update

DESCRIPTION

@@ -1,12 +1,13 @@
 Package: Rtauchen
 Type: Package
-Title: Discretization of AR(1) processes using the method by Tauchen (1986)
+Title: Discretization of AR(1) Processes using the Method by Tauchen
+        (1986)
 Version: 1.0
 Date: 2016-08-01
 Author: David Zarruk Valencia & Rodrigo Azuero Melo
-URL: https://github.com/rodazuero/Rtauchen
+URL: https://github.com/davidzarruk/Rtauchen
 Maintainer: David Zarruk Valencia <davidzarruk@gmail.com>
-Description: Method for discretizing AR(1) processes by choosing a finite set of points and transition probabilities, so the resulting finite-state Markov Chain mimics the original process. 
+Description: Description: Discretize AR process. 
 License: GPL (>= 2)
-Imports: stats, pnorm
-Packaged: 2016-08-04 21:56:05 UTC; rodrigoazuero
+Imports: stats
+Packaged: 2016-08-05 21:47:04 UTC; rodrigoazuero

NAMESPACE

@@ -1,2 +1,2 @@
 exportPattern("^[[:alpha:]]+")
-importFrom("stats", "pnorm")
+importFrom("stats","pnorm")

R/tauchen.R

@@ -1,73 +1,41 @@
 
 # Corro un archivo con los parametros
 
-Tgrid = function(n, ssigma, llambda, m){
+Tgrid=function(ne, ssigma_eps, llambda_eps, m){
 
-  grid = array(0, dim = n)
-  ssigma_y = sqrt(ssigma^2/(1-llambda^2))
+  grid = array(0, dim = ne)
+  ssigma_y = sqrt(ssigma_eps^2/(1-llambda_eps^2))
 
-  estep = 2*ssigma_y*m / (n-1)
-  for(ii in 1:1:n){
-    grid[ii] = -m*sqrt((ssigma^2)/(1-llambda^2)) + (ii-1)*estep
+  estep = 2*ssigma_y*m / (ne-1)
+  for(ii in 1:1:ne){
+    grid[ii] = -m*sqrt((ssigma_eps^2)/(1-llambda_eps^2)) + (ii-1)*estep
   }
-	return(grid)
+  return(grid)
 }
 
-Tmatrix = function(n, ssigma, llambda, m){
+Rtauchen=function(ne, ssigma_eps, llambda_eps, m){
 
-  grid = Tgrid(n, ssigma, llambda, m)
+  grid = Tgrid(ne, ssigma_eps, llambda_eps, m)
 
-  if(n > 1){
+  if(ne > 1){
     w = grid[2] - grid[1]
-    P = array(0, dim = c(n,n))
+    P = array(0, dim = c(ne,ne))
 
-    for(jj in 1:n){
-      for(kk in 1:n){
+    for(jj in 1:ne){
+      for(kk in 1:ne){
         if(kk == 1){
-          P[jj,kk] = pnorm((grid[kk] - llambda*grid[jj]+(w/2))/ssigma)
+          P[jj,kk] = pnorm((grid[kk] - llambda_eps*grid[jj]+(w/2))/ssigma_eps)
         }
-        else if(kk == n){
-          P[jj,kk] = 1- pnorm((grid[kk] - llambda*grid[jj]-(w/2))/ssigma)
+        else if(kk == ne){
+          P[jj,kk] = 1- pnorm((grid[kk] - llambda_eps*grid[jj]-(w/2))/ssigma_eps)
         }
         else{
-          P[jj,kk] = pnorm((grid[kk] - llambda*grid[jj]+(w/2))/ssigma)-pnorm((grid[kk] - llambda*grid[jj]-(w/2))/ssigma)
+          P[jj,kk] = pnorm((grid[kk] - llambda_eps*grid[jj]+(w/2))/ssigma_eps)-pnorm((grid[kk] - llambda_eps*grid[jj]-(w/2))/ssigma_eps)
         }
       }
     }
   } else{
     P = 1
   }
   return(P)
-}
-
-
-
-nTgrid = function(n, ssigma, llambda, m, dim){
-
-  n = 9
-  m = 3
-  dimension = dim(ssigma)
-  grid = array(0, dim = c(dimension[1], n))
-
-  llambda = array(c(0.7, 0.2, 0.3,  0.5), dim = c(2,2))
-  ssigma = array(c(0.1, 0, 0, 0.1), dim = c(2,2))
-
-  ssigma_y = array(0, dim = c(dimension[1], dimension[1]))
-  error = 1
-  tol = 10^-5
-  while(error > tol){
-    ssigma_y_p = llambda%*%ssigma_y%*%t(llambda) + ssigma
-    error = norm(ssigma_y_p - ssigma_y)
-    ssigma_y = ssigma_y_p
-  }
-
-  for(jj in 1:1:dimension[1]){
-    estep = 2*sqrt(ssigma_y[jj, jj])*m / (n-1)
-    for(ii in 1:1:n){
-      grid[jj, ii] = -sqrt(ssigma_y[jj, jj])*m + (ii-1)*estep
-    }
-  }
-
-
-  return(grid)
-}
+}

README.md

@@ -1,6 +1,6 @@
 Rtauchen
 =======
-[![Build Status](https://travis-ci.org/rodazuero/Rtauchen.png)](https://travis-ci.org/rodazuero/Rtauchen) 
+[![Build Status](https://travis-ci.org/davidzarruk/Rtauchen.png)](https://travis-ci.org/rodazuero/Rtauchen) 
 [![CRAN_Status_Badge](http://www.r-pkg.org/badges/version/Rtauchen)](https://cran.r-project.org/package=Rtauchen)
 
 RTauchen uses [Tauchen's (1986)](http://apps.eui.eu/Personal/Researchers/georgd/tauchen.pdf) method for discretizing AR(1) processes by choosing a finite set of points and transition probabilities, so the resulting finite-state Markov Chain mimics the original process. This method is computationally faster and numerically more stable than other methods (for example, Gaussian quadrature). This method is very popular in computational macroeconomics (See for example [Deaton (1989)](http://www.nber.org/papers/w3196), [Sargent and Ljungqvist (2012)](https://books.google.com/books?hl=es&lr=&id=H-PxCwAAQBAJ&oi=fnd&pg=PR5&ots=T-BOYPfzA7&sig=BMqWTp_3mAsSiyqQKRGm6xGClNU#v=onepage&q&f=false) and [Aiyagari (1994)](http://www.jstor.org/stable/2118417?seq=1#page_scan_tab_contents)).
@@ -15,15 +15,15 @@ install.packages("Rtauchen")
 
 # Github installation
 # install.packages("devtools")
-devtools::install_github("rodazuero/Rtauchen")
+devtools::install_github("davidzarruk/Rtauchen")
 ```
 
 
 ## Example 1
 This example computes the transition probability matrix of the finite-state Markov chain approximation of an AR(1) process with:
 n = 5 (points in the Markov chain), ssigma = 0.02, lambda = 0.95, m = 3
 ``` r
-results = Tmatrix(5, 0.02, 0.98, 3)
+results = Rtauchen(5, 0.02, 0.98, 3)
 
 results
 #               [,1]         [,2]         [,3]         [,4]         [,5]
@@ -45,8 +45,3 @@ results
 
 ```
 
-
-
-
-
-

demo/example1.R

@@ -5,7 +5,7 @@ library(Rtauchen)
 # ssigma = 0.02
 # lambda = 0.95
 # m = 3
-results = Tmatrix(5, 0.02, 0.98, 3)
+results = Rtauchen(5, 0.02, 0.98, 3)
 
 results
 #               [,1]         [,2]         [,3]         [,4]         [,5]

demo/example2.R

@@ -1,11 +1,4 @@
 library(Rtauchen)
 
-# This example computes the grid of points of the finite-state Markov chain approximation of an AR(1) process with:
-# n = 5 points in the Markov chain
-# ssigma = 0.02
-# lambda = 0.95
-# m = 3
-results = Tgrid(5, 0.02, 0.98, 3)
-
-results
-# [1] -0.3015113 -0.1507557  0.0000000  0.1507557  0.3015113
+# Include description here
+results = Rtauchen(2, 1.0e-5, 0.1,0.4)

man/Rtauchen.Rd

@@ -6,22 +6,22 @@
 Rtauchen(ne, ssigma_eps, llambda_eps, m)
 }
 \arguments{
-\item{ne}{Number of points in grid}
+\item{ne}{Number of points of the grid of the finite-state Markov chain that mimics the AR(1) process}
 
-\item{ssigma_eps}{Standard deviation of exogenous shock}
+\item{ssigma_eps}{Standard deviation of exogenous shock in the AR(1) process}
 
-\item{llambda_eps}{persistence parameter}
+\item{llambda_eps}{Persistence parameter of the AR(1) process}
 
-\item{m}{Tauchen parameter}
+\item{m}{Tauchen parameter for the width of the process (number of standard deviations of the AR(1) process covered by the grid)}
 }
 \value{
-a matrix with the corresponding transition matrix
+A matrix with the corresponding to the transition matrix of the finite-state Markov chain that approximates the AR(1) process
 }
 \description{
-Discretize AR1 process
+This function generates a matrix of transition probabilites of a finite-state Markov chain that mimics an AR(1) process with persistence parameter llamda, standard deviation ssigma and a fixed parameter m.
 }
 \details{
-Discretize AR process
+See Tauchen (1986) for details.
 }
 \examples{
 results = Rtauchen(2, 1.0e-5, 0.1,0.4)

man/Tgrid.Rd

@@ -3,16 +3,16 @@
 \alias{Tgrid}
 \title{Tgrid}
 \usage{
-Tgrid(n, ssigma, llambda, m)
+Tgrid(ne, ssigma_eps, llambda_eps, m)
 }
 \arguments{
-\item{n}{Number of points of the grid of the finite-state Markov chain that mimics the AR(1) process.}
+\item{ne}{Number of points of the grid of the finite-state Markov chain that mimics the AR(1) process}
 
-\item{ssigma}{Standard deviation of exogenous shock in the AR(1) process.}
+\item{ssigma_eps}{Standard deviation of exogenous shock in the AR(1) process}
 
-\item{llambda}{Persistence parameter of the AR(1) process.}
+\item{llambda_eps}{Persistence parameter of the AR(1) process}
 
-\item{m}{Tauchen parameter for the width of the process (number of standard deviations of the AR(1) process covered by the grid).}
+\item{m}{Tauchen parameter for the width of the process (number of standard deviations of the AR(1) process covered by the grid)}
 }
 \value{
 An array with the grid points of a finite-state Markov chain which approximates the original AR(1) process.