ArmaUtils-package: Fast RcppArmadillo and RcppEigen Helper Functions for Dynamic...
In floswald/ArmaUtils: Rcpp Helper Functions

Description Details Author(s) References Examples

Implements several functions useful for discrete and continuous state dynamic programming. A prototypical problem would be to approximate V(S) = max_x u(S,x) + b * V(S'|S,x) where S stands for the state space, x for control variables, V() and u() are real valued functions. The package is useful if S is large. It has the following functionality:

compute u(c,h):	`utilfun` computes u(c,h)
compute u(c,h):	`ufun_Atta` computes u(c,h) after Wakefield et al.
compute u(c,l,h):	`ufun_Atta_L` computes u(c,l,h) with continuous l
compute u(c,l,h):	`ufun_Atta_disc` computes u(c,l,h) with discrete l
max u(S,x) + b * V(S'\|S,x):	`armamax` rowwise max
approximate V(S'):	`armakron` obtain B-spline coefficients
evaluate approximations of V(S'):	`krons` evaluate B-spline coeffs

Package:	ArmaUtils
Type:	Package
Version:	1.0
Date:	2012-09-11
License:	What license is it under?

florian.oswald@gmail.com

C. De Boor. Efficient computer manipulation of tensor products. ACM Transactions on Mathematical Software (TOMS), 5(2):173-182, 1979
M. Miranda and P. Fackler. Applied computational economics and finance. MIT Press, 2002
Orazio P. Attanasio, Renata Bottazzi, Hamish W. Low, Lars Nesheim, and Matthew Wakefield. Modelling the demand for housing over the life cycle. Review of Economic Dynamics, 15(1):1 – 18, 2012.

## Not run: 
x1 <- matrix(rnorm(25),nrow=5,ncol=5)
x2 <- matrix(rnorm(9),nrow=3,ncol=3)
x3 <- matrix(rnorm(4),nrow=2,ncol=2)
y  <- rnorm(5*3*2)

R.kron   <- kronecker(x3,kronecker(x2,x1))	# form R kronecker product. impossible beyond a certain size
R.result <- R.kron 

Arma.result <- armakron(y=y,list(x1,x2,x3))	# armakron() computes result directly without forming kronecker product

all.equal(R.result,Arma.result)		# TRUE

# armamax()
result <- armamax(x1)
all.equal(apply(x1,1,max),as.numeric(result$values))	# TRUE
all.equal(apply(x1,1,which.max),as.numeric(result$indices))	# TRUE

## End(Not run)