Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
NCTM or The Math Forum.


Punjab
Posts:
1
Registered:
12/12/04


Re: Help, ideas for module problem
Posted:
Jun 16, 1996 4:13 PM


Have you ever looked at the way that pocket calculators solve their problems? NO two companies (in some cases models from the same company) solve their problems in a similar algorithm. For instance the solution to sin (2 radians) is processed differently by almost each type of TI calculator...Although it may be slightly difficult to identify exactly how each does it... (try timing)
In article <4pjgqu$e5l@newsrocq.inria.fr>, pessaux@couchey.inria.fr says... > > > >Hi everybody, > > >Here is my problem... > >I'm researcher student at INRIA (France). I currently work on module >typing in functionnal languages, strongly typed (in fact Caml Special >Light). >My work is to add a conditionnal primitive in the module langage, in >order to be able to discriminate on types and give different >implementations of a module in function of this type discrimination. > >So, now I'm looking for an idea of use for this feature ;) >You could help me if you could give me an example of problem (math >problem, for instance) which could be solved in different ways, >depending of the type of data used in it. > >For example, I'd like to hear about something like: (I know it is >stupid, but it is only to show the kind of problem) >"We want to solve polynomial equations = 0. We know that there exit a >generic method, but we also know that if exposants are naturals, there >is a best way to solve, and if exposants are floats there is another >method... So we would like to have a programsolver which takes >advantages of looking at exposants type, in order to go faster/better..." > >This example is a non sense, Ok (I heard polynomials with float exposants >seems not to have solution for = 0, but as I'm not a mathematician, I >don't know exactly the story ;) > >But such an example would be an excellent one to expose the utility of >my work. I already have example, but they are not very interesting. > > >An other style of problem could be useful: using discrimination on >values in a module. For instance, depending of the value of something, >the solution of the problem is an integer, a float, a complex... >Something like "if the coefficient X of my equation E is 0 then I know >that my solution will be an integer. If X is 5 then I know the >solution is a complex, and so on..." > > >So, thank's for any ideas... > > > Francois > > > >



