Learn R Programming

genlasso (version 1.6.1)

iterate: Iterate a genlasso object

Description

Given an incomplete genlasso path object, this function continues the path computation from the last computed knot, either until the complete path has been computed or the step limit specified by moresteps has been reached. All options are assumed to be the same as those in the initial call to a genlasso function (as in genlasso, fusedlasso, or trendfilter), with the exception of minlam and verbose, which can be changed with a call to iterate.

Usage

iterate(object, moresteps=200, minlam=0, verbose=FALSE)

Value

Returns an list of the same class typing and same structure as the passed

object.

Arguments

object

a genlasso object with an incomplete path.

moresteps

an integer specifying the number of additional steps to take, starting from termination point of the passed (incomplete) path object.

minlam

a numeric variable indicating the value of lambda at which the path should terminate. Default is 0.

verbose

a logical variable indicating if progress should be reported after each knot in the path.

See Also

genlasso, trendfilter, fusedlasso

Examples

Run this code
  # Sparse 2d fused lasso
  library(genlasso)
  set.seed(1)
  dim1 = dim2 = 10
  n = 100
  y = as.numeric(row(diag(dim1)) > 5 & col(diag(dim2)) > 5) * 3 + rnorm(n)

  a10 = fusedlasso2d(y, dim1=dim1, dim2=dim2, gamma=0.5, maxsteps=10)
  a20 = fusedlasso2d(y, dim1=dim1, dim2=dim2, gamma=0.5, maxsteps=20)
  a30 = fusedlasso2d(y, dim1=dim1, dim2=dim2, gamma=0.5, maxsteps=30)
  b20 = iterate(a10, moresteps=10)
  b30 = iterate(b20, moresteps=10)

  # Check for equality; should match on all but 'call'
  b20$call = a20$call
  b30$call = a30$call
  all.equal(target=a20, current=b20)
  all.equal(target=a30, current=b30)

Run the code above in your browser using DataLab