Gadget3 array-handling utilities

Tools to make munging array reports easier

g3_array_agg(
        ar,
        margins = NULL,
        agg = sum,
        opt_time_split = !("time" %in% margins || "time" %in% ...names()),
        opt_length_midlen = FALSE,
        ... )

Arguments

ar

Input array, e.g. dstart_fish__num from a model report

margins

dimension names to include in the final array, e.g. c("age", "year") to generate a report by-year & age. If NULL, no aggregation is done

agg

Function to use when aggregating

opt_time_split

Boolean, should we split up "time" into separate "year" & "step" dimensions?

opt_length_midlen

Boolean, should we convert "length"

...

Filters to apply to any dimension, including "year" / "step" if opt_time_split is TRUE. e.g. length = 40, age = 5, step = 1

Details

g3_array_agg allows you to both filter & aggregate an array at the same time.

Specifying a filter in ... is simplfied in comparison to a regular R subset:

  1. You can give the dimensions in any order

  2. Values are always interpreted, age = 3 will be interpreted as "age3", not the third age.

For particular dimensions we have extra helpers:

age

Numeric ages e.g. age = 5 are converted to "age5", as generated by gadget3

length

Numeric lengths will pick a value within groups, e.g. with lengths "10:20", "20:30", length = 15 will pick the smaller lengthgroup

Value

An array, filtered by ... and aggregated by margins

Examples

# Generate an array to test with
dn <- list(
    length = c("50:60", "60:70", "70:Inf"),
    age = paste0("age", 0:5),
    time = paste0(rep(1990:1996, each = 2), c("-01", "-02")) )
ar <- array(
    seq_len(prod(sapply(dn, length))),
    dim = sapply(dn, length),
    dimnames = dn)
ar[,,"1994-02", drop = FALSE]
#> , , time = 1994-02
#> 
#>         age
#> length   age0 age1 age2 age3 age4 age5
#>   50:60   163  166  169  172  175  178
#>   60:70   164  167  170  173  176  179
#>   70:Inf  165  168  171  174  177  180
#> 

# Generate by-year report for ages 2..4
g3_array_agg(ar, c('age', 'year'), age = 2:4)
#>       year
#> age    1990 1991 1992 1993 1994 1995 1996
#>   age2  102  318  534  750  966 1182 1398
#>   age3  120  336  552  768  984 1200 1416
#>   age4  138  354  570  786 1002 1218 1434

# ...for only step 1
g3_array_agg(ar, c('age', 'year'), age = 2:4, step = 1)
#>       year
#> age    1990 1991 1992 1993 1994 1995 1996
#>   age2   24  132  240  348  456  564  672
#>   age3   33  141  249  357  465  573  681
#>   age4   42  150  258  366  474  582  690

# Report on smallest length group, for each timestep
g3_array_agg(ar, c('length', 'time'), length = 55)
#>        time
#> length  1990-01 1990-02 1991-01 1991-02 1992-01 1992-02 1993-01 1993-02 1994-01
#>   50:60      51     159     267     375     483     591     699     807     915
#>        time
#> length  1994-02 1995-01 1995-02 1996-01 1996-02
#>   50:60    1023    1131    1239    1347    1455
# Use midlen as the dimension name
g3_array_agg(ar, c('length', 'time'), length = 55, opt_length_midlen = TRUE)
#>       time
#> length 1990-01 1990-02 1991-01 1991-02 1992-01 1992-02 1993-01 1993-02 1994-01
#>     55      51     159     267     375     483     591     699     807     915
#>       time
#> length 1994-02 1995-01 1995-02 1996-01 1996-02
#>     55    1023    1131    1239    1347    1455