Author malte
Recipients malte, silvan
Date 2015-07-27.11:53:50
Two additional notes that I made when we recently read the IJCAI 2015 paper on
simulation in M&S by Álvaro and Jörg:

- They use a "max_transitions" limit to control M&S generation. (See Section 8,
second paragraph.) It would be useful for us to implement this. The transitions
are harder to control than the states, but they are also more important. (They
use this to abort the construction process, rather than to control how much to
shrink. If we copy this behaviour, perhaps the name should be sufficiently
distinguished from max_states etc. to make this clear.)

- There are cases where exact label reduction is possible even though the two
labels have different cost. Perhaps we want to look into this.
Date User Action Args
2015-07-27 11:53:50maltesetmessageid: <>
2015-07-27 11:53:50maltesetrecipients: + malte, silvan
2015-07-27 11:53:50maltelinkissue567 messages
2015-07-27 11:53:50maltecreate