Unexpected behavior of modulo (%%) operations on negative cell values

[klaus-huebert]

Many thanks for the amazing terra package! I hope this bug report is helpful.

I searched open issues on GitHub and checked stack overflow [terra] for previous discussions of this topic and did not find any.

There seems to be a problem with the terra implementation of %% operating on negative cell values. A simple example is that -5%%2 is 1 in base R arithmetic, but a SpatRaster cell of -5 %% 2 yields a SpatRaster cell of -1. Looking more closely, it seems that for negative numbers the terra implementation of %% uses the next/larger integer multiple as a reference point, instead of the previous/smaller one (see example 2 below). I replicated this behavior across various platforms and operating systems, and included the session info for one of them.

Cheers,
-Klaus Huebert

library(terra)
# terra 1.7.83
# Warning message:
# package ‘terra’ was built under R version 4.4.2

### example 1

-5%%2 # as expected
# [1] 1

matrix(-5,1,1)%%2 # as expected
#      [,1]
# [1,]    1

rast(matrix(-5,1,1))%%2 # not as expected
# class       : SpatRaster 
# dimensions  : 1, 1, 1  (nrow, ncol, nlyr)
# resolution  : 1, 1  (x, y)
# extent      : 0, 1, 0, 1  (xmin, xmax, ymin, ymax)
# coord. ref. :  
# source(s)   : memory
# name        : lyr.1 
# min value   :    -1 
# max value   :    -1 

### example 2

c(-5:5)%%3
# [1] 1 2 0 1 2 0 1 2 0 1 2

matrix(c(-5:5),11,1)%%3
#       [,1]
#  [1,]    1
#  [2,]    2
#  [3,]    0
#  [4,]    1
#  [5,]    2
#  [6,]    0
#  [7,]    1
#  [8,]    2
#  [9,]    0
# [10,]    1
# [11,]    2

values(rast(matrix(c(-5:5),1,11))%%3)
#       lyr.1
#  [1,]    -2
#  [2,]    -1
#  [3,]     0
#  [4,]    -2
#  [5,]    -1
#  [6,]     0
#  [7,]     1
#  [8,]     2
#  [9,]     0
# [10,]     1
# [11,]     2

sessionInfo()
# R version 4.4.1 (2024-06-14 ucrt)
# Platform: x86_64-w64-mingw32/x64
# Running under: Windows 11 x64 (build 22631)
# 
# Matrix products: default
# 
# 
# locale:
# [1] LC_COLLATE=English_United States.utf8  LC_CTYPE=English_United States.utf8    LC_MONETARY=English_United States.utf8 LC_NUMERIC=C                          
# [5] LC_TIME=English_United States.utf8    
# 
# time zone: America/New_York
# tzcode source: internal
# 
# attached base packages:
# [1] stats     graphics  grDevices utils     datasets  methods   base     
# 
# other attached packages:
# [1] terra_1.7-83
# 
# loaded via a namespace (and not attached):
# [1] compiler_4.4.1   tools_4.4.1      Rcpp_1.0.12      codetools_0.2-20

terra::gdal(lib="all")
#     gdal     proj     geos 
#  "3.9.3"  "9.4.1" "3.12.2" 

[rhijmans]

Thank you for pointing out this inconsistency. This is a bit tricky (see wikipedia and ?Arithmetic; and returning a negative modulo seems more informative; and you could do abs(x %% 2). But consistency with the R base language is probably more important, so I have changed to code to achieve that:

library(terra)
v <- -4:4
x <- rast(nrow=3, ncol=3, vals=v)
y <- rast(nrow=3, ncol=3, vals=2)
 
v %% 2
#[1] 0 1 0 1 0 1 0 1 0
values(x %% y)[,1]
#[1] 0 1 0 1 0 1 0 1 0
values(x %% 2)[,1]
#[1] 0 1 0 1 0 1 0 1 0

[klaus-huebert]

A bit tricky indeed - thanks for the quick response!