Faster trimmed mean algorithm?

[HenrikBengtsson]

Filing this for the record:

Gleb Beliakov, Fast Computation of Trimmed Means, Journal of Statistical Software, March 2011 presents an algorithm for trimmed means that is significantly faster than default ones, especially for large number of values.

Should/could matrixStats provide an implementation for this?

[karoliskoncevicius]

This issue seems to be a few years old but just wanted to vote in favour of this.

I am slowly developing a package called matrixTests and having something like rowMeans2(x, trim=0.1) would help in implementing some versions of Leven's test.