[R-Forge #2422] Handle columns with complex numbers

[arunsrinivasan]

Submitted by: Paul Metzner; Assigned to: Nobody; R-Forge link

Necessary to use data.table for signal processing, especially for the analysis of power spectra.

[MichaelChirico]

I suppose this issue is open because complex columns are not feature-complete?

E.g,

set.seed(102439)
DT <- data.table(C = c(1+2i, 1+2i, 1-2i, 1-2i), 
                 N = runif(4), I = c(1L, 1L, 2L, 2L))
DT[ , .N, by = C]

Error in``[.data.table(DT, , .N, by = C):
column or expression 1 of `by` or `keyby` is type `complex`. Do not quote column names. Usage: `DT[,sum(colC),by=list(colA,month(colB))]`

Other standard operations seem to work:

DT[, abs(C)]
# [1] 2.236068 2.236068 2.236068 2.236068

Though not always...

DT[ , sum(C), by = I]

Error in gsum(C) :
Type complex not supported by GForce sum (gsum). Either add the prefix base::sum(.) or turn off GForce optimization using options(datatable.optimize=1)

But

DT[ , base::sum(C), by = I]

Error in``[.data.table(DT, , base::sum(C), by = I) :
Type 15 in `.SD` column 0

[jangorecki]

The last one DT[ , base::sum(C), by = I] looks like a bug. This sapply(split(DT$C, DT$I), sum) works.

[arunsrinivasan]

dup of #1444.