analysis_online is a function to analyse the online measurements
and output files of the tmLQCD software, see references. The function
operates on a subdirectory either passed via rundir or automatically
constructed from the various function arguments. Depending on which
parts of the analysis are requested, this subdirectory is expected
to contain onlinemeas.%06d files with online correlator measurements,
output.data containing the plaquette and energy violation, amongst others
and monomial-%02d.data with measurements of the extremal eigenvalues of the
analysis_online(L, Time, t1, t2, beta, kappa, mul, cg_col, evals_id, rundir,
cg.ylim, type = "", csw = 0, musigma = 0, mudelta = 0, muh = 0,
addon = "", skip = 0, rectangle = TRUE, plaquette = TRUE,
dH = TRUE, acc = TRUE, trajtime = TRUE, omeas = TRUE, plotsize = 5,
debug = FALSE, trajlabel = FALSE, title = FALSE, pl = FALSE,
method = "uwerr", fit.routine = "optim", oldnorm = FALSE, S = 1.5,
stat_skip = 0, omeas.samples = 1, omeas.stride = 1, omeas.avg = 1,
omeas.stepsize = 1, evals.stepsize = 1, boot.R = 1500, boot.l = 2,
outname_suffix = "", verbose = FALSE)integer. spatial lattice extent
integer. temporal lattice extent
integer. initial time of fit range
integer. end time of fit range
numeric. inverse squared gauge coupling
numeric. hopping parameter
numeric. light sea twisted quark mass
integer. column of CG iteration counts from output.data to use
Integer. Monomial ID of the monomial for which eigenvalues are measured.
Function will attempt to open monomial-%02d.data.
string. run directory. If not specified, run directory will be constructed automatically. See construct_onlinemeas_rundir for details.
numeric. y-limits for CG iteration counts
string. Type specifier for the gauge action, this might be 'iwa' for Iwasaki, for example.
numeric. clover coefficient
numeric. average 1+1 sea twisted quark mass
numeric. splitting 1+1 sea twisted quark mass
numeric. "heavy" twisted mass in the case of a n_f=2+2 run
string. addon to output filenames
integer. number of initial measurements to skip in analysis
boolean. If true, rectangle plaquettes are analysed
boolean. If true, square plaquettes are analysed
boolean. If true, delta H is analysed
boolean. If true, the acceptance rate is analysed
boolean. If true, the time per trajectory is analysed
boolean. If true, online measurements are analysed (onlinemeas.%06d)
numeric. size of plots being generated
boolean. provide debug information
boolean or string. If not FALSE, use as trajectory labels
bolean or string. If not FALSE, use as main title of plots
boolean. If set to TRUE plots will be generated
string. method to compute errors, can be "uwerr", "boot" or "all"
string. minimisation routine for chisq, can be "optim"
boolean. If TRUE, the function assumes that the onlinemeas.%06d are
in old tmLQCD normalisation.
numeric. S parameter of uwerr
integer. By passing this parameter, the various timeseries plots
will include stat_skip measurements,
but these will be skipped in the corresponding statistical analysis.
This maybe useful, for example, to visualise thermalisation.
integer. number of stochastic samples per online measurement
integer. stride length in the reading of online measurements
integer. Block average over this many subsequent measurements.
integer. Number of trajectories between online measurements. Autocorrelation times of online measurement data will be scaled by this factor.
integer. Numer of trajectories between (strange-charm Dirac opertoar) eigenvalue measurements. Autocorrelation times of eigenvalues will be scaled by this factor.
integer. number of bootstrap samples to use in bootstrap-based parts of analysis.
integer. bootstrap block size
string. suffix for output files
boolean. If TRUE, function produces verbose output.
#'
a list is returned with all the accumulated results. Moreover, a
PDF file with statistics and analytics is created and the results
are written into .Rdata files. On the one hand, the result of the
call to the onlinemeas function is written to
onlineout.%s.Rdata, where %s is replaced with a label
built from meta information based on the arguments above.
On the other hand, summary data across many calls of this
function is silently accumulated in the file
omeas.summary.Rdata which contains the named list 'resultsum' with
element names based on rundir.
K. Jansen and C. Urbach, Comput.Phys.Commun. 180 (2009) 2717-2738