Mode "profile": the objective is evaluated at the best parameter set of this ELF's ledger row (its first set, the best by replicated mean), and at that set with each coefficient of profileCoefs in turn set to each value of profileAt, profileReps times each on paired seeds (fireSenseUtils::profileCoefficients()). The values may lie outside the fit's bounds, e.g. below the fuel coefficients' lower bound of 0. The objective and its data are the fit's (spreadFitDEArgs(), emulatorObjectiveArgs()), on a cluster built as the fit's (withObjectiveCluster()). Nothing is fitted and the ledger is not written.

profileLedgerFit(sim, covs, runName)

Arguments

sim

a simList, after spreadFitPrepare.

covs

mod$covsX1000.

runName

character; labels the cluster and the output file.

Value

the profileCoefficients() table, with ledgerObjFunVal (the ledger's replicated mean of the best set, to compare with the reference row's mean). It is also written to file.path(outputPath(sim), currentModule(sim), "spreadFitProfile_<runName><fitFileSuffix>.csv").

Details

Stops when the ledger has no row for this ELF, or when the row was fitted on other parameters or other data than this run would fit: other parameter names, another covariate centre, or another power-law gamma. A profile is only about the fit when the objective is the fit's.