Re: [fonc] Morphic 3 defensive disclosure
On Tue, 03 Dec 2013 23:24:12 -0300 J. Vuletich (mail lists) juanli...@jvuletich.org wrote: Hi Folks, The first defensive disclosure about Morphic 3 has been accepted and published at http://www.defensivepublications.org/publications/prefiltering-antialiasing-for-general-vector-graphics and http://ip.com/IPCOM/000232657 .. Morphic 3 is described at http://www.jvuletich.org/Morphic3/Morphic3-201006.html On http://www.jvuletich.org/Morphic3/Morphic3-201006.html, you claim: Anti-aliasing is usually considered a technique to avoid stairway artifacts on rendered images. This is a simplistic view on the problem. Aliasing is a consequence of sampling continuous functions (images, photos, sound, etc). Makers of digital cameras and audio software know and use the theory behind it. You can read more at http://en.wikipedia.org/wiki/Nyquist–Shannon_sampling_theorem. Researches know all this. The best text books say it. However, existing graphics software completely ignore the theory. ... This allows for mathematically proved alias free rendering. As no existing application does this ... I'm sort of puzzled by this. I've always thought that this was the whole idea behind the stochastic sampling thingy that the ILM/Pixar people patented (http://www.google.com/patents/US4897806) in the 1980's to achieve mathematically proven alias-free rendering (as you said) of arbitrarily shaded arbitrary geometry (even shaded with non-analytical functions). Of course, it trades aliasing for noise, but I believe that you can have the noise arbitrarily low (and for animations, it may not matter all that much anyway since one can expect some grain or noisiness on live footage so completely noise-free sampling may even look unnatural). They certainly didn't ignore the problem; they had been studying numerous analytical and non-analytical solutions for a better part of the 1980s and then finally striked gold with stochastic sampling and PRMan. -- Gath ___ fonc mailing list fonc@vpri.org http://vpri.org/mailman/listinfo/fonc
Re: [fonc] Layering, Thinking and Computing
On Sat, Apr 13, 2013 at 8:29 PM, David Barbour dmbarb...@gmail.com wrote: On this forum, 'Nile' is sometimes proffered as an example of the power of equational reasoning, but is a domain specific model. Isn't one of the points of idst/COLA/Frank/whatever-it-is-called-today to simplify the development of domain-specific models to such an extent that their casual application becomes conceivable, and indeed even practical, as opposed to designing a new one-size-fits-all language every decade or so? I had another idea the other day that could profit from a domain-specific model: a model for compiler passes. I stumbled upon the nanopass approach [1] to compiler construction some time ago and found that I like it. Then it occurred to me that if one could express the passes in some sort of a domain-specific language, the total compilation pipeline could be assembled from the individual passes in a much more efficient way that would be the case if the passes were written in something like C++. In order to do that, however, no matter what the intermediate values in the pipeline would be (trees? annotated graphs?), the individual passes would have to be analyzable in some way. For example, two passes may or may not interfere with each other, and therefore may or may not be commutative, associative, and/or fusable (in the same respect that, say, Haskell maps over lists are fusable). I can't imagine that C++ code would be analyzable in this way, unless one were to use some severely restricted subset of C++ code. It would be ugly anyway. Composing the passes by fusing the traversals and transformations would decrease the number of memory accesses, speed up the compilation process, and encourage the compiler writer to write more fine-grained passes, in the same sense that deep inlining in modern language implementations encourages the programmer to write small and reusable routines, even higher-order ones, without severe performance penalties. Lowering the barrier to implementing such a problem-specific language seems to make such an approach viable, perhaps even desirable, given how convoluted most production compilers seem to be. (If I've just written something that amounts to complete gibberish, please shoot me. I just felt like writing down an idea that occurred to me recently and bouncing it off somebody.) - Gath [1] Kent Dybvig, A nanopass framework for compiler education (2005), http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.72.5578 ___ fonc mailing list fonc@vpri.org http://vpri.org/mailman/listinfo/fonc
Re: [fonc] Natural Language Wins
On Fri, Apr 5, 2013 at 7:59 PM, Kirk Fraser overcomer@gmail.com wrote: I was pointing out that innovation for its own sake is worthless then was agreeing with the view that not all the world's inventions come from people who think in English yet pointing out communicating in English is best for world wide distribution. I don't really know how many Jews who won Nobel Prizes thought in Hebrew, English, or even Russian. But it is as you wrote possible that Hebrew is more efficient. No, it's not. Whorfianism has been all but refuted. The only area in which the idea hold water, quite ironically, is formal/computer/programming languages (or so Paul Graham says, but he's right, as far as I can tell). Something about their culture tends to be productive compared to others. Perhaps it's their orientation toward God, which is defined as absolute spiritual perfection. That in itself would tend to produce more efficient thought. They've been oppressed by intellectually impoverished Christians for two millennia, denied the right to work in the fields of agriculture and crafts, and were forced to work in knowledge oriented professions such as medicine or finances. Of course that this nurtures a specific culture, and with the (most likely involuntary) need to become as indispensable for others as possible in over to avoid getting killed by hilt-happy Easter celebrators, they were virtually forced into what is usually referred to as overachievement (although here I have to admit, despite my former point, that you English people have the weirdest notions in your language). English has a property that unfortunately allows it to be redefined with liberal definitions which are inefficient. ^^^ This is a thoroughly nonsensical and meaningless statement. Computers need smarter software to exceed the performance of Watson and OpenCyc to create worthwhile innovations automatically. I think working to automate Bible analysis is an efficient way to produce smarter software. But based the failures of automatic translators, computers may be slow to think flawlessly. Again, you're completely ignoring the actual nature of speech, demonstrated in such phenomena as the existence of idiolects, referential indeterminacy, diachronic shifts etc. Language is what it is because there's a common sense component to its processing in our brains, and once you have that, you've successfully replicated a human being in silicon. Until that happens, all bets are off. (I'm tempted to wager that the inverse also holds, has_human_intelligence(X) :- understands_language(X). Although the fact that an average human being picked from your general population often fails at simple logical reasoning sort of suggests that the intelligence is of a slightly different kind that what we usually mean by saying he's intelligent/he's a genius.) - Gath ___ fonc mailing list fonc@vpri.org http://vpri.org/mailman/listinfo/fonc
Re: [fonc] Natural Language Wins
The first math language Fortran was soon displaced in business by more readable code afforded by Cobol's longer variable names. Fortran was displaced in business because early Fortran had no structures and random record-oriented file access, and because of some silly government requirements for computer system procurement. There are three basic statements in any computer language: assignment, If then else, and loop. ...except for those languages that have none of these three? I'd rather argue that all languages have 1) primitives, 2) means of composition, 3) means of abstraction. Some languages lack the third (Excel?) but these are not especially useful on large scale. Now to complete the project without corporate resources, it is necessary to select an NLP application which is both more powerful and physically smaller than IBM's Watson which won against Jeopardy's best players. The most powerful NLP text in history is the Bible which is only 4 Mb instead of Watson's 4 Tb. I have absolutely no idea what powerful is supposed to mean in this context, but I'd bet the Reuters corpora against the Bible any day of the week. Bible sounds like a horrible source material for any automated NLP endeavor, no matter whether research oriented or production-oriented, since it's on all levels (lexical, semantic, factual) schizophrenically disconnected from modern textual material. This level of NLP mastery in or external to an outside and indoor robot could be used to end poverty, illiteracy, crime, terrorism, and war around the world by growing and serving food, educating and entertaining a family with the same language and religion cradle to Ph.D what? O_o; - Gath ___ fonc mailing list fonc@vpri.org http://vpri.org/mailman/listinfo/fonc
Re: [fonc] Natural Language Wins
On Thu, Apr 4, 2013 at 11:13 PM, Kirk Fraser overcomer@gmail.comwrote: Liberal dictionaries have definitions that are by default wrong. There's no such thing as liberal dictionaries. For evidence of language decay, read definitions from the 1988 Webster's Collegiate vs. the current Webster's. Pure word and definition is needed to understand truth. There's no such thing as pure words. Language is a dynamic, evolving, feedback-driven entity that grows and adapts to new conditions, with meanings of words broadening (dog), narrowing (hound), shifting(computer) etc. People who love to lie get along without words meaning things. ...I won't comment on that nonsense. For example the current political fight on marriage demonstrates some people couldn't care less for truth, only for employer's spouse benefits to be shared with roommates. Political fight on marriage? I don't live in the US, so I have little understanding what you're talking about, but the word marriage seems to be applied in most cultures over the globe for some sort of binding social contract between individuals related to nurturing younglings for the next generation, yielding vastly different rights and obligations from such union across the different cultures. This makes the meaning of the word marriage highly contextual. (But I admit freely that my understanding of cultural anthropology is limited to having skimmed through the Encyclopedia of World Cultures. It was worth it, though - and quite fascinating at that.) - Gath ___ fonc mailing list fonc@vpri.org http://vpri.org/mailman/listinfo/fonc
Re: [fonc] education experimental target
On Sun, Mar 3, 2013 at 5:40 PM, David Girle davidgi...@gmail.com wrote: Given the interest the Raspberry Pi is enjoying in education, the new platform coming out of TI towards the end of April, might be an interesting target for any fonc experiments runnable on ARM. *http://beagleboard.org/unzipped/ * Is this going to require another dose of proprietary binary blobs? With Pi, you at least have to prospect of being able to compile your graphics stuff from Nile into something that actually uses the graphics hardware the way it's supposed to be used. (For the same reason, current AMD APU offerings seem much more. At least you have the documentation.) Cheers, Gath ___ fonc mailing list fonc@vpri.org http://vpri.org/mailman/listinfo/fonc
Re: [fonc] education experimental target
On Sun, Mar 3, 2013 at 8:02 PM, Loup Vaillant-David l...@loup-vaillant.frwrote: Err, what did you mean exactly? Could Nile use hardware graphics on the rPi? Why (not)? My current understanding is that to do this, we would have to interface with the rPi's firmware blob, effectively compiling Nile into using using openGL. It looks that it's possible, though not optimal (not only in terms of speed, but also because it may require more code). Is that correct? You ought to be able to compile it directly into the GPU's instruction set: https://github.com/hermanhermitage/videocoreiv https://github.com/hermanhermitage/videocoreiv/wiki/VideoCore-IV-Programmers-Manual If we can't do that, it's a waste of good silicon. If I were to port the FoNC SW onto something new, I'd never go into HW without public specs. I have quite a lot of ideas for applications (AI, ML, OCR etc.) that would benefit from extra processing power. So much power, and it should go to waste? The very idea just gives me the creeps. But I somehow can't avoid the feeling that from this perspective, anything ARM-based on the market is still inferior to stuff like AMD T40E and the like. Do you know of any truly FLOSS friendly ARM SoCs, by any chance? Gath ___ fonc mailing list fonc@vpri.org http://vpri.org/mailman/listinfo/fonc
