[klee-dev] Why there are so many weighted random search algorithms in KLEE?

[Loi Luu]

Hi Greg,

What I am curious is that when we select the state, we do it randomly.
Therefore, the effect of distribution type is reduced or maybe zero. So am
I right to think that?

Regards,


On Tue, Apr 30, 2013 at 6:57 PM, <gwpublic at wp.pl> wrote:

> Hello,
>
> > I noticed that there are several NURS algorithms in klee, and each will
> > randomly select a state based on its own distribution. So why do we need
> > many of them in KLEE while after all we do randomly select states?
>
>
> For me it feels like you've answered your own question:
> Because they have different distributions :).
>
> If you would have findings about best distribution for selecting next
> node (maybe taking into account control/data flow structure) this
> might be nice contribution.
>
> Cheers,
> Greg
>



-- 
Loi, Luu The (Mr.)
University of  Engineering and Technology, Vietnam National University,
Hanoi.
-------------- next part --------------
HTML attachment scrubbed and removed