jade_path_planned: Fit JADE at a sequence of gamma values.
In jean997/jadeTF:

Description Usage Arguments Value

This function requires having previously fit the data at gamma=0. JADE will be fit at a serries of gamma values evenly spaced on the log scale. There are three ways to specify which values of gamma will be fit: 1. Specify log.gamma.start, log.gamma.stop and n.fits 1. Specify log.gamma.start and step.size 1. Specify gammas

The start and end points should be pretty good guesses but don't need to be exact. If log.gamma.start is too large jade_path_planned will back up until it finds a value of gamma with no fusion between profiles. Similarly if log.gamma.stop is too small and the path isn't complete, the function will keep going until all the profiles are fused.log.gamma.stop. Guessing the start and stop locations can have a few negative consequences:

-If log.gamma.start is much too large it may take a (prohibitively) long time to fit the initial fit or convergence might not be reached. Warm starts help a lot with convergence times so starting in the middle of the path can be costly.

- If log.gamma.start and log.gamma.stop are way too far apart, the chosen step size will be too large which will have a similar effect as starting in the middle of the path. It will also result in a sparsely populated path which might not be very useful.

- If the endpoints are way too close together the step size will be way small and many more fits than n.fits will be calculated.

jade_path_planned(fit0, out.file, log.gamma.start = NULL, n.fits = NULL,
  log.gamma.stop = NULL, step.size = NULL, gammas = NULL,
  return.object = TRUE, tol = 0.001, verbose = TRUE, temp.file = NULL,
  max.it = 10000, log.gamma.max = 20, restart.file = NULL,
  hard.stop = FALSE, adjust.rho.alpha = TRUE)

`fit0`	Either a jade object or a file containing a jade object produced by fitting with `gamma=0`
`out.file`	Name of a file to save the results to.
`log.gamma.start`	Starting value of `log10(gamma)`.
`n.fits`	Desired number of fits along the path. Actual results may vary. `n.fits` can only be missing if `step.size` or `gammas` are provided.
`log.gamma.stop`	Speculative stopping point. If `log.gamma.stop` and `n.fits` are provided `step.size = (log.gamma.stop-log.gamma.start)/n.fits`.
`step.size`	Alternative to specifying `log.gamma.stop` and `n.fits`
`gammas`	Alternative to providing start/stop etc - just provide an exact list of all the `gamma` values to fit.
`temp.file`	Name a temp file. If missing will use a default based on `out.file`. This file will be deleted at the end.
`max.it`	Maximum number of jade iterations.
`log.gamma.max`	Absolute largest value of `gamma` that will be fit.
`restart.file`	Provide if restarting from a temporary file.
`hard.stop`	Stop at `log.gamma.stop` regardless of if the path is done or not.

The output of this function is both returned and saved to a file. Partial results are saved to a temporary file along the way. The path objects is a list with several components:

gammas: A list of length n of gamma values at which JADE was fit. The first element is always zero.
JADE_fits: A list of n JADE objects in the same order as gammas.
l1.total: Vector of length n giving the total L1 distance between pairs of profiles at each value of gamma
sep.total: Vector of length n giving the total number of separated sites between all pairs of profiles
sep: List of lists of matrices giving the pairwise separation between profiles. sep[[i]][[j-i]] is a p \times n matrix which describes the separation between profiles for group i and group j.
tol: The tolerance at which the seaparation in sep.total and sep were calculated