>> I understand that you somehow want a more efficient datastructure for integer >> partitions, but if that is all, then it is not worth the trouble.
> I disagree, at least slightly. If it's just a little overhead, OK. But the > algorithm for generating integer partitions is probably a lot faster. So, you should use the faster algorithm. But that does *not* mean that you have to export that special data structure to the outside world. > There is a second issue: If we only implement MultiSetPartitions, then we can > only provide an efficient coercion to MultiSet MultiSet Integer. To get a > SetSpecies SetSpecies Integer, we first have to check that we really have a > SetPartition. And so on. I think, I was aware of this when I wrote http://sourceforge.net/mailarchive/message.php?msg_id=37814844 in particular the part that starts with ... So for the design of AC I would like to propose the following. structures: List L -> Generator % isomorphismTypes: List L -> Generator % with the following interpretation: Would that fit your needs? >> I have reasons to avoid an extra IsomorphismType domain/category. >> >> 1) It makes the code more complicated. > > But you can generate isomorphismtypes of compositions now. As far as I know, > MuPAD Combinat only does special cases. Same holds for the Derivative, and for > Pointing, I think. I hope, Martin, you understand me correctly. I don't have something against including the computation of iso types. I worry about the design of the code. >> 2) you don't gain anything if you work with decomposable structures >> (Look at how you build the IsomorphismType for Plus and Times.) > I don't understand exactly what you mean here. The code for isomorphismTypes > may look similar, but it's doing something quite different. No that is not what I mean. I meant that <<representation: Times>>= Rep == Record(tag: Subset L, left: F L, right: G L); @ <<representation: MultiTimes>>= Rep == Record( tag: MultiSubset L, left: IsomorphismType$(F L), right: IsomorphismType$(G L) ); @ are actually very much the same construction, ie, you get a record of 3 things. So the argument of getting more efficient data structures does not apply for MultiPlus, MultiTimes, MultiCompose. For me it just looks like doubling code. And further, I would guess, that the current compiler has problems with using a domain exported by another domain. There must be something better. Ralf ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys - and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV _______________________________________________ Aldor-combinat-devel mailing list Aldor-combinat-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/aldor-combinat-devel
