[Pellet-users] java.lang.OutOfMemoryError: Java heap space

[Andrew McKnight]

Hello everyone,

When classifying, I get a "java.lang.OutOfMemoryError: Java heap space"
problem with much less data than seems reasonable. Here's the situation:

- using OWLAPI
- using all the default PelletOptions
- using the default heap space size for java (128M?)
- expressivity sHIN(Q)
  - I use, for example, equivalent class axioms, "exactly 1" axioms,
disjunctions
- the ontology has about 200 axioms
- about 4700:
  - individuals
  - class assertion axioms
  - object property relationships between individuals (all one of 5 types)
- all the object properties used explicitly on individuals have inverses
defined in the ontology
- not using SWRL rules

It classifies in less than 3 seconds for any number of axioms below 4500 and
then spikes between 4500 and 4700

Is it normal to have a heap overflow with this few axioms? If so, what is
the reason and is there a way around it? If an overflow isn't normal with
this few axioms, what are the usual limits?
Would removing classes and properties that are irrelevant in the ontology
make much of a difference?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.owldl.com/pipermail/pellet-users/attachments/20080612/8f5de24e/attachment.htm