There is a cost to using logarithms. So while you might be able to say what the level of precision on a resultant would be if the logarithmic value was converted to the number that it represented, that does not mean that you can discard the error while the number is in logarithmic form. My recollection is that many integers have to be expressed with non-rational numbers if in logarithmic form so the implicit complexity is all over the place and can be infinite for many simple numbers. This means that you cannot eliminate error while the number is in logarithmic form and that error will be carried and quickly amplified in chain calculations.
Jim Bromer On Sat, Jun 6, 2015 at 8:44 AM, Jim Bromer <[email protected]> wrote: > Do you know a formula for computing the fudge factor for logarithmic > addition and the precision of using it? Do you know where I can find more > about this? > Incidentally, I don't think you have considered the possible implications > of a SAT solver in p. I don't know what those implications are but I know > that what might seem reasonable to you now might seem less reasonable ten > years from now. > > Jim Bromer > > On Sat, Jun 6, 2015 at 2:29 AM, Steve Richfield <[email protected] > > wrote: > >> TJ, >> >> The resend of my paper is attached to this posting. Can you see it OK? >> >> Steve >> =============== >> >> On Fri, Jun 5, 2015 at 5:33 PM, TJ Katal <[email protected]> wrote: >> >>> resend paper attatchment please i didnnt get it the first >>> On May 18, 2015 1:43 PM, "Alan Grimes" <[email protected]> wrote: >>> >>>> Steve Richfield wrote: >>>> > Alan, >>>> > >>>> > The methods to circumvent your concerns are known, but each has been >>>> > prematurely dismissed in the absence of the others. Please read my >>>> > attached FPGA-oriented paper, and then identify what (if anything) I >>>> > might have been missing. >>>> > >>>> > Specific comments follow... >>>> >>>> Okay, that was a perfectly splendid paper. >>>> >>>> 1. The reason we are using instead of x86_64 instead of <monks chanting >>>> worshipfully>the 21364 </monks> is the software... The software guys >>>> will be even more annoyed if they have to use a special version of Ada >>>> and can't rely on the chip to be in a specific configuration. >>>> >>>> 2. /me sobs pathetically about the $8,000 price tag on the Altera >>>> Stratix 5 kit and the inavailability of the Startix 10 that they've been >>>> teasing us about for a few years now. =( >>>> >>>> 3. Have u seen the AMD Kevari APU? It actually puts CPU and GPU >>>> resources into a single address space making it actually feasible to use >>>> GPU resources for more general computations without memory gymnastics. >>>> >>>> 4. Where is the paper that describes what computations you actually need >>>> for AGI? -- that would seem to be a pre-requisite for going nuts >>>> designing chips. >>>> >>>> -- >>>> IQ is a measure of how stupid you feel. >>>> >>>> Powers are not rights. >>>> >>>> >>>> >>>> ------------------------------------------- >>>> AGI >>>> Archives: https://www.listbox.com/member/archive/303/=now >>>> RSS Feed: >>>> https://www.listbox.com/member/archive/rss/303/26971539-9c060d2f >>>> Modify Your Subscription: https://www.listbox.com/member/?&; >>>> Powered by Listbox: http://www.listbox.com >>>> >>> *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> >>> <https://www.listbox.com/member/archive/rss/303/10443978-6f4c28ac> | >>> Modify <https://www.listbox.com/member/?&;> Your Subscription >>> <http://www.listbox.com> >>> >> >> >> >> -- >> Full employment can be had with the stoke of a pen. Simply institute a >> six hour workday. That will easily create enough new jobs to bring back >> full employment. >> >> *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> >> <https://www.listbox.com/member/archive/rss/303/24379807-653794b5> | >> Modify >> <https://www.listbox.com/member/?&;> >> Your Subscription <http://www.listbox.com> >> > > ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657 Powered by Listbox: http://www.listbox.com
