Error in df(X0) : argument "df1" is missing, with no default--tracing R code

Question

I have written two gradient descent functions and in the second one I just have the alpha parameter and the initial alpha is different. I receive a weird error and was unable to trace the reason for it.

Here's the code:

k=19000
rho.prime<-function(t,k) ifelse (abs(t)<=k,2*t,(2*k*sign(t)))

dMMSE <- function(b,k=19000, y=farmland$farm, x=farmland$land){

  n = length(y)
  a=0
  d=0
  for (i in 1:n) {

    a = a + rho.prime(y[i]-b[1]-b[2]*x[i],k)
    d = d + x[i]*rho.prime(y[i]-b[1]-b[2]*x[i],k)
  }
  a <- (-a/n)
  d <- (-d/n)
  return(c(a,d))
}

grd=gr.descent(dMMSE, c(3500,0.33),alpha=0.0001, verbose=TRUE)

gr.descent2 <- function(dMMSE,x0, alpha=0.1, eps=0.001, max.it = 50, verbose = FALSE){
  X1 <- x0
  cond <- TRUE
  iteration <- 0
  if(verbose) cat("X0 =",X1,"\n")
  while(cond){
    iteration <- iteration + 1
    X0 <- X1
    X1 <- X0 - alpha * df(X0)
    alpha <- alpha/2
    cond <- sum((X1 - X0)^2) > eps & iteration < max.it
    if(verbose) cat(paste(sep="","X",iteration," ="), X1, "\n")
  }
  print("mona2")
  print(X1)
  return(X1)
}

grd2=gr.descent2(dMMSE, c(3500,0.33),alpha=0.1, verbose=TRUE)
#(beta0=grd2[1])
#(beta1=grd2[2])

So when I run the code I receive this error:

[1] "mona"
[1]    3496.409 -259466.640
X0 = 3500 0.33 
 Show Traceback

 Rerun with Debug
 Error in df(X0) : argument "df1" is missing, with no default 

Which is related to gr.descent2 function. Any thought?

Answer 1

Here's the answer:

farmland <- read.csv("http://pages.stat.wisc.edu/~gvludwig/327-5/FarmLandArea.csv")
str(farmland)
plot(farm~land,data=farmland)
fit=lm(farm~land,data=farmland)
abline(fit) #lease square regression line
abline(rlm(farm~land,data=farmland),col="red")
gr.descent <- function(der_f, x0, alpha=0.0001, eps=0.001, max.it = 50, verbose = FALSE){
  X1 <- x0
  cond <- TRUE
  iteration <- 0
  if(verbose) cat("X0 =",X1,"\n")
  while(cond){
    iteration <- iteration + 1
    X0 <- X1
    X1 <- X0 - alpha * der_f(X0)
    cond <- sum((X1 - X0)^2) > eps & iteration < max.it
    if(verbose) cat(paste(sep="","X",iteration," ="), X1, "\n")
  }
  print("mona")
  print(X1)
  return(X1)
}



rho<-function(t,k) ifelse(abs(t)<=k,t^2,(2*k*abs(t))-k^2)
k=19000
rho.prime<-function(t,k) ifelse (abs(t)<=k,2*t,(2*k*sign(t)))

dMMSE <- function(b,k=19000, y=farmland$farm, x=farmland$land){

  n = length(y)
  a=0
  d=0
  for (i in 1:n) {

    a = a + rho.prime(y[i]-b[1]-b[2]*x[i],k)
    d = d + x[i]*rho.prime(y[i]-b[1]-b[2]*x[i],k)
  }
  a <- (-a/n)
  d <- (-d/n)
  return(c(a,d))
}

grd=gr.descent(dMMSE, c(3500,0.33),alpha=0.0001, verbose=TRUE)

gr.descent2 <- function(der_f,x0, alpha=0.1, eps=0.001, max.it = 50, verbose = FALSE){
  X1 <- x0
  cond <- TRUE
  iteration <- 0
  if(verbose) cat("X0 =",X1,"\n")
  while(cond){
    iteration <- iteration + 1
    X0 <- X1
    X1 <- X0 - alpha * der_f(X0)
    alpha <- alpha/2
    cond <- sum((X1 - X0)^2) > eps & iteration < max.it
    if(verbose) cat(paste(sep="","X",iteration," ="), X1, "\n")
  }
  print("mona2")
  print(X1)
  return(X1)
}

#plot(farm~land,data=farmland)
#curve(rho(k=19000),xlim=c(-10,10),,col="blue", add="TRUE")
grd2=gr.descent2(dMMSE, c(3500,0.33),alpha=0.1, verbose=TRUE)
#(beta0=grd2[1])
#(beta1=grd2[2])

Answer 2

Type this:

?df     # the F distribution density

And notice that the df1 and df2 arguments are not assumed to be any particular value so they do need to be supplied.

integrate( function(x) df(x, 1, 100), 0, 3.84)
# 0.9471727 with absolute error < 1.4e-05

And notice the similarity of result:

> integrate( function(x) dchisq(x, 1), 0, 3.84)
0.9499565 with absolute error < 1.4e-05