Forum Replies Created
-
Posts
-
I have had a couple of issues with clingo4.x that I thought it might be useful to share.
1) There are some issues associated with grounding aggregates and also left assignment unifications (see bugs 100 and 102 in http://sourceforge.net/p/potassco/bugs/milestone/gringo-4.x/)
They are quite subtle, occasionally literals may be dropped from a rule or a rule disappear entirely. Localising the problem is quite tricky, sometimes reordering rules or changing them slightly without actually changing the semantics of the rule can lead to different results.2) When using the incremental grounder/solver with weak constraints – the optimisation levels reported don’t always reflect the answer set. They seem to only count the weights introduced in the grounding iteration.
Both these problems have been addressed in the trunk branch (4.5.0). This has not yet been released but can be downloaded from SourceForge. If you are experiencing any “odd” behaviour, it may be worth trying a quick test with this version.
I have tried a few of the standard options available in 4.3.0 – they do perform differently but not in a predictable way.
One thing that really does consistently speed things up, particularly when optimising, is using a solver that has been compiled with multi threading switched on.I am getting along ok with clingo 4.3.0.
I have found that it seems to work fine – once one gets familiar with the various syntax changes.
There are examples with the source code for both 4.3.0 and 4.4.0 that are quite useful.
Those together with the readme comments and the standardised language spec are just about enough to go on.I tried out clingo 4.4.0 but have found some problems when using Lua functions.
Roland identified a problem in the implementation that can occur for programs having large groundings. As far as I am aware the fix is not published yet, so tread carefully. I am sticking with 4.3.0 for the moment.
