Rounding error in output time

[Andy00000000000]

Hi,

Unfortunately, I cannot provide a full reproducible example. In a simulation of 10 subjects with identical dosing times, I noticed that a summary of mean concentrations vs. time was jaggedly jumping up and back down at certain times. It ended up that at the specific erroneous times, two summary records were being created (for one such input time = 2184 (attached); N = 9 was as expected, while N = 1 appeared to have a rounding error). This error was only occurring at observational outputs at times of dose, but not at all of the dose times (attached is a list of dose times with the error times marked). My workaround currently is to round mrgsolve time output to same digits as supplied input times.

Running mrgsolve version 1.4.1 on R 4.4.1 on Ubuntu 24.04 LTS. Have not yet tested whether error occurs with newest CRAN release of mrgsolve.

Best,
Andy

Simulation calls/options:

out<- mod2 %>%
  data_set(evD) %>%
  idata_set(demo) %>%
  param(x = x)%>%
  carry.out(evid,x) %>%
  mrgsim(recsort=3, tgrid = sort(unique( c( seq(0,(ndose*lcycle),1), seq(0,simendtime,4) ))), atol=1E-25, Req = "x")
as.data.frame(out)

[kylebaron]

Thanks for the report, @Andy00000000000 . This looks like some issue with representing the number. I'm not sure where number 9 turned different.

If you can, please keep working on a reproducible example that we could run. That would help us start in on this.