locits (version 1.7.7)

StoreStatistics: Interogates calculation store to see how well we are reusing previous calculations (debugging)

Description

The computation of the variance of the lacf estimator is intensive and we try to speed that up by reusing calculations. These calculations are stored in internal C arrays. This function interrogates those arrays and can provide details on how well the storage is working and provide hints if more storage needs to be allocated. For very large time series it is possible that values need to be calculated that can be stored and this function can monitor this.

Usage

StoreStatistics()

Value

None.

Author

Guy Nason.

Details

The function prints out the state of the storage. Three numbers are reported on. 1. The number of values that were calculated but not stored "outside framework". Ideally you want this number to be low, if it gets persistently high then more storage needs to be allocated in the C code (notably MAXELL, MAXJ, MAXK, MAXD for the ThmStore and ValExists arrays).

The other two numbers are "Number stored" and "Number found". The first number corresponds to the number of values calculated once and then stored. The second number contains the number of times the software interogated the store and found a value that it did not have to then calculate. So, ideally, you'd like the latter number to be a high percentage of the former number, as this means the store is working efficiently.

Note, this function is definitely not intended for casual users. However, for users of very large series, who have the computational resources, these storage parameters might need to be increased.

The values will be zero if Rvarlacf has not yet been called, and only refer to the last call to that function (as the function zeroes the store on invocation).

References

Nason, G.P. (2013) A test for second-order stationarity and approximate confidence intervals for localized autocovariances for locally stationary time series. J. R. Statist. Soc. B, 75, 879-904. tools:::Rd_expr_doi("10.1111/rssb.12015")

See Also

Rvarlacf

Examples

#
# Simulate some data
#
x <- tvar1sim()
#
# Calculate lacf and confidence intervals
#
x.lacf <- Rvarlacf(x=x, nz=50, var.lag.max=20)
#
# Find out how the store did
#
StoreStatistics()
#Number calculated outside framework:  0 
#Number calculated then stored:  154440 
#Number found in store:  14980680 
#
#Overall % calculated:  1.020408 
#% outside framework:  0