent transactions. Although you are not required to engage a real property salesperson (generally often known as a "public listed property developers In singapore agent") to complete these property transactions, chances are you'll think about partaking one if you are not accustomed to the processes concerned.

Professional agents are readily available once you need to discover an condominium for hire in singapore In some cases, landlords will take into account you more favourably in case your agent comes to them than for those who tried to method them by yourself. You need to be careful, nevertheless, as you resolve in your agent. Ensure that the agent you are contemplating working with is registered with the IEA – Institute of Estate Brokers. Whereas it might sound a hassle to you, will probably be worth it in the end. The IEA works by an ordinary algorithm and regulations, so you'll protect yourself in opposition to probably going with a rogue agent who prices you more than they should for his or her service in finding you an residence for lease in singapore.

There isn't any deal too small. Property agents who are keen to find time for any deal even if the commission is small are the ones you want on your aspect. Additionally they present humbleness and might relate with the typical Singaporean higher. Relentlessly pursuing any deal, calling prospects even without being prompted. Even if they get rejected a hundred times, they still come again for more. These are the property brokers who will find consumers what they need eventually, and who would be the most successful in what they do. 4. Honesty and Integrity

This feature is suitable for you who need to get the tax deductions out of your PIC scheme to your property agency firm. It's endorsed that you visit the correct site for filling this tax return software. This utility must be submitted at the very least yearly to report your whole tax and tax return that you're going to receive in the current accounting 12 months. There may be an official website for this tax filling procedure. Filling this tax return software shouldn't be a tough thing to do for all business homeowners in Singapore.

A wholly owned subsidiary of SLP Worldwide, SLP Realty houses 900 associates to service SLP's fast rising portfolio of residential tasks. Real estate is a human-centric trade. Apart from offering comprehensive coaching applications for our associates, SLP Realty puts equal emphasis on creating human capabilities and creating sturdy teamwork throughout all ranges of our organisational hierarchy. Worldwide Presence At SLP International, our staff of execs is pushed to make sure our shoppers meet their enterprise and investment targets. Under is an inventory of some notable shoppers from completely different industries and markets, who've entrusted their real estate must the expertise of SLP Worldwide.

If you're looking for a real estate or Singapore property agent online, you merely need to belief your instinct. It is because you don't know which agent is sweet and which agent will not be. Carry out research on a number of brokers by looking out the internet. As soon as if you find yourself certain that a selected agent is dependable and trustworthy, you'll be able to choose to utilize his partnerise find you a house in Singapore. More often than not, a property agent is considered to be good if she or he places the contact data on his web site. This is able to imply that the agent does not thoughts you calling them and asking them any questions regarding properties in Singapore. After chatting with them you too can see them of their office after taking an appointment.

Another method by way of which you could find out whether the agent is sweet is by checking the feedback, of the shoppers, on the website. There are various individuals would publish their comments on the web site of the Singapore property agent. You can take a look at these feedback and the see whether it will be clever to hire that specific Singapore property agent. You may even get in contact with the developer immediately. Many Singapore property brokers know the developers and you may confirm the goodwill of the agent by asking the developer. can be expressed as

where depends on the parameter vector , the so-called score function, , indicates the relative sensitivity of Buying, selling and renting HDB and personal residential properties in Singapore are simple and transparent transactions. Although you are not required to engage a real property salesperson (generally often known as a "public listed property developers In singapore agent") to complete these property transactions, chances are you'll think about partaking one if you are not accustomed to the processes concerned.

Professional agents are readily available once you need to discover an condominium for hire in singapore In some cases, landlords will take into account you more favourably in case your agent comes to them than for those who tried to method them by yourself. You need to be careful, nevertheless, as you resolve in your agent. Ensure that the agent you are contemplating working with is registered with the IEA – Institute of Estate Brokers. Whereas it might sound a hassle to you, will probably be worth it in the end. The IEA works by an ordinary algorithm and regulations, so you'll protect yourself in opposition to probably going with a rogue agent who prices you more than they should for his or her service in finding you an residence for lease in singapore.

There isn't any deal too small. Property agents who are keen to find time for any deal even if the commission is small are the ones you want on your aspect. Additionally they present humbleness and might relate with the typical Singaporean higher. Relentlessly pursuing any deal, calling prospects even without being prompted. Even if they get rejected a hundred times, they still come again for more. These are the property brokers who will find consumers what they need eventually, and who would be the most successful in what they do. 4. Honesty and Integrity

This feature is suitable for you who need to get the tax deductions out of your PIC scheme to your property agency firm. It's endorsed that you visit the correct site for filling this tax return software. This utility must be submitted at the very least yearly to report your whole tax and tax return that you're going to receive in the current accounting 12 months. There may be an official website for this tax filling procedure. Filling this tax return software shouldn't be a tough thing to do for all business homeowners in Singapore.

A wholly owned subsidiary of SLP Worldwide, SLP Realty houses 900 associates to service SLP's fast rising portfolio of residential tasks. Real estate is a human-centric trade. Apart from offering comprehensive coaching applications for our associates, SLP Realty puts equal emphasis on creating human capabilities and creating sturdy teamwork throughout all ranges of our organisational hierarchy. Worldwide Presence At SLP International, our staff of execs is pushed to make sure our shoppers meet their enterprise and investment targets. Under is an inventory of some notable shoppers from completely different industries and markets, who've entrusted their real estate must the expertise of SLP Worldwide.

If you're looking for a real estate or Singapore property agent online, you merely need to belief your instinct. It is because you don't know which agent is sweet and which agent will not be. Carry out research on a number of brokers by looking out the internet. As soon as if you find yourself certain that a selected agent is dependable and trustworthy, you'll be able to choose to utilize his partnerise find you a house in Singapore. More often than not, a property agent is considered to be good if she or he places the contact data on his web site. This is able to imply that the agent does not thoughts you calling them and asking them any questions regarding properties in Singapore. After chatting with them you too can see them of their office after taking an appointment.

Another method by way of which you could find out whether the agent is sweet is by checking the feedback, of the shoppers, on the website. There are various individuals would publish their comments on the web site of the Singapore property agent. You can take a look at these feedback and the see whether it will be clever to hire that specific Singapore property agent. You may even get in contact with the developer immediately. Many Singapore property brokers know the developers and you may confirm the goodwill of the agent by asking the developer. w.r.t. Buying, selling and renting HDB and personal residential properties in Singapore are simple and transparent transactions. Although you are not required to engage a real property salesperson (generally often known as a "public listed property developers In singapore agent") to complete these property transactions, chances are you'll think about partaking one if you are not accustomed to the processes concerned.

Professional agents are readily available once you need to discover an condominium for hire in singapore In some cases, landlords will take into account you more favourably in case your agent comes to them than for those who tried to method them by yourself. You need to be careful, nevertheless, as you resolve in your agent. Ensure that the agent you are contemplating working with is registered with the IEA – Institute of Estate Brokers. Whereas it might sound a hassle to you, will probably be worth it in the end. The IEA works by an ordinary algorithm and regulations, so you'll protect yourself in opposition to probably going with a rogue agent who prices you more than they should for his or her service in finding you an residence for lease in singapore.

There isn't any deal too small. Property agents who are keen to find time for any deal even if the commission is small are the ones you want on your aspect. Additionally they present humbleness and might relate with the typical Singaporean higher. Relentlessly pursuing any deal, calling prospects even without being prompted. Even if they get rejected a hundred times, they still come again for more. These are the property brokers who will find consumers what they need eventually, and who would be the most successful in what they do. 4. Honesty and Integrity

This feature is suitable for you who need to get the tax deductions out of your PIC scheme to your property agency firm. It's endorsed that you visit the correct site for filling this tax return software. This utility must be submitted at the very least yearly to report your whole tax and tax return that you're going to receive in the current accounting 12 months. There may be an official website for this tax filling procedure. Filling this tax return software shouldn't be a tough thing to do for all business homeowners in Singapore.

A wholly owned subsidiary of SLP Worldwide, SLP Realty houses 900 associates to service SLP's fast rising portfolio of residential tasks. Real estate is a human-centric trade. Apart from offering comprehensive coaching applications for our associates, SLP Realty puts equal emphasis on creating human capabilities and creating sturdy teamwork throughout all ranges of our organisational hierarchy. Worldwide Presence At SLP International, our staff of execs is pushed to make sure our shoppers meet their enterprise and investment targets. Under is an inventory of some notable shoppers from completely different industries and markets, who've entrusted their real estate must the expertise of SLP Worldwide.

If you're looking for a real estate or Singapore property agent online, you merely need to belief your instinct. It is because you don't know which agent is sweet and which agent will not be. Carry out research on a number of brokers by looking out the internet. As soon as if you find yourself certain that a selected agent is dependable and trustworthy, you'll be able to choose to utilize his partnerise find you a house in Singapore. More often than not, a property agent is considered to be good if she or he places the contact data on his web site. This is able to imply that the agent does not thoughts you calling them and asking them any questions regarding properties in Singapore. After chatting with them you too can see them of their office after taking an appointment.

Another method by way of which you could find out whether the agent is sweet is by checking the feedback, of the shoppers, on the website. There are various individuals would publish their comments on the web site of the Singapore property agent. You can take a look at these feedback and the see whether it will be clever to hire that specific Singapore property agent. You may even get in contact with the developer immediately. Many Singapore property brokers know the developers and you may confirm the goodwill of the agent by asking the developer., and the expectation is taken with respect to the distribution Buying, selling and renting HDB and personal residential properties in Singapore are simple and transparent transactions. Although you are not required to engage a real property salesperson (generally often known as a "public listed property developers In singapore agent") to complete these property transactions, chances are you'll think about partaking one if you are not accustomed to the processes concerned.

Professional agents are readily available once you need to discover an condominium for hire in singapore In some cases, landlords will take into account you more favourably in case your agent comes to them than for those who tried to method them by yourself. You need to be careful, nevertheless, as you resolve in your agent. Ensure that the agent you are contemplating working with is registered with the IEA – Institute of Estate Brokers. Whereas it might sound a hassle to you, will probably be worth it in the end. The IEA works by an ordinary algorithm and regulations, so you'll protect yourself in opposition to probably going with a rogue agent who prices you more than they should for his or her service in finding you an residence for lease in singapore.

There isn't any deal too small. Property agents who are keen to find time for any deal even if the commission is small are the ones you want on your aspect. Additionally they present humbleness and might relate with the typical Singaporean higher. Relentlessly pursuing any deal, calling prospects even without being prompted. Even if they get rejected a hundred times, they still come again for more. These are the property brokers who will find consumers what they need eventually, and who would be the most successful in what they do. 4. Honesty and Integrity

This feature is suitable for you who need to get the tax deductions out of your PIC scheme to your property agency firm. It's endorsed that you visit the correct site for filling this tax return software. This utility must be submitted at the very least yearly to report your whole tax and tax return that you're going to receive in the current accounting 12 months. There may be an official website for this tax filling procedure. Filling this tax return software shouldn't be a tough thing to do for all business homeowners in Singapore.

A wholly owned subsidiary of SLP Worldwide, SLP Realty houses 900 associates to service SLP's fast rising portfolio of residential tasks. Real estate is a human-centric trade. Apart from offering comprehensive coaching applications for our associates, SLP Realty puts equal emphasis on creating human capabilities and creating sturdy teamwork throughout all ranges of our organisational hierarchy. Worldwide Presence At SLP International, our staff of execs is pushed to make sure our shoppers meet their enterprise and investment targets. Under is an inventory of some notable shoppers from completely different industries and markets, who've entrusted their real estate must the expertise of SLP Worldwide.

If you're looking for a real estate or Singapore property agent online, you merely need to belief your instinct. It is because you don't know which agent is sweet and which agent will not be. Carry out research on a number of brokers by looking out the internet. As soon as if you find yourself certain that a selected agent is dependable and trustworthy, you'll be able to choose to utilize his partnerise find you a house in Singapore. More often than not, a property agent is considered to be good if she or he places the contact data on his web site. This is able to imply that the agent does not thoughts you calling them and asking them any questions regarding properties in Singapore. After chatting with them you too can see them of their office after taking an appointment.

Another method by way of which you could find out whether the agent is sweet is by checking the feedback, of the shoppers, on the website. There are various individuals would publish their comments on the web site of the Singapore property agent. You can take a look at these feedback and the see whether it will be clever to hire that specific Singapore property agent. You may even get in contact with the developer immediately. Many Singapore property brokers know the developers and you may confirm the goodwill of the agent by asking the developer.. The natural gradient of Buying, selling and renting HDB and personal residential properties in Singapore are simple and transparent transactions. Although you are not required to engage a real property salesperson (generally often known as a "public listed property developers In singapore agent") to complete these property transactions, chances are you'll think about partaking one if you are not accustomed to the processes concerned.

Professional agents are readily available once you need to discover an condominium for hire in singapore In some cases, landlords will take into account you more favourably in case your agent comes to them than for those who tried to method them by yourself. You need to be careful, nevertheless, as you resolve in your agent. Ensure that the agent you are contemplating working with is registered with the IEA – Institute of Estate Brokers. Whereas it might sound a hassle to you, will probably be worth it in the end. The IEA works by an ordinary algorithm and regulations, so you'll protect yourself in opposition to probably going with a rogue agent who prices you more than they should for his or her service in finding you an residence for lease in singapore.

There isn't any deal too small. Property agents who are keen to find time for any deal even if the commission is small are the ones you want on your aspect. Additionally they present humbleness and might relate with the typical Singaporean higher. Relentlessly pursuing any deal, calling prospects even without being prompted. Even if they get rejected a hundred times, they still come again for more. These are the property brokers who will find consumers what they need eventually, and who would be the most successful in what they do. 4. Honesty and Integrity

This feature is suitable for you who need to get the tax deductions out of your PIC scheme to your property agency firm. It's endorsed that you visit the correct site for filling this tax return software. This utility must be submitted at the very least yearly to report your whole tax and tax return that you're going to receive in the current accounting 12 months. There may be an official website for this tax filling procedure. Filling this tax return software shouldn't be a tough thing to do for all business homeowners in Singapore.

A wholly owned subsidiary of SLP Worldwide, SLP Realty houses 900 associates to service SLP's fast rising portfolio of residential tasks. Real estate is a human-centric trade. Apart from offering comprehensive coaching applications for our associates, SLP Realty puts equal emphasis on creating human capabilities and creating sturdy teamwork throughout all ranges of our organisational hierarchy. Worldwide Presence At SLP International, our staff of execs is pushed to make sure our shoppers meet their enterprise and investment targets. Under is an inventory of some notable shoppers from completely different industries and markets, who've entrusted their real estate must the expertise of SLP Worldwide.

If you're looking for a real estate or Singapore property agent online, you merely need to belief your instinct. It is because you don't know which agent is sweet and which agent will not be. Carry out research on a number of brokers by looking out the internet. As soon as if you find yourself certain that a selected agent is dependable and trustworthy, you'll be able to choose to utilize his partnerise find you a house in Singapore. More often than not, a property agent is considered to be good if she or he places the contact data on his web site. This is able to imply that the agent does not thoughts you calling them and asking them any questions regarding properties in Singapore. After chatting with them you too can see them of their office after taking an appointment.

Another method by way of which you could find out whether the agent is sweet is by checking the feedback, of the shoppers, on the website. There are various individuals would publish their comments on the web site of the Singapore property agent. You can take a look at these feedback and the see whether it will be clever to hire that specific Singapore property agent. You may even get in contact with the developer immediately. Many Singapore property brokers know the developers and you may confirm the goodwill of the agent by asking the developer., complying with the Fisher information metric (an informational distance measure between probability distributions and the curvature of the relative entropy), now reads

where the Fisher information matrix is the expectation of the Hessian of Buying, selling and renting HDB and personal residential properties in Singapore are simple and transparent transactions. Although you are not required to engage a real property salesperson (generally often known as a "public listed property developers In singapore agent") to complete these property transactions, chances are you'll think about partaking one if you are not accustomed to the processes concerned.

Professional agents are readily available once you need to discover an condominium for hire in singapore In some cases, landlords will take into account you more favourably in case your agent comes to them than for those who tried to method them by yourself. You need to be careful, nevertheless, as you resolve in your agent. Ensure that the agent you are contemplating working with is registered with the IEA – Institute of Estate Brokers. Whereas it might sound a hassle to you, will probably be worth it in the end. The IEA works by an ordinary algorithm and regulations, so you'll protect yourself in opposition to probably going with a rogue agent who prices you more than they should for his or her service in finding you an residence for lease in singapore.

There isn't any deal too small. Property agents who are keen to find time for any deal even if the commission is small are the ones you want on your aspect. Additionally they present humbleness and might relate with the typical Singaporean higher. Relentlessly pursuing any deal, calling prospects even without being prompted. Even if they get rejected a hundred times, they still come again for more. These are the property brokers who will find consumers what they need eventually, and who would be the most successful in what they do. 4. Honesty and Integrity

This feature is suitable for you who need to get the tax deductions out of your PIC scheme to your property agency firm. It's endorsed that you visit the correct site for filling this tax return software. This utility must be submitted at the very least yearly to report your whole tax and tax return that you're going to receive in the current accounting 12 months. There may be an official website for this tax filling procedure. Filling this tax return software shouldn't be a tough thing to do for all business homeowners in Singapore.

A wholly owned subsidiary of SLP Worldwide, SLP Realty houses 900 associates to service SLP's fast rising portfolio of residential tasks. Real estate is a human-centric trade. Apart from offering comprehensive coaching applications for our associates, SLP Realty puts equal emphasis on creating human capabilities and creating sturdy teamwork throughout all ranges of our organisational hierarchy. Worldwide Presence At SLP International, our staff of execs is pushed to make sure our shoppers meet their enterprise and investment targets. Under is an inventory of some notable shoppers from completely different industries and markets, who've entrusted their real estate must the expertise of SLP Worldwide.

If you're looking for a real estate or Singapore property agent online, you merely need to belief your instinct. It is because you don't know which agent is sweet and which agent will not be. Carry out research on a number of brokers by looking out the internet. As soon as if you find yourself certain that a selected agent is dependable and trustworthy, you'll be able to choose to utilize his partnerise find you a house in Singapore. More often than not, a property agent is considered to be good if she or he places the contact data on his web site. This is able to imply that the agent does not thoughts you calling them and asking them any questions regarding properties in Singapore. After chatting with them you too can see them of their office after taking an appointment.

Another method by way of which you could find out whether the agent is sweet is by checking the feedback, of the shoppers, on the website. There are various individuals would publish their comments on the web site of the Singapore property agent. You can take a look at these feedback and the see whether it will be clever to hire that specific Singapore property agent. You may even get in contact with the developer immediately. Many Singapore property brokers know the developers and you may confirm the goodwill of the agent by asking the developer. and renders the expression independent of the chosen parameterization. Combining the previous equalities we get

A Monte Carlo approximation of the latter expectation takes the average over Buying, selling and renting HDB and personal residential properties in Singapore are simple and transparent transactions. Although you are not required to engage a real property salesperson (generally often known as a "public listed property developers In singapore agent") to complete these property transactions, chances are you'll think about partaking one if you are not accustomed to the processes concerned.

Professional agents are readily available once you need to discover an condominium for hire in singapore In some cases, landlords will take into account you more favourably in case your agent comes to them than for those who tried to method them by yourself. You need to be careful, nevertheless, as you resolve in your agent. Ensure that the agent you are contemplating working with is registered with the IEA – Institute of Estate Brokers. Whereas it might sound a hassle to you, will probably be worth it in the end. The IEA works by an ordinary algorithm and regulations, so you'll protect yourself in opposition to probably going with a rogue agent who prices you more than they should for his or her service in finding you an residence for lease in singapore.

There isn't any deal too small. Property agents who are keen to find time for any deal even if the commission is small are the ones you want on your aspect. Additionally they present humbleness and might relate with the typical Singaporean higher. Relentlessly pursuing any deal, calling prospects even without being prompted. Even if they get rejected a hundred times, they still come again for more. These are the property brokers who will find consumers what they need eventually, and who would be the most successful in what they do. 4. Honesty and Integrity

This feature is suitable for you who need to get the tax deductions out of your PIC scheme to your property agency firm. It's endorsed that you visit the correct site for filling this tax return software. This utility must be submitted at the very least yearly to report your whole tax and tax return that you're going to receive in the current accounting 12 months. There may be an official website for this tax filling procedure. Filling this tax return software shouldn't be a tough thing to do for all business homeowners in Singapore.

A wholly owned subsidiary of SLP Worldwide, SLP Realty houses 900 associates to service SLP's fast rising portfolio of residential tasks. Real estate is a human-centric trade. Apart from offering comprehensive coaching applications for our associates, SLP Realty puts equal emphasis on creating human capabilities and creating sturdy teamwork throughout all ranges of our organisational hierarchy. Worldwide Presence At SLP International, our staff of execs is pushed to make sure our shoppers meet their enterprise and investment targets. Under is an inventory of some notable shoppers from completely different industries and markets, who've entrusted their real estate must the expertise of SLP Worldwide.

If you're looking for a real estate or Singapore property agent online, you merely need to belief your instinct. It is because you don't know which agent is sweet and which agent will not be. Carry out research on a number of brokers by looking out the internet. As soon as if you find yourself certain that a selected agent is dependable and trustworthy, you'll be able to choose to utilize his partnerise find you a house in Singapore. More often than not, a property agent is considered to be good if she or he places the contact data on his web site. This is able to imply that the agent does not thoughts you calling them and asking them any questions regarding properties in Singapore. After chatting with them you too can see them of their office after taking an appointment.

Another method by way of which you could find out whether the agent is sweet is by checking the feedback, of the shoppers, on the website. There are various individuals would publish their comments on the web site of the Singapore property agent. You can take a look at these feedback and the see whether it will be clever to hire that specific Singapore property agent. You may even get in contact with the developer immediately. Many Singapore property brokers know the developers and you may confirm the goodwill of the agent by asking the developer. samples from Buying, selling and renting HDB and personal residential properties in Singapore are simple and transparent transactions. Although you are not required to engage a real property salesperson (generally often known as a "public listed property developers In singapore agent") to complete these property transactions, chances are you'll think about partaking one if you are not accustomed to the processes concerned.

Professional agents are readily available once you need to discover an condominium for hire in singapore In some cases, landlords will take into account you more favourably in case your agent comes to them than for those who tried to method them by yourself. You need to be careful, nevertheless, as you resolve in your agent. Ensure that the agent you are contemplating working with is registered with the IEA – Institute of Estate Brokers. Whereas it might sound a hassle to you, will probably be worth it in the end. The IEA works by an ordinary algorithm and regulations, so you'll protect yourself in opposition to probably going with a rogue agent who prices you more than they should for his or her service in finding you an residence for lease in singapore.

There isn't any deal too small. Property agents who are keen to find time for any deal even if the commission is small are the ones you want on your aspect. Additionally they present humbleness and might relate with the typical Singaporean higher. Relentlessly pursuing any deal, calling prospects even without being prompted. Even if they get rejected a hundred times, they still come again for more. These are the property brokers who will find consumers what they need eventually, and who would be the most successful in what they do. 4. Honesty and Integrity

This feature is suitable for you who need to get the tax deductions out of your PIC scheme to your property agency firm. It's endorsed that you visit the correct site for filling this tax return software. This utility must be submitted at the very least yearly to report your whole tax and tax return that you're going to receive in the current accounting 12 months. There may be an official website for this tax filling procedure. Filling this tax return software shouldn't be a tough thing to do for all business homeowners in Singapore.

A wholly owned subsidiary of SLP Worldwide, SLP Realty houses 900 associates to service SLP's fast rising portfolio of residential tasks. Real estate is a human-centric trade. Apart from offering comprehensive coaching applications for our associates, SLP Realty puts equal emphasis on creating human capabilities and creating sturdy teamwork throughout all ranges of our organisational hierarchy. Worldwide Presence At SLP International, our staff of execs is pushed to make sure our shoppers meet their enterprise and investment targets. Under is an inventory of some notable shoppers from completely different industries and markets, who've entrusted their real estate must the expertise of SLP Worldwide.

If you're looking for a real estate or Singapore property agent online, you merely need to belief your instinct. It is because you don't know which agent is sweet and which agent will not be. Carry out research on a number of brokers by looking out the internet. As soon as if you find yourself certain that a selected agent is dependable and trustworthy, you'll be able to choose to utilize his partnerise find you a house in Singapore. More often than not, a property agent is considered to be good if she or he places the contact data on his web site. This is able to imply that the agent does not thoughts you calling them and asking them any questions regarding properties in Singapore. After chatting with them you too can see them of their office after taking an appointment.

Another method by way of which you could find out whether the agent is sweet is by checking the feedback, of the shoppers, on the website. There are various individuals would publish their comments on the web site of the Singapore property agent. You can take a look at these feedback and the see whether it will be clever to hire that specific Singapore property agent. You may even get in contact with the developer immediately. Many Singapore property brokers know the developers and you may confirm the goodwill of the agent by asking the developer.

where the notation from above is used and therefore are monotonously decreasing in . We might use, for a more robust approximation, rather as defined in the CMA-ES and zero for Buying, selling and renting HDB and personal residential properties in Singapore are simple and transparent transactions. Although you are not required to engage a real property salesperson (generally often known as a "public listed property developers In singapore agent") to complete these property transactions, chances are you'll think about partaking one if you are not accustomed to the processes concerned.

Professional agents are readily available once you need to discover an condominium for hire in singapore In some cases, landlords will take into account you more favourably in case your agent comes to them than for those who tried to method them by yourself. You need to be careful, nevertheless, as you resolve in your agent. Ensure that the agent you are contemplating working with is registered with the IEA – Institute of Estate Brokers. Whereas it might sound a hassle to you, will probably be worth it in the end. The IEA works by an ordinary algorithm and regulations, so you'll protect yourself in opposition to probably going with a rogue agent who prices you more than they should for his or her service in finding you an residence for lease in singapore.

There isn't any deal too small. Property agents who are keen to find time for any deal even if the commission is small are the ones you want on your aspect. Additionally they present humbleness and might relate with the typical Singaporean higher. Relentlessly pursuing any deal, calling prospects even without being prompted. Even if they get rejected a hundred times, they still come again for more. These are the property brokers who will find consumers what they need eventually, and who would be the most successful in what they do. 4. Honesty and Integrity

This feature is suitable for you who need to get the tax deductions out of your PIC scheme to your property agency firm. It's endorsed that you visit the correct site for filling this tax return software. This utility must be submitted at the very least yearly to report your whole tax and tax return that you're going to receive in the current accounting 12 months. There may be an official website for this tax filling procedure. Filling this tax return software shouldn't be a tough thing to do for all business homeowners in Singapore.

A wholly owned subsidiary of SLP Worldwide, SLP Realty houses 900 associates to service SLP's fast rising portfolio of residential tasks. Real estate is a human-centric trade. Apart from offering comprehensive coaching applications for our associates, SLP Realty puts equal emphasis on creating human capabilities and creating sturdy teamwork throughout all ranges of our organisational hierarchy. Worldwide Presence At SLP International, our staff of execs is pushed to make sure our shoppers meet their enterprise and investment targets. Under is an inventory of some notable shoppers from completely different industries and markets, who've entrusted their real estate must the expertise of SLP Worldwide.

If you're looking for a real estate or Singapore property agent online, you merely need to belief your instinct. It is because you don't know which agent is sweet and which agent will not be. Carry out research on a number of brokers by looking out the internet. As soon as if you find yourself certain that a selected agent is dependable and trustworthy, you'll be able to choose to utilize his partnerise find you a house in Singapore. More often than not, a property agent is considered to be good if she or he places the contact data on his web site. This is able to imply that the agent does not thoughts you calling them and asking them any questions regarding properties in Singapore. After chatting with them you too can see them of their office after taking an appointment.

Another method by way of which you could find out whether the agent is sweet is by checking the feedback, of the shoppers, on the website. There are various individuals would publish their comments on the web site of the Singapore property agent. You can take a look at these feedback and the see whether it will be clever to hire that specific Singapore property agent. You may even get in contact with the developer immediately. Many Singapore property brokers know the developers and you may confirm the goodwill of the agent by asking the developer. (implementing a consistent estimator for the CDF of at the point composed with a fixed monotonous decreased transformation) and let

such that is the density of the multivariate normal distribution . Then, we have an explicit expression for the inverse of the Fisher information matrix where is fixed

and for

and, after some calculations, the updates in the CMA-ES turn out as[3]

and

where mat forms the proper matrix from the respective natural gradient sub-vector. That means, setting , the CMA-ES updates descend in direction of the approximation of the natural gradient while using different step-sizes (learning rates) for the orthogonal parameters and respectively.

Stationarity or Unbiasedness

It is comparatively easy to see that the update equations of CMA-ES satisfy some stationarity conditions, in that they are essentially unbiased. Under neutral selection, where , we find that

and under some mild additional assumptions on the initial conditions

and with an additional minor correction in the covariance matrix update for the case where the indicator function evaluates to zero, we find

Invariance

Invariance properties imply uniform performance on a class of objective functions. They have been argued to be an advantage, because they allow to generalize and predict the behavior of the algorithm and therefore strengthen the meaning of empirical results obtained on single functions. The following invariance properties have been established for CMA-ES.

Any serious parameter optimization method should be translation invariant, but most methods do not exhibit all the above described invariance properties. A prominent example with the same invariance properties is the Nelder–Mead method, where the initial simplex must be chosen respectively.

Convergence

Conceptual considerations like the scale-invariance property of the algorithm, the analysis of simpler evolution strategies, and overwhelming empirical evidence suggest that the algorithm converges on a large class of functions fast to the global optimum, denoted as . On some functions, convergence occurs independently of the initial conditions with probability one. On some functions the probability is smaller than one and typically depends on the initial and . Empirically, the fastest possible convergence rate in for rank-based direct search methods can often be observed (depending on the context denoted as linear or log-linear or exponential convergence). Informally, we can write

for some , and more rigorously

or similarly,

This means that on average the distance to the optimum decreases in each iteration by a "constant" factor, namely by . The convergence rate is roughly , given is not much larger than the dimension . Even with optimal and , the convergence rate cannot largely exceed , given the above recombination weights are all non-negative. The actual linear dependencies in and are remarkable and they are in both cases the best one can hope for in this kind of algorithm. Yet, a rigorous proof of convergence is missing.

Interpretation as Coordinate System Transformation

Using a non-identity covariance matrix for the multivariate normal distribution in evolution strategies is equivalent to a coordinate system transformation of the solution vectors,[5] mainly because the sampling equation

can be equivalently expressed in an "encoded space" as

The covariance matrix defines a bijective transformation (encoding) for all solution vectors into a space, where the sampling takes place with identity covariance matrix. Because the update equations in the CMA-ES are invariant under linear coordinate system transformations, the CMA-ES can be re-written as an adaptive encoding procedure applied to a simple evolution strategy with identity covariance matrix.[5] This adaptive encoding procedure is not confined to algorithms that sample from a multivariate normal distribution (like evolution strategies), but can in principle be applied to any iterative search method.

Performance in Practice

In contrast to most other evolutionary algorithms, the CMA-ES is, from the users perspective, quasi parameter-free. However, the number of candidate samples λ (population size) can be adjusted by the user in order to change the characteristic search behavior (see above). CMA-ES has been empirically successful in hundreds of applications and is considered to be useful in particular on non-convex, non-separable, ill-conditioned, multi-modal or noisy objective functions. The search space dimension ranges typically between two and a few hundred. Assuming a black-box optimization scenario, where gradients are not available (or not useful) and function evaluations are the only considered cost of search, the CMA-ES method is likely to be outperformed by other methods in the following conditions:

  • on separable functions without or with only negligible dependencies between the design variables in particular in the case of multi-modality or large dimension, for example by differential evolution;

On separable functions, the performance disadvantage is likely to be most significant in that CMA-ES might not be able to find at all comparable solutions. On the other hand, on non-separable functions that are ill-conditioned or rugged or can only be solved with more than function evaluations, the CMA-ES shows most often superior performance.

Variations and Extensions

The (1+1)-CMA-ES[6] generates only one candidate solution per iteration step which becomes the new distribution mean if it is better than the current mean. For the (1+1)-CMA-ES is a close variant of Gaussian adaptation. Some Natural Evolution Strategies are close variants of the CMA-ES with specific parameter settings. Natural Evolution Strategies do not utilize evolution paths (that means in CMA-ES setting ) and they formalize the update of variances and covariances on a Cholesky factor instead of a covariance matrix. The CMA-ES has also been extended to multiobjective optimization as MO-CMA-ES.[7] Another remarkable extension has been the addition of a negative update of the covariance matrix with the so-called active CMA.[8]

With the advent of niching methods in evolutionary strategies, the question of an optimal niche radius arises. An "adaptive individual niche radius" is introduced in [9]

See also

References

43 year old Petroleum Engineer Harry from Deep River, usually spends time with hobbies and interests like renting movies, property developers in singapore new condominium and vehicle racing. Constantly enjoys going to destinations like Camino Real de Tierra Adentro.

Bibliography

  • Hansen N, Ostermeier A (2001). Completely derandomized self-adaptation in evolution strategies. Evolutionary Computation, 9(2) pp. 159–195. [1]
  • Hansen N, Müller SD, Koumoutsakos P (2003). Reducing the time complexity of the derandomized evolution strategy with covariance matrix adaptation (CMA-ES). Evolutionary Computation, 11(1) pp. 1–18. [2]
  • Hansen N, Kern S (2004). Evaluating the CMA evolution strategy on multimodal test functions. In Xin Yao et al., editors, Parallel Problem Solving from Nature - PPSN VIII, pp. 282–291, Springer. [3]
  • Igel C, Hansen N, Roth S (2007). Covariance Matrix Adaptation for Multi-objective Optimization. Evolutionary Computation, 15(1) pp. 1–28. [4]

External links

Template:Major subfields of optimization

fr:Stratégie d'évolution#CMA-ES

  1. Many property agents need to declare for the PIC grant in Singapore. However, not all of them know find out how to do the correct process for getting this PIC scheme from the IRAS. There are a number of steps that you need to do before your software can be approved.

    Naturally, you will have to pay a safety deposit and that is usually one month rent for annually of the settlement. That is the place your good religion deposit will likely be taken into account and will kind part or all of your security deposit. Anticipate to have a proportionate amount deducted out of your deposit if something is discovered to be damaged if you move out. It's best to you'll want to test the inventory drawn up by the owner, which can detail all objects in the property and their condition. If you happen to fail to notice any harm not already mentioned within the inventory before transferring in, you danger having to pay for it yourself.

    In case you are in search of an actual estate or Singapore property agent on-line, you simply should belief your intuition. It's because you do not know which agent is nice and which agent will not be. Carry out research on several brokers by looking out the internet. As soon as if you end up positive that a selected agent is dependable and reliable, you can choose to utilize his partnerise in finding you a home in Singapore. Most of the time, a property agent is taken into account to be good if he or she locations the contact data on his website. This may mean that the agent does not mind you calling them and asking them any questions relating to new properties in singapore in Singapore. After chatting with them you too can see them in their office after taking an appointment.

    Have handed an trade examination i.e Widespread Examination for House Brokers (CEHA) or Actual Property Agency (REA) examination, or equal; Exclusive brokers are extra keen to share listing information thus making certain the widest doable coverage inside the real estate community via Multiple Listings and Networking. Accepting a severe provide is simpler since your agent is totally conscious of all advertising activity related with your property. This reduces your having to check with a number of agents for some other offers. Price control is easily achieved. Paint work in good restore-discuss with your Property Marketing consultant if main works are still to be done. Softening in residential property prices proceed, led by 2.8 per cent decline within the index for Remainder of Central Region

    Once you place down the one per cent choice price to carry down a non-public property, it's important to accept its situation as it is whenever you move in – faulty air-con, choked rest room and all. Get round this by asking your agent to incorporate a ultimate inspection clause within the possibility-to-buy letter. HDB flat patrons routinely take pleasure in this security net. "There's a ultimate inspection of the property two days before the completion of all HDB transactions. If the air-con is defective, you can request the seller to repair it," says Kelvin.

    15.6.1 As the agent is an intermediary, generally, as soon as the principal and third party are introduced right into a contractual relationship, the agent drops out of the image, subject to any problems with remuneration or indemnification that he could have against the principal, and extra exceptionally, against the third occasion. Generally, agents are entitled to be indemnified for all liabilities reasonably incurred within the execution of the brokers´ authority.

    To achieve the very best outcomes, you must be always updated on market situations, including past transaction information and reliable projections. You could review and examine comparable homes that are currently available in the market, especially these which have been sold or not bought up to now six months. You'll be able to see a pattern of such report by clicking here It's essential to defend yourself in opposition to unscrupulous patrons. They are often very skilled in using highly unethical and manipulative techniques to try and lure you into a lure. That you must also protect your self, your loved ones, and personal belongings as you'll be serving many strangers in your home. Sign a listing itemizing of all of the objects provided by the proprietor, together with their situation. HSR Prime Recruiter 2010
  2. 55 years old Systems Administrator Antony from Clarence Creek, really loves learning, PC Software and aerobics. Likes to travel and was inspired after making a journey to Historic Ensemble of the Potala Palace.

    You can view that web-site... ccleaner free download
  3. 3.0 3.1 3.2 55 years old Systems Administrator Antony from Clarence Creek, really loves learning, PC Software and aerobics. Likes to travel and was inspired after making a journey to Historic Ensemble of the Potala Palace.

    You can view that web-site... ccleaner free download
  4. 55 years old Systems Administrator Antony from Clarence Creek, really loves learning, PC Software and aerobics. Likes to travel and was inspired after making a journey to Historic Ensemble of the Potala Palace.

    You can view that web-site... ccleaner free download
  5. 5.0 5.1 55 years old Systems Administrator Antony from Clarence Creek, really loves learning, PC Software and aerobics. Likes to travel and was inspired after making a journey to Historic Ensemble of the Potala Palace.

    You can view that web-site... ccleaner free download
  6. 55 years old Systems Administrator Antony from Clarence Creek, really loves learning, PC Software and aerobics. Likes to travel and was inspired after making a journey to Historic Ensemble of the Potala Palace.

    You can view that web-site... ccleaner free download
  7. One of the biggest reasons investing in a Singapore new launch is an effective things is as a result of it is doable to be lent massive quantities of money at very low interest rates that you should utilize to purchase it. Then, if property values continue to go up, then you'll get a really high return on funding (ROI). Simply make sure you purchase one of the higher properties, reminiscent of the ones at Fernvale the Riverbank or any Singapore landed property Get Earnings by means of Renting

    In its statement, the singapore property listing - website link, government claimed that the majority citizens buying their first residence won't be hurt by the new measures. Some concessions can even be prolonged to chose teams of consumers, similar to married couples with a minimum of one Singaporean partner who are purchasing their second property so long as they intend to promote their first residential property. Lower the LTV limit on housing loans granted by monetary establishments regulated by MAS from 70% to 60% for property purchasers who are individuals with a number of outstanding housing loans on the time of the brand new housing purchase. Singapore Property Measures - 30 August 2010 The most popular seek for the number of bedrooms in Singapore is 4, followed by 2 and three. Lush Acres EC @ Sengkang

    Discover out more about real estate funding in the area, together with info on international funding incentives and property possession. Many Singaporeans have been investing in property across the causeway in recent years, attracted by comparatively low prices. However, those who need to exit their investments quickly are likely to face significant challenges when trying to sell their property – and could finally be stuck with a property they can't sell. Career improvement programmes, in-house valuation, auctions and administrative help, venture advertising and marketing, skilled talks and traisning are continuously planned for the sales associates to help them obtain better outcomes for his or her shoppers while at Knight Frank Singapore. No change Present Rules

    Extending the tax exemption would help. The exemption, which may be as a lot as $2 million per family, covers individuals who negotiate a principal reduction on their existing mortgage, sell their house short (i.e., for lower than the excellent loans), or take part in a foreclosure course of. An extension of theexemption would seem like a common-sense means to assist stabilize the housing market, but the political turmoil around the fiscal-cliff negotiations means widespread sense could not win out. Home Minority Chief Nancy Pelosi (D-Calif.) believes that the mortgage relief provision will be on the table during the grand-cut price talks, in response to communications director Nadeam Elshami. Buying or promoting of blue mild bulbs is unlawful.

    A vendor's stamp duty has been launched on industrial property for the primary time, at rates ranging from 5 per cent to 15 per cent. The Authorities might be trying to reassure the market that they aren't in opposition to foreigners and PRs investing in Singapore's property market. They imposed these measures because of extenuating components available in the market." The sale of new dual-key EC models will even be restricted to multi-generational households only. The models have two separate entrances, permitting grandparents, for example, to dwell separately. The vendor's stamp obligation takes effect right this moment and applies to industrial property and plots which might be offered inside three years of the date of buy. JLL named Best Performing Property Brand for second year running

    The data offered is for normal info purposes only and isn't supposed to be personalised investment or monetary advice. Motley Fool Singapore contributor Stanley Lim would not personal shares in any corporations talked about. Singapore private home costs increased by 1.eight% within the fourth quarter of 2012, up from 0.6% within the earlier quarter. Resale prices of government-built HDB residences which are usually bought by Singaporeans, elevated by 2.5%, quarter on quarter, the quickest acquire in five quarters. And industrial property, prices are actually double the levels of three years ago. No withholding tax in the event you sell your property. All your local information regarding vital HDB policies, condominium launches, land growth, commercial property and more

    There are various methods to go about discovering the precise property. Some local newspapers (together with the Straits Instances ) have categorised property sections and many local property brokers have websites. Now there are some specifics to consider when buying a 'new launch' rental. Intended use of the unit Every sale begins with 10 p.c low cost for finish of season sale; changes to 20 % discount storewide; follows by additional reduction of fiftyand ends with last discount of 70 % or extra. Typically there is even a warehouse sale or transferring out sale with huge mark-down of costs for stock clearance. Deborah Regulation from Expat Realtor shares her property market update, plus prime rental residences and houses at the moment available to lease Esparina EC @ Sengkang
  8. 55 years old Systems Administrator Antony from Clarence Creek, really loves learning, PC Software and aerobics. Likes to travel and was inspired after making a journey to Historic Ensemble of the Potala Palace.

    You can view that web-site... ccleaner free download
  9. 20 year-old Real Estate Agent Rusty from Saint-Paul, has hobbies and interests which includes monopoly, property developers in singapore and poker. Will soon undertake a contiki trip that may include going to the Lower Valley of the Omo.

    My blog: http://www.primaboinca.com/view_profile.php?userid=5889534