java.lang.ArrayIndexOutOfBoundsException error when calling accessibility

[phily5051]

I am trying to calculate accessibility for for a medium size number of points in Montreal using walk and transit but ran into the following error: java.util.concurrent.ExecutionException: java.lang.ArrayIndexOutOfBoundsException
1106 out of 2522 origins processed.
2455 out of 2522 origins processed.

Before I updated the Java SE Development Kit 11 to 21, it calculated accessibility for both modes, but now it is not working. I've included a subset of points and the files that should reproduce the error.

Data files:

• https://liveuclac-my.sharepoint.com/:u:/g/personal/ucfnpje_ucl_ac_uk/EQ8TbVR2vgBAn9X9loqTmHoBhUAAufDtpqynKfS_jF4v9w?e=FA0hgm

# assign more RAM to java
options(java.parameters = "-Xmx12G")

# essential packages
library(r5r)
library(accessibility)
library(sf)
library(data.table)
library(ggplot2)
library(interp)
library(h3jsr)
library(dplyr)

# build network dataset
r5r_core <- setup_r5(data_path = 'data/Montreal/')

# setting up the modes for general transit access
mode <- c("WALK", "TRANSIT") 
max_walk_time <- 30 
travel_time_cutoff <- 30  
departure_datetime <- as.POSIXct("12-04-2023 08:00:00", format = "%d-%m-%Y %H:%M:%S") 
time_window <- 30 
percentiles <- 50

# data points
orgpoints <- fread(file.path("data/Montreal/PopGrid1km.csv"))
destpoints <- fread(file.path("data/Montreal/POI.csv"))

# accessibility
access <- accessibility(r5r_core,
                        origins = orgpoints,
                        destinations = destpoints,
                        mode = mode,
                        opportunities_colnames = c("supermarket"),
                        decay_function = "step",
                        cutoffs = travel_time_cutoff,
                        departure_datetime = departure_datetime,
                        max_walk_time = max_walk_time,
                        time_window = time_window,
                        percentiles = percentiles,
                        progress = TRUE)

The output and the error messages are as follows:

Warning in assign_points_input(origins, "origins") :
  'origins$id' forcefully cast to character.
Warning in assign_points_input(destinations, "destinations") :
  'destinations$id' forcefully cast to character.
621 out of 2522 origins processed.Error in .jcall("RJavaTools", "Ljava/lang/Object;", "invokeMethod", cl,  : 
  java.util.concurrent.ExecutionException: java.lang.ArrayIndexOutOfBoundsException
624 out of 2522 origins processed.
2455 out of 2522 origins processed.

No error occurs when I use mode <- "WALK". Any opinion will be much appreciated.

[rafapereirabr]

Could you please run r5r::r5r_sitrep() and share the output below ?

[phily5051]

Hello, @rafapereirabr.
I ran r5r::r5r_sitrep() and the below is the output:

$r5r_package_version
[1] ‘2.0’

$r5_jar_version
[1] "7.1"

$java_version
[1] "21.0.4"

$set_memory
[1] "-Xmx12G"

$session_info
R version 4.4.1 (2024-06-14 ucrt)
Platform: x86_64-w64-mingw32/x64
Running under: Windows 10 x64 (build 19045)

Matrix products: default


locale:
[1] LC_COLLATE=Korean_Korea.utf8  LC_CTYPE=Korean_Korea.utf8   
[3] LC_MONETARY=Korean_Korea.utf8 LC_NUMERIC=C                 
[5] LC_TIME=Korean_Korea.utf8    

time zone: Europe/London
tzcode source: internal

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods  
[7] base     

other attached packages:
[1] dplyr_1.1.4         h3jsr_1.3.1         interp_1.1-6       
[4] ggplot2_3.5.1       data.table_1.15.4   sf_1.0-16          
[7] accessibility_1.4.0 r5r_2.0             rJava_1.0-11       

loaded via a namespace (and not attached):
 [1] utf8_1.2.4         generics_0.1.3     tidyr_1.3.1       
 [4] class_7.3-22       KernSmooth_2.23-24 digest_0.6.36     
 [7] magrittr_2.0.3     evaluate_0.24.0    grid_4.4.1        
[10] fastmap_1.2.0      jsonlite_1.8.8     backports_1.5.0   
[13] e1071_1.7-14       DBI_1.2.3          purrr_1.0.2       
[16] fansi_1.0.6        scales_1.3.0       Rdpack_2.6        
[19] cli_3.6.3          rlang_1.1.4        units_0.8-5       
[22] rbibutils_2.2.16   bit64_4.0.5        munsell_0.5.1     
[25] withr_3.0.0        yaml_2.3.9         tools_4.4.1       
[28] deldir_2.0-4       checkmate_2.3.1    colorspace_2.1-0  
[31] geojsonsf_2.0.3    curl_5.2.1         vctrs_0.6.5       
[34] R6_2.5.1           proxy_0.4-27       lifecycle_1.0.4   
[37] classInt_0.4-10    bit_4.0.5          V8_4.4.2          
[40] pkgconfig_2.0.3    pillar_1.9.0       gtable_0.3.5      
[43] glue_1.7.0         Rcpp_1.0.12        xfun_0.45         
[46] tibble_3.2.1       tidyselect_1.2.1   rstudioapi_0.16.0 
[49] knitr_1.48         htmltools_0.5.8.1  rmarkdown_2.27    
[52] compiler_4.4.1    

[rafapereirabr]

Thanks, @phily5051 . I was able to reproduce the error. It only occurs when using "TRANSIT" as a transportation mode. The code works fine for walking trips, for example.

This leads me to think the error is not in r5r but in the GTFS data. I would strongly suggest you check for any issues in the transit data using the GTFS validator available in the {gtfstools} package..

Once the issue in the GTFS is fixed, you will need to build the network again setting overwrite = TRUE.

r5r_core <- setup_r5(data_path = 'data/Montreal/', overwrite = TRUE)

[mattwigway]

@rafapereirabr is there a way to get rJava to print a full traceback for these Java exceptions? That would make debugging a lot easier.

[rafapereirabr]

@mattwigway , yes it is possible to traceback Java exceptions with {rJava}. See this SO a Q&A. We use this in the setup_r5() function already (see here). I've looked into the accessibility() and travel_time_matrix() functions but I didn't find a way to implement this there.

[mattwigway]

If we can't do it on the r side we should be able to catch it on the Java side, print the traceback, and then re-throw so that R knows it failed