hyperSpec Arithmetic with row vector: extra data columns

[cbeleites]

here's the current behavior:

> chondro - chondro[1,]
hyperSpec object
   875 spectra
   8 data columns
   300 data points / spectrum
wavelength: Delta * tilde(nu)/cm^-1 [numeric] 602 606 ... 1798 
data:  (875 rows x 8 columns)
   1. y: y [numeric] -4.77 -4.77 ... 19.23 
   2. x: x [numeric] -11.55 -10.55 ... 22.45 
   3. filename: filename [character] rawdata/chondro.txt rawdata/chondro.txt ... rawdata/chondro.txt 
   4. clusters: clusters [factor] matrix matrix ... lacuna + NA
   5. spc: I / a.u. [matrix300] 0.000000 -1.364222 ... -117.3327 
   6. y.y: y [numeric] -4.77 -4.77 ... -4.77 
   7. x.y: x [numeric] -11.55 -11.55 ... -11.55 
   8. clusters.y: clusters [factor] matrix matrix ... matrix 

Note the "duplicated" columns. These are not exactly duplicated, but give the one value of the row vector:

> head(chondro - chondro[1,])$..
      y      x            filename clusters   y.y    x.y clusters.y
1 -4.77 -11.55 rawdata/chondro.txt   matrix -4.77 -11.55     matrix
2 -4.77 -10.55 rawdata/chondro.txt   matrix -4.77 -11.55     matrix
3 -4.77  -9.55 rawdata/chondro.txt   matrix -4.77 -11.55     matrix
4 -4.77  -8.55 rawdata/chondro.txt   matrix -4.77 -11.55     matrix
5 -4.77  -7.55 rawdata/chondro.txt   matrix -4.77 -11.55     matrix
6 -4.77  -6.55 rawdata/chondro.txt   matrix -4.77 -11.55     matrix

So the behavior is strictly logical - but this may get annoying.

Question: should we drop columns of a row vector that correspond (same name and label) to columns in the hyperSpec object?

[eoduniyi]

Hi @cbeleites,

I'm not sure why we wouldn't drop the duplicate columns?

[cbeleites]

@eoduniyi They are not exactly duplicates: the 2nd version contains the same value for all rows, since it was originally the first row only and the values are copied.

[eoduniyi]

@eoduniyi They are not exactly duplicates: the 2nd version contains the same value for all rows, since it was originally the first row only and the values are copied.

Right, well it seems to make sense for chondro - chondro[1, ] to remove the first row in the original hyperSpec object and get rid of the "duplicates":

chondro$y.y <- NULL 
chondro$x.y <- NULL 

[ximeg]

chondro - chondro[1,] subtracts one spectrum from all other spectra. I, as a user, would be very much surprised if this operation changes anything else apart of the $spc matrix. So I don't find this behavior logical.

[cbeleites]

@ximeg which this? dropping columns from y or merging them into the result?