R语言 Rmalschains包 Rmalschains-package()函数中文帮助文档(中英文对照)

loveR

Rmalschains-package(Rmalschains)
Rmalschains-package()所属R语言包：Rmalschains

                                        Getting started with the Rmalschains package
                                         入门的Rmalschains包

                                         译者：生物统计家园网 机器人LoveR

描述----------Description----------

This package implements an algorithm family for continuous optimization called memetic algorithms with local search chains
这个包实现的算法族称为“模因算法的不断优化与当地的搜索链


Details

详细信息----------Details----------

One of the main issues to optimize a real-coded function is the capability of the algorithm to realize a good exploration of the search space and, at the same time, to exploit the most promising region to obtain accurate solutions.
优化一个真正的编码的功能的主要问题之一是该算法的搜索空间，实现了良好的勘探，并在相同的时间，利用最有前途的区域，以获得准确的解决方案的能力。

Memetic algorithms are hybridizations of genetic algorithms with local search methods.  They are especially suited for continuous optimization, as they combine the power of evolutionary algorithms to explore the search space with a local search method to find the local optimum of a  promising region. In these algorithms, it is recommended to increase the effort invested in the  local search (measured in number of evaluations, called intensity) in the improvement of the  most promising solution. However, it is not easy to decide the right intensity for each solution.
文化基因算法是杂交遗传算法与局部搜索方法。特别适合于不断优化，因为它们进化算法结合起来，探索本地搜索的方法找到一个有前途的区域局部最优的搜索空间。在这些算法中，建议增加投资的最有希望的解决方案的改进中的局部搜索（测量评价的数量，被称为强度）的努力。然而，它是不容易为每个解决方案来决定正确的强度。

MA-LS-Chains is a steady-state memetic algorithm, which combines a steady-state genetic algorithm with various different local search methods. In contrast to the generational approach, where all  individuals are substituted in an iteration, in the steady-state genetic algorithm  in each iteration only one solution, the worst one, is subtituted in the population. This makes it possible to not lose the improvement obtained by the local search over the individuals.
MA-LS-链是一个稳态的Memetic算法，它结合了稳态遗传算法的各种不同的本地搜索方法。在对比度的代的方法，其中被取代所有个人在某次迭代中，在稳态的遗传算法在每次迭代中，只有一个解，最差的一个，是在人口subtituted。这使得有可能不失去在个人通过本地检索得到的改善。

For MA-LS-Chains, the current state of the local search algorithm is stored along with the individuals. So, it becomes possible to run the local search a fixed number of iterations, stop it, and possibly later continue the previous  local search over the same individual. In this way, MA-LS-Chains controls the application of the local search to the most promising solutions.
对于MA-LS-链，局部搜索算法的当前状态一起存储的个人。因此，它变得可以运行本地搜索一个固定次数的迭代，停止，并有可能在同一个人后继续以前的本地搜索。在这种方式中，MA-LS-链控制的局部搜索的最有前景的解决方案的应用程序。

The package implements various different local search strategies:
包实现各种不同的本地搜索战略：

CMA-ES The Covariance Matrix Adaptation Evolution Strategy
CMA-ES的协方差矩阵适应进化策略

SW A Solis Wets solver
SW A的索利斯才子解算器

SSW Subgrouping Solis Wets
西南的分集团索利斯才子

Simplex
单

CMA-ES is a very effective local search strategy, but quite complicated, and it does not scale well if the amount of  parameters to optimize is huge. The Solis Wets solver is pretty simple and therewith fast. The SSW strategy is an adapted version of the Solis Wets solver for high dimensional data, so that the algorithm with this type of local search scales well with the dimensionality of the data. It applies the Solis Wets solver to randomly chosen subgroups of  variables (Subgrouping Solis Wets).
CMA-ES是一种非常有效的局部搜索策略，但相当复杂的，它不很好，如果扩展量的参数，以优化是巨大的。索利斯的才子解算器是非常简单的，又用舌头快速的。西南的战略是改编版的索利斯才子解高维数据，因此，这种类型的本地搜索尺度的数据的维数的算法。它适用的的索利斯的才子解算器，以随机选取的变量分组（分集团索利斯才子）。

The package contains some demos illustrating its use. To get a list of them, type:
该软件包包含了一些演示说明其使用。为了得到它们的列表，键入：

library(Rmalschains)
library(Rmalschains)

demo()
demo()

The demos currently available are claw, rastrigin, sphere, and rastrigin_highDim. So in order to,  e.g., execute the claw demo, type
目前的演示是claw，rastrigin，sphere和rastrigin_highDim。因此，为了，例如，执行claw演示，类型

demo(claw)
demo(claw)

For theoretical background of the algorithm, the reader may refer to the cited literature, where the algorithms where originally proposed.
的理论背景的算法，读者可参考引用的文献，其中的算法最初提出。


（作者）----------Author(s)----------


Christoph Bergmeir <a href="mailto:c.bergmeir@decsai.ugr.es">c.bergmeir@decsai.ugr.es</a>

Daniel Molina <a href="mailto:dmolina@decsai.ugr.es">dmolina@decsai.ugr.es</a>

Jos茅 M. Ben铆tez <a href="mailto:j.m.benitez@decsai.ugr.es">j.m.benitez@decsai.ugr.es</a>

DiCITS Lab, Sci2s group, DECSAI, University of Granada. <a href="http://dicits.ugr.es">http://dicits.ugr.es</a>.

Additional information is also available at our group's website of continuous optimization:

<a href="http://sci2s.ugr.es/EAMHCO/">http://sci2s.ugr.es/EAMHCO/</a>



参考文献----------References----------

Memetic algorithms based on local search chains for large scale continuous optimisation problems: MA-SSW-Chains (2011) Soft Computing, 15 (11), pp. 2201-2220.
MA-SW-Chains: Memetic algorithm based on local search chains for large scale continuous global optimization (2010) 2010 IEEE World Congress on Computational Intelligence, WCCI 2010 - 2010 IEEE Congress on Evolutionary Computation, CEC 2010.
Memetic algorithms for continuous optimisation based on local search chains

参见----------See Also----------

malschains, malschains.control
malschains，malschains.control


实例----------Examples----------


#Example for maximization of the claw function[实施例的爪函数最大化]
##############################################[＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃]

claw <- function(xx) {
x <- xx[1]
y <- (0.46 * (dnorm(x, -1, 2/3) + dnorm(x, 1, 2/3)) +
(1/300) * (dnorm(x, -0.5, 0.01) + dnorm(x, -1,
0.01) + dnorm(x, -1.5, 0.01)) + (7/300) *
(dnorm(x, 0.5, 0.07) + dnorm(x, 1, 0.07) + dnorm(x,
1.5, 0.07)))
return(y)
}

res.claw <- malschains(function(x) {-claw(x)}, lower=c(-3), upper=c(3),
maxEvals=50000, control=malschains.control(popsize=50,
istep=300, ls="cmaes", optimum=-5))

x <- seq(-3, 3,length=1000)
claw_x <- NULL
for (i in 1:length(x)) claw_x[i] <- claw(x[i])

plot(x,claw_x, type="l")
points(res.claw$sol, -res.claw$fitness, col="red")


##############################################[＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃]
#Example for the rastrigin function[实例的rastrigin功能]
##############################################[＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃＃]

rastrigin <- function(x) {

dimension <- length(x)

res <- 0.0
for (i in 1:dimension) {
res <- res + (x[i]*x[i] - 10.0*cos(2.0*pi*x[i]) + 10.0)
}

res
}

res.rastrigin1 <- malschains(rastrigin, lower=seq(-1.0, -1.0, length=30),
upper=seq(1.0, 1.0, length=30), maxEvals=50000,
control=malschains.control(effort=0.8, alpha=0.3,
popsize=20, istep=100, ls="simplex"))


res.rastrigin2 <- malschains(rastrigin, lower=seq(-1.0, -1.0, length=30),
upper=seq(1.0, 1.0, length=30), maxEvals=50000,
initialpop = seq(0.1, 0.1, length=30),
control=malschains.control(popsize=50,
istep=300, ls="cmaes"))

res.rastrigin1

转载请注明:出自 生物统计家园网(http://www.biostatistic.net)。


注：
注1：为了方便大家学习，本文档为生物统计家园网机器人LoveR翻译而成，仅供个人R语言学习参考使用，生物统计家园保留版权。
注2：由于是机器人自动翻译，难免有不准确之处，使用时仔细对照中、英文内容进行反复理解，可以帮助R语言的学习。
注3：如遇到不准确之处，请在本贴的后面进行回帖，我们会逐渐进行修订。