Page 1 of 1

Floating point error source not being shown

Posted: Fri Feb 12, 2021 12:00 pm
by damani
In my sensitivity run, I get the message shown below. Problem is it does not tell me the variable whose computation is giving this error.

Floating point error at time 24 - filling with NA
with sensitivity parameters:
Zone 1 elevation = 431.643
Double cropped land Trad init = 217.139
Rainfall variation from normal for sensitivity = 1.10768
Area2 = 533.049
Bandhara 2 capacity = 328.077
Price sensitivity multiplier = 0.901749
Conductivity = 38.8466
Floating point error at time 37 - filling with NA
****and so on*****

Re: Floating point error source not being shown

Posted: Fri Feb 12, 2021 2:43 pm
by tomfid
You can use the parameters shown to rerun a single run that will expose the problem.

Re: Floating point error source not being shown

Posted: Sat Feb 13, 2021 7:33 am
by damani
Sure. But please take it as a feature request. Simulator already knows the information - it can easily show it at that point. Often just by knowing the variable and looking at the formula, one will know what went wrong. Thanks,

Re: Floating point error source not being shown

Posted: Sat Feb 13, 2021 4:54 pm
by tomfid
It may not be known, because optimization and sensitivity runs preserve less information for the sake of speed, but I agree that this is a useful request. I'll see what's possible.