**Vincent Zoonekynd's Blog**, and kindly contributed to R-bloggers)

Many problems in statistics or machine learning

are of the form “find the values of the parameters

that minimize some measure of error”.

But in some cases, constraints are also imposed

on the parameters: for instance, that they should sum up to 1,

or that at most 10 of them should be non-zero —

this adds a combinatorial layer to the problem,

which makes it much harder to solve.

In this note,

I will give a guide to (some of) the optimization packages in R

and explain (some of) the algorithms behind them.

The solvers accessible from R have some limitations,

such as the inability to deal with binary or integral constraints

(in non-linear problems): we will see how to solve such problems.

When you start to use optimization software,

you struggle to coax the problem into the form expected

by the software (you often have to reformulate it to make it

linear or quadratic, and then write it in matrix form).

This is not very user-friendly.

We will see that it is possible to specify optimization problems

in a perfectly readable way.

# Actual R code x <- variable(5) minimize( sum(abs(x)) + sum(x^2) - .2*x[1]*x[2] ) x >= 0 x <= 1 sum(x) == 1 x[1] == x[2] r <- solve()

Many of the examples will be taken from finance and portfolio optimization.

<p><a href=”http://zoonek.free.fr/blosxom/R/2012-06-01_Optimization.html?seemore=y” class=”seemore”>See more …</a></p>

**leave a comment**for the author, please follow the link and comment on their blog:

**Vincent Zoonekynd's Blog**.

R-bloggers.com offers

**daily e-mail updates**about R news and tutorials on topics such as: visualization (ggplot2, Boxplots, maps, animation), programming (RStudio, Sweave, LaTeX, SQL, Eclipse, git, hadoop, Web Scraping) statistics (regression, PCA, time series, trading) and more...