search

cmlegault / IBMWG

Index Based Methods Working Group
0 stars 1 forks source link

replicate scenarios #49

Closed cmlegault closed 3 years ago

cmlegault commented 4 years ago

Discovered scenarios.csv has both M_CC and CatchCurve_FequalM IBMs - these are equivalent

result is that iscens 25-32 are the same as 89-96 and iscens 138-145 are the same as 202-209

too late to renumber scenarios will need to do some filtering to remove replicate runs from results will need to filter out replicate scenarios from future runs to save time will need to adjust output tables and plots to deal with the missing scenarios

gavinfay commented 4 years ago

Commit @de44288 adds code to do_mse.R to filter replicated scenarios when putting the 'todo' list together for new runs.

dupes <- duplicated(mse_sim_setup[,-(1:2)]) not_dupes <- mse_sim_setup$rowid[!dupes] ..... filter(rowid %in% not_dupes) %>%

gavinfay commented 4 years ago

@cmlegault I made some changes to your my_summarize_results.R that should help with processing and filter the bad results out.

cmlegault commented 4 years ago

Thanks Gavin! Very helpful. I modified demo_make_tables_figures.R as well to use this trick,

cmlegault commented 4 years ago

Ran into problems trying to map large numbers of large files in my_summarize_results.R (many GB of data was too much for little laptop to handle all at once). So reverted to looping through runs. Appreciate the hint about map - would be nice if running on network machine with lots of memory.

gavinfay commented 4 years ago

Ah yes. Makes sense!

cmlegault commented 3 years ago

Looks like we've got everything under control here.