[ODE] recent dSetZero() patch wrong?

chris caj at cs.york.ac.uk
Thu Oct 23 11:39:48 MST 2003


Adam D. Moss wrote:

>
>
> Nguyen Binh wrote:
>
>> AP> It did make StepFast working incorrectly. The cause was in 
>> stepfast.cpp,
>> AP> around line 740: "dSetZero (J, 2 *
>> AP> m * 8);" shouldn't be commented out. I don't know the "meaning" 
>> of it, but
>> AP> it seems from the code that dSetZero is actually needed there...
>>
>>     Yes, you are right! like *c, J is used as place holder for
>>     Jacobian sub matrices and it should be set to zero on initialize.
>
>
> Fixed in CVS.
>
> (We must have gotten lucky with testing and gotten memory which
> was already zero'd!)
>
OK, just checking.. did you look through the code for dSetZeros which 
can be safely mathematically removed, or just do throw and keep chucking 
ones which didn't seem to break anything? Sorry to ask so bluntly, but I 
would be worried if someone has gone through ODE throwing out lines they 
don't think are needed without being sure...

Chris




More information about the ODE mailing list