Parameters for prosecution hypothesis, as generated by optimisation.params.
D.pars
Parameters for prosecution hypothesis, as generated by optimisation.params.
tolerance
Tolerance for the final chunk of optimisation. If the relative difference
between the current result, and the last checked result is less than this value, then it is
classed as converged.
n.steps
Number of steps to run. Defaults to NULL. if n.steps is NULL , the number
of steps to run is determined by the mean of the standard deviation of the initial phase of
the intial chunk of optimisation for prosecution and defence.
progBar
Logical, stating whether to display a graphical progress basr or not. This should
be set to FALSE is the user does not have graphical capabilities e.g. if running from command
line on a server.
interim
Logical, stating whether or not to generate interim reports. If set to TRUE a basic
set of results after each step is output to "Interim.csv", and an image of the content of
evaluate() or evaluate.from.interim() "interim.RData" are both stored in the current working
directory. The latter can be used by evaluate.from.interim() to continue an evaluation from its
previous state. Each step will write over the results from the previous step.
CR.start
Numerical, between 0 and 1, used by DEoptim as CR argument, at the start of the
search. Gradually moves towards CR.end to allow a broader initial search, gradually becoming
more localised in parameter space. See DEoptim for further details.
CR.end
Numerical, between 0 and 1,see details for CR.start.
seed.input
An integer that should be specified if the user wishes to set a particular seed.
If not specified, the program sets the seed to an integer representation of the present time, date and process ID.
Details
Optimize over parameter space, using a geometric progression of crossover rate and tolerance.
Both prosecution and defence cases are optimized simultaneously.
Value
A list containing five elements:
Pros
Prosecution results, structured as results from DEoptim::DEoptim.
Def
Defence results, structured as results from DEoptim::DEoptim.
WoE
WoE for each chunk. The final value if the final WoE.
seed.used
Seed that is set at the beginning of computation.
seed.input
Seed that is input by the user.
See Also
DEoptim,DEoptimLoop
Examples
## Not run:
# datapath to example files
datapath = file.path(system.file("extdata", package="likeLTD"),"hammer")
# File paths and case name for allele report
admin = pack.admin.input(
cspFile = file.path(datapath, 'hammer-CSP.csv'),
refFile = file.path(datapath, 'hammer-reference.csv'),
caseName = "hammer",
kit= "SGMplus"
)
# Enter arguments
args = list(
nUnknowns = 1,
doDropin = FALSE,
ethnic = "EA1",
adj = 1,
fst = 0.02,
relatedness = c(0,0)
)
# Create hypotheses
hypP = do.call(prosecution.hypothesis, append(admin,args))
hypD = do.call(defence.hypothesis, append(admin,args))
# Get parameters for optimisation
paramsP = optimisation.params(hypP)
paramsD = optimisation.params(hypD)
# Run optimisation
# n.steps set for demonstration purposes
results = evaluate(paramsP, paramsD, n.steps=1)
## End(Not run)