From clive at xtra.co.nz Mon Aug 21 07:20:05 2006 From: clive at xtra.co.nz (Clive Gifford) Date: Mon Aug 21 07:21:16 2006 Subject: [icfp-discuss] Eigenratios Message-ID: <000601c6c513$c6a4afd0$0301a8c0@Owhango1> I wonder who is still listening on this list? (Possibly not that many after all the recent "eigenratio" SPAM!) Sorry, but here's a little more... I'm still obsessing over tall towers of self-interpeters and other exotica that I haven't played with before. I want to try some self-interpreters in high level languages, especially running them on different platforms/implementations to get a feeling for how the "eigenratio" thing works (or doesn't as the case may be!). I.e. As N increases, does the running time for N+1 levels of the interpeter (plus some other program "on top") divided by the running time for just N levels (running the same other top level program) tend towards some constant (what I've called the "eigenratio")? I've been collecting examples of various self-interpreters. There are links to some of them http://eigenratio.blogspot.com but the only high level one I've tried to run so far is a Scheme version as given in the link below. http://groups.google.com/group/comp.lang.scheme/msg/dc37c81fb146b37a I don't know much (almost nothing really) about Scheme so I can't really comment about this particular self-interpreter - just the first one I picked. I did manage to run it on the latest realease (352) of PLT Scheme (on my old 500Mhz PIII Win2K machine) and (with a couple more levels of the "Fak" test added at the end) and the eigenratio seems to be heading towards something in the mid 50s. (Note that the "time" measurements sometimes do very weird things on my machine producing negative values and so on for CPU and gc. Maybe a bug?) Anyway, is anybody out there willing and able to try this out in different environments to see if the numbers are reasonably consistent? Pleas tell me if you think there are "better" examples of Scheme self-interpreters that I should be trying instead? (I have found a few others but haven't tried them yet.) Perhaps more portable than this one also? I naively tried it on MIT Scheme 7.7.1 and it wasn't happy. Probably very easy to fix if you know more about Scheme than me (which is extremely likely!). Anyway, I'm currently interested in any and all examples of self-interpeters (in any language at all) if you are able to help me with any links, references, etc. Thanks, Clive From davidad at mit.edu Mon Aug 21 07:39:47 2006 From: davidad at mit.edu (David Dalrymple) Date: Mon Aug 21 07:39:50 2006 Subject: [icfp-discuss] Eigenratios In-Reply-To: <000601c6c513$c6a4afd0$0301a8c0@Owhango1> References: <000601c6c513$c6a4afd0$0301a8c0@Owhango1> Message-ID: <504b2780608210439q42efc675v29b4e3b4ffeaeb71@mail.gmail.com> > I wonder who is still listening on this list? (Possibly not that many after > all the recent "eigenratio" SPAM!) I've been getting all of it and marking it "To Do" for some time when I have lots of time to kill.... it looks tremendously interesting but it also looks tremendously time-consuming to read and understand :D --David From nelsoneci at gmail.com Mon Aug 21 14:19:26 2006 From: nelsoneci at gmail.com (Nelson Castillo) Date: Mon Aug 21 14:19:32 2006 Subject: [icfp-discuss] Eigenratios In-Reply-To: <000601c6c513$c6a4afd0$0301a8c0@Owhango1> References: <000601c6c513$c6a4afd0$0301a8c0@Owhango1> Message-ID: <2accc2ff0608211119j598b9e09l586c5c5766eead7a@mail.gmail.com> On 8/21/06, Clive Gifford wrote: > I wonder who is still listening on this list? (Possibly not that many after > all the recent "eigenratio" SPAM!) I still listen to it, even when I don't follow it. I wonder how the eigenratio thing could be useful to me :) Seems like fun, anyway (I like theory), but I don't have time to kill now on things that don't pay off soon. > to some of them http://eigenratio.blogspot.com but the only high level one http://eigenratios.blogspot.com Regards. -- http://arhuaco.org/