AWS is a library containing functions to perform the Propagation-Separation approach as introduced in

    Polzehl, J. and Spokoiny, V. (2006). Local Likelihood Modeling by Adaptive Weights Smoothing.
    Probability Theory and Related Fields, Vol.135, 335--362 .

For local polynomial smoothing see  

    Polzehl, J. and Spokoiny, V. (2007). Structural Adaptive Smoothing by Propagation-Separation-Methods,
    In: Handbook of Data Visualization (Edts. Chen, C., Haerdle, W. and Unwin, A), Springer Handbooks of Computational Statistics.

The package can also be used ( parameters aws=FALSE, memory=TRUE) to perform
the stagewise aggregation approach, see 

    Belomestny, D. and Spokoiny, V. (2007). Spatial aggregation of local likelihood estimates with applications to classification,  Ann. Statist.  25 ,  2287--2311. 

As an alternative procedure the package now includes an implementation of
the Intersection of Confidence Intervals (ICI) methods from

     Katkovnik, V.  Egiazarian, K. and  Astola, J. (2006), Local Approximation Techniques in Signal And Image Processing},
     SPIE Society of Photo-Optical Instrumentation Engin., PM157, Chapter 6.


The library has been reimplementated using S4-classes. Functionality has been rearranged. 
Functions now handle 1D/2D/3D data (if implemented for the specified model).
Functions for density estimation and tail-index estimation have been removed.
Additional functionality includes local constant Gaussian models for
irregular design (function aws.irreg) and Gaussian models with global mean-variance model (aws.gaussian).
A revised strategy for parameter selection is now provided (function awstestprop).

AWS 1.9-0 or higher supports OpenMP.  

To set the number of cores use the R-method setCores.



Joerg Polzehl
email: polzehl@wias-berlin.de
URL: http://www.wias-berlin.de/people/polzehl