update

Author: James Ramsay

.Rbuildignore

@@ -5,3 +5,4 @@ revdep
 ^.*\.Rproj$
 ^\.Rproj\.user$
 fda_release_to_CRAN.R
+^CRAN-SUBMISSION$

DESCRIPTION

@@ -1,10 +1,10 @@
 Package: fda
-Version: 6.1.3
-Date: 2023-04-24
+Version: 6.1.4
+Date: 2023-05-23
 Title: Functional Data Analysis
 Authors@R: c(person("James", "Ramsay", 
             role=c("aut","cre"),
-            email="james.ramsay@mcgill.ca"), 
+            email="ramsay@psych.mcgill.ca"), 
           person("Giles", "Hooker",
             role="ctb",
             email="gjh27@cornell.edu"),
@@ -29,5 +29,5 @@ Description: These functions were developed to support functional data
 License: GPL (>= 2)
 URL: http://www.functionaldata.org
 LazyData: true
-NeedsCompilation: no
+NeedsCompilation: yes
 VignetteBuilder: knitr

NAMESPACE

@@ -195,6 +195,8 @@ S3method(density,      fd)
 S3method(deriv,        fd)
 S3method(lines,        fd)
 S3method(mean,         fd)
+S3method(matrix,       matplot)
+S3method(numeric,      matplot)
 S3method(norder,       fd)
 S3method(plot,         fd)
 S3method(predict,      fd)

R/smooth.morph.R

@@ -40,12 +40,18 @@ smooth.morph <- function(x, y, ylim, WfdPar,
   #  WFD       Functional data object for W.  It's coefficient vector
   #               contains the optimized coefficients.
 
-  #  last modified 3 June 2022 by Jim Ramsay
+  #  last modified 17 May 2023 by Jim Ramsay
 
-  nobs <- length(x)        #  number of observations
+  #  number of observations and weights on x-values
+  
+  nobs <- length(x)        
   wt   <- matrix(1,nobs,1)
   wt[nobs] <- 10
 
+  #  -----------------------------------------------------
+  #                  Check arguments
+  #  -----------------------------------------------------
+  
   # check consistency of x and y and convert to column matrices
 
   if (length(y) != nobs) {
@@ -81,10 +87,13 @@ smooth.morph <- function(x, y, ylim, WfdPar,
   }
 
   #  -----------------------------------------------------
-  #      extract information from WfdPar
+  #                  Initialize optimization
   #  -----------------------------------------------------
 
+  #      extract information from WfdPar
+  
   Wfdobj   <- WfdPar$fd
+  cvec     <- Wfdobj$coef   #  initial coefficients
   Wbasis   <- Wfdobj$basis      #  basis for Wfdobj
   Wnbasis  <- Wbasis$nbasis      #  no. basis functions
   Wrange   <- Wbasis$rangeval
@@ -96,20 +105,22 @@ smooth.morph <- function(x, y, ylim, WfdPar,
     stop("One or more weights are negative.") 
   }
 
-  cvec <- Wfdobj$coef   #  initial coefficients
-  Zmat <- fda::zerobasis(length(cvec))
-  bvec <- t(Zmat) %*% cvec
-  cvec <- Zmat %*% bvec
-  
-  #  check range of x
+  #  check that values in x are within limits in xlim
 
   if (abs(x[1] - xlim[1]) > 1e-7 || abs(x[nobs] - xlim[2]) > 1e-7) {
-    stop("Values in x are out of bounds by more than 1e-7")
+    print("Argument vector x is out of range:")
+    stop("Values are out of bounds by more than 1e-7")
   }  else {
     x[1   ] <- xlim[1]
     x[nobs] <- xlim[2]
   }
 
+  #  transform coefficients to zero column sum 
+  
+  Zmat <- fda::zerobasis(length(cvec))
+  bvec <- t(Zmat) %*% cvec
+  cvec <- Zmat %*% bvec
+  
   #  initialize matrix Kmat defining penalty term
 
   if (Wlambda > 0) {
@@ -160,10 +171,6 @@ smooth.morph <- function(x, y, ylim, WfdPar,
   iterhist <- matrix(0,iterlim+1,length(status))
   iterhist[1,]  <- status
 
-  #  -----------------------------------------------------
-  #  -------------  Begin main iterations  ---------------
-  #  -----------------------------------------------------
-  
   STEPMAX <- 10
   itermax <- 20
   TOLX    <- 1e-10
@@ -220,7 +227,7 @@ smooth.morph <- function(x, y, ylim, WfdPar,
   }
 
   #  -----------------------------------------------------
-  #         construct output objects
+  #    construct output objects and return in a list
   #  -----------------------------------------------------
 
   Wfdobj  <- fd(cvec, Wbasis)
@@ -233,7 +240,8 @@ smooth.morph <- function(x, y, ylim, WfdPar,
   hwidth  <- hmax - hmin
   hfine   <- (ylim[1] - hmin) + hfine*(ywidth/hwidth)
 
-  return(list(Wfdobj=Wfdobj, f=f, grad=grad, hmat=hmat, norm=norm, hfine=hfine, 
+  return(list(Wfdobj=Wfdobj, f=f, grad=grad, hmat=hmat, 
+              norm=norm, hfine=hfine, 
               iternum=iternum, iterhist=iterhist))
 
 }

data/StatSciChinese.rda

@@ -67,10 +67,11 @@ as.array3(x)
   ##
   ## 4-d array 
   ##
-  
-  if(!CRAN()) {
-    as.array3(array(1:24, 1:4)) 
-  }
+  # These lines throw an error because the dimensionality woud be 4
+  # and as.array3 only allows dimensions 3 or less.
+  # if(!CRAN()) {
+  #   as.array3(array(1:24, 1:4)) 
+  # }
 }
 \keyword{utilities}
 

fda_release_to_CRAN.R

@@ -113,9 +113,9 @@ lines(x., eval.fd(x., fn.), col='green', lty='dotted', lwd=3)
 
 if(!CRAN()) {
 # Will NOT translate a periodic spline
-fp <- splinefun(x, y, method='periodic')
-as.fd(fp)
-#Error in as.fd.function(fp) :
+# fp <- splinefun(x, y, method='periodic')
+# as.fd(fp)
+# Error in as.fd.function(fp) :
 #  x (fp)  uses periodic B-splines, and as.fd is programmed
 #   to translate only B-splines with coincident boundary knots.
 }

man/StatSciChinese.Rd

@@ -19,8 +19,8 @@ eval.surp(evalarg, Wfdobj, nderiv = 0)
   the set \code{c(0,1,2)}.}
 }
 \details{A surprisal \code{M}-vector is information measured in \code{M}-bits.
-Since a multinomial probability vector must sum to one, it follows that the
-surprisal vector \code{S} must satisfy the constraint 
+Since a multinomial probability vector must sum to one, it follows 
+that the surprisal vector \code{S} must satisfy the constraint 
 \code{log_M(sum(M^(-S)) = 0.} That is, surprisal vectors lie within a
 curved \code{M-1}-dimensional manifold.