Title: A (Fractal) Theory Of Everything? Post by: Gary Gaulin on January 03, 2010, 07:53:12 AM I could not resist explaining something I know about that would make the ultimate collaboration project for a fractal forum. Just in case anyone else here is up to this challenge.
I have known about fractals and their basics but am relatively new to their details. From what I read here and elsewhere I appear to be working with one but never knew what to call it. Which brought me here looking for more information on them. It still appears to be a fractal, but other opinions on that are welcome. My primary occupation is in the graphic arts (printing) and to some degree science. With the economic crisis having devastated local industry and there being occasional money in software I write I went back to work on one of the projects that I have been experimenting with over the years that figures out how to "crosscut" paper to get the most cut pieces out of a given sheet possible. (http://3.bp.blogspot.com/_Gyn0-PcXrJY/Sz4kBtjYlHI/AAAAAAAAAuo/YdPjMQUXeII/s320/2x5.JPG) After reading Paul N. Lee's excellent definition of fractal (http://www.Nahee.com/PNL/Fractals.html) I better add that since it's only necessary to show the layouts with the greatest number out most of the fractal is never shown. It would be possible to have infinite layouts inside of layouts (and in the trims) that keep the same proportions of full to cut dimensions, but a printer just wants to know how to cut their stock down to size (not make fractal art) so that recursion is not included. The algorithm is now decribed here: http://crosscutfractal.blogspot.com/ If it is possible to conclude that it is not a fractal then I can zap the blog. It's not necessary for it to be one, it's just that a fractal is the only thing in math/science that I have ever found to explain it. And it very much relates to other recursive related phenomena. One of the most interesting comes from 1970's robotics, the work of David Heiserman who wrote books on building relatively simple electronically hardwired self-learning systems that model the behavior of biological systems that makes it to the protozoan or simple insect level. Since it's a lot cheaper and easier to model on a personal computer (that did not exist back then) than build a board for a robot to crash around the house I studied/simplified the algorithm in cyberspace. A tutorial on how it works with VB6 source code is the link below. http://intelligencegenerator.blogspot.com/ Self-similarity should be observable in a computer model of a cell where each of its molecules is its own intelligence loop learning to behave like its biochemical counterpart. The cell's behavior, something that fractaled out of that. Where you have many aggregating cells in the model able to change in form to adapt to survive you have stem cells that differentiate into a complex multicellular organism. Here's the loop from the above blog that through fractal recursion (and with computing power we only wish we had) could theoretically form societies of virtual plants and creatures: Code: LoopStart: It is important to notice it is "confidence driven" (like we are) and only needs to know how successful it was. And you are not modeling the exact physical shape of things, just their behavior which in turn can define form. Would here have to predict that a complex organism at our scale that could fractal out of these virtual atoms or molecules would also be confidence driven, without having to design that into their brain (which would likewise fractal out of the molecular behavior without us having to first design its brain ourselves). The "circuit" that it is simulating has two way communication like our brain to our motor muscles that have neuron to turn it on/off and another for feedback to see how well it's doing in response, which would predict two way muscle control will emerge on its own by it already being present in the behavior of the fractal it came from. A massively parallel computer to fully test the idea does not yet exist, but PC's are still powerful enough to get started. And instead of starting at molecules, the loop could model one cell body each but require software that simulates an even more complex system. The least complex would be where the emergent behavior of matter (hence living things) begins, that scientists would love to discover more about with the new CERN superconducting supercollider. Anyway, if you're looking for a challenging project then that is easily able to keep you busy for a lifetime too. Should be possible to go from what I have at this point to a new fractal theory to explain how to model the behavior of natural systems from quarks on up to stars. A theory of everything, that explains a fractal that fractals into everything. Or at least demonstrates how to get a part of the way there in a PC... Title: Re: A (Fractal) Theory Of Everything? Post by: Nahee_Enterprises on January 05, 2010, 11:03:18 AM I could not resist explaining something I know about.... My primary occupation is in the graphic arts (printing).... ....I have been experimenting with over the years that figures out how to "crosscut" paper to get the most cut pieces out of a given sheet possible. An interesting project, two interesting blogs, and some interesting ideas. I only had a few moments to glance over all of it, including the blogs, so will have to come back when I can spend a bit more time. Just wanted to say Greetings, and Welcome you to this particular Forum !!!! :) After reading Paul N. Lee's excellent definition of fractal (http://www.Nahee.com/PNL/Fractals.html) I better add that since it's only necessary to show the layouts with the greatest number out most of the fractal is never shown. Very kind of you to say so, but the first paragraph of that definition is mainly from a dictionary I was using way back when. The rest of it is my own writing though. Title: Re: A (Fractal) Theory Of Everything? Post by: Gary Gaulin on January 06, 2010, 03:59:12 PM Hi Paul, thanks for responding!
It was this later paragraph that was most useful. Wikipedia and other sources will cover it but it's hard to tell whether that is what people who experiment with fractals are normally working with, and the accuracy of information. The way you explained it here was all I needed to begin to make sense of all the formulas that I have been seeing: The standard Mandelbrot fractal equation takes the form z(n+1) = z(n)^2 + c, where c is the complex number x+iy corresponding to any point on the (x,y) coordinate plane. Fractal equations are iterative, in that the result of one calculation of the fractal equation becomes the z input to the next calculation. Over repeated evaluations of a fractal equation, values for each point in the (x,y) coordinate space either converge at single points, move toward the (0,0) origin point, or move toward infinity. The diverse colors in fractal plots reflect the rate of this movement for each point. Discussions of chaos theory frequently use fractals as examples, because slight variations in the fractal equation produce radically different results. I later found "Introduction to the Mandelbrot Set, A guide for people with little math experience. By David Dewey" (http://www.ddewey.net/mandelbrot/) which helped make sense of the imaginary numbers and complex number plane. I also looked at how fractal trees were calculated. But the method is so entirely different and in a way such a comparatively simple draw that along with what I read on Wiki I have to (with some humor) wonder whether plotting the points of a circle might also somehow qualify as drawing a fractal. In an intelligence algorithm, iteration is used to keep it going from frame to frame (time). The behavior of the pixels/entities produce the final image (as opposed to behavior of an equation that places the pixels) like in this experiment: (http://1.bp.blogspot.com/_Gyn0-PcXrJY/SEeCUd2y60I/AAAAAAAAAC4/PCnXaHr3yHc/s400/Training500.JPG) http://selflearningbots.blogspot.com/ In this case, to draw a Mandelbrot fractal the intelligence generator algorithm would use the Mandelbrot equation to determine whether the last step each entity/pixel took was successful in getting closer to forming one, or it failed. If successful then confidence is incremented, else decremented. In time they would learn to behave like the equation. Of course a "fractal of everything" should like in matter produce competing entities that increase in complexity to become competing entities at the next highest level, then the next. The iteration would be the same as you mentioned "Fractal equations are iterative, in that the result of one calculation of the fractal equation becomes the z input to the next calculation." but since an intelligence generator algorithm places/moves the particles it does not need an equation to form an image which at least simplifies the "complex number" part of the equation for those who have a hard time imagining an imaginary number! :D Title: Re: A (Fractal) Theory Of Everything? Post by: kram1032 on January 06, 2010, 07:07:07 PM nice stuff :)
I wonder what OpenCL could do here :) Title: Re: A (Fractal) Theory Of Everything? Post by: Gary Gaulin on January 07, 2010, 03:46:40 PM nice stuff :) I wonder what OpenCL could do here :) Thanks for the compliment! And you must jest about OpenCL. High performance multi-core parallel computing with OpenGL 3D would be the ultimate! Quote OpenCL (Open Computing Language) is a framework for writing programs that execute across heterogeneous platforms consisting of CPUs, GPUs, and other processors. OpenCL includes a language (based on C99) for writing kernels (functions that execute on OpenCL devices), plus APIs that are used to define and then control the platforms. OpenCL provides parallel computing using task-based and data-based parallelism. OpenCL is analogous to the open industry standards OpenGL and OpenAL, for 3D graphics and computer audio, respectively. OpenCL extends the power of the GPU beyond graphics (GPGPU). OpenCL is managed by the non-profit technology consortium Khronos Group. http://en.wikipedia.org/wiki/OpenCL Do you program in OpenCL? If yes, then give the 3D "Particles In A Box Bots" idea a try! (http://2.bp.blogspot.com/_Gyn0-PcXrJY/SGePcnumWWI/AAAAAAAAADI/bXnwWjxDfTU/s400/ParticlesInABoxScreen.jpg) http://selflearningbots.blogspot.com/ I would love to see that recoded into a language worth migrating to from VB6 which still compiles into fast loops and all but the graphics slows it way down. It is far from optimized by setting single bits with "+ (YMR(N) * 4) +" and such that could be performed with an "Or" and use "And" also but I try to avoid making it harder for others to figure out the code. The main loop that does the thinking is the last subroutine. Code: Private Sub WriteMainMemory(N As Long) Ideally the "MainLoop" should be reduced to four lines so that the four "requirements" that the system has stay sorted out and simple to visualize and correlate with how this relates to biology. And please don't mind my pushing the limits of theory by the ending of this one but where scientifically possible, I can't resist: Quote From nonrandom behavior of matter comes a progression of self-assembling emergent behaviors where at the molecular, cellular and multicellular levels each is an increasingly complex fully autonomous self-learning associative memory confidence driven intelligence system that in turn produces fractal-similar emergence at the next intelligent level on up to us. Computer models of this common intelligence system that is present at each level shows its mechanism reduces to four necessary requirements; Something for intelligence to control (motors, muscles, metabolic cycle), sensory addressable memory to store motor actions in response, feedback to gauge failure or success in actions taken, and a guess mechanism that can try a new action which may include good guesses such as in crossover exchange recombination that makes offspring a little different from each other (not clones) and recombining of small conserved domains that are the small nuts and bolts and motors of complex molecular machinery that all together keep living things alive. Designs that successfully reproduce remain in memory in the population (gene pool) to keep going the billions year old learning process that is the cycle of life where through continual reproduction of previous state of genetic memory one replication at a time builds upon previous designs in memory. Thus a cladogram of resultant lineage shows a progression of adapting designs evidenced by the fossil record where never once was there not a predecessor of like design present in memory for the descendant design to have come from. It is this progression of intelligent causality from nonrandom subatomic behavior in matter that makes possible the complexity of cells, speciation, Cambrian Explosion and all existing biodiversity. Without this intelligent cause, living things that we now see would not exist. This in as few words as possible explains how to detect the confidence driven behavior that is in the system, which is relatively easy to do at the program level in computer models because the variables influenced by it (like battery/energy level, hungry, overall confidence) only need to be monitored: Quote Intelligence Detection To determine whether a system is "intelligent" we look for the previously explained "four requirements" that must be met. This will rule out (as being intelligent) a simple heating thermostat, toilet float and water molecules. Where a system does not include all requirements for intelligence we have a system with a "behavior" that may possibly appear to be intelligent but would not qualify as an "intelligence system" or "intelligence". The four requirements will be listed as: REQUIREMENT #1 of 4 - SOMETHING TO CONTROL REQUIREMENT #2 of 4 - ADDRESSABLE MEMORY REQUIREMENT #3 of 4 - FEEDBACK TO GAUGE FAILURE AND SUCCESS REQUIREMENT #4 of 4 - ABILITY TO TAKE A GUESS Where all four processes required to produce intelligence are properly functioning together there should also be detectable synchronized cycles indicative of proper functioning and good health. Where these cycles are no longer present the intelligence is then nonfunctional. The above will be needed after the particles are behaving in a way that an emergent level above it begins to form, which is not expected to be as easy to detect but we still have its whole world reduced to variables. Just need to know what to look for in the data that needs to be put on the screen that will rule out say an image of a panda that looks alive when the model was more like forming a cloudlike substance with no memory of any kind to tap into (thus ruled out as intelligent) that just happened to take the shape of one. In biology Self-Replicating RNA's are a very simple nucleotide coded memory system, with an analogy that might form in the virtual environment where there is no doubt they are coded entities competing with other and getting better at controlling each other and their environment as time goes on. Prions are proteins that compete with each other and their environment much the same way but not enough seems to be known about them to rule one way or the other. Hopefully that helped explain missing details you or anyone else needs to know to program it. And if a new more fractal based theory is possible I'll leave that up to consensus because the title of the one I already have is overwhelming enough. To make a long story short the releasing of the Ben Stein movie that made it seem like the Discovery Institute had a useful intelligence theory stirred me up real good. After weeks of brainstorming an idea to counter it, all the science that was scattered around blogs and ftp came together in my mind then I wrote it down before forgetting how the logic connected together to become the theory they asked for. Only problem is it's a theory that's supposed to be impossible to write and illegal in the school district of Dover, PA so I made sure the teachers there were the first to know through their local newspaper forum so that reporters and others will know what it is and not worry about it. I was also published for a self-assembly experiment that developed in another forum (and was inspired by a few Creationists) that was published by the National Science Teacher Association (#1 science education journal) (http://www.nsta.org/store/product_detail.aspx?id=10.2505/4/tst07_074_07_72) to help introduce the self-assembly concept to US schools. Since it demonstrates an entirely "nonrandom" process the dwelling on the origin of life having been from "random" processes met its match with just a little egg yolk in oil in water and shaken into phospholipid membranes as makes our cell membranes. All cellular organelles replicate without genetic code just self-assembly that kinda goes "Poof!" and it's there, can't evolve. So even though it's just science and I cannot see myself as a Creationist (just follow the science where it leads) it just sometimes goes in very surprising directions. But those who genuinely want the controversy to end with everyone having learned something can see who is making progress and who is not. And it is vitally important that a lot of honest people (who are not looking to destroy science ) who put a lot of faith in a theory of ID understand why Dover went so badly for the Discovery Institute's theory. It was a "premise" for a theory and arguments against another when a theory has to stand on it's own scientific merit. Must explain how the stated phenomena works in a way that others can experiment with it. In time that becomes a part of science without needing to influence school boards or anything messy like that. Power the Discovery Institute could only wish it had, which only belongs to those who are genuinely into science entirely for science's sake. When you are and you know how to evidence what supposedly 99% of scientists agree was impossible it's impossible not to try to at least see what does develop. It's one way of solving a complex scientific problem where you start off with the goal to evidence something you only have an abstract generalization for that describes it in this case "Intelligent Cause" then do the best you can to reach that goal. Where you fall short you still end up further than anyone has ever been along that path of discovery, so you write up what you have minus the far reaching goal that would be out of place where it's obvious to others that it was not really evidenced. But in this case "Intelligent Cause" is probably the best two word phrase to describe the type of causation being experimented with and theories change how words and phrases are defined, not the other way around. Since you might find (or already found) the http://theoryofid.blogspot.com/ site I had to explain what that is for. It currently has the info on previously mentioned RNA's and other things pertaining to biology and other sciences plus origin of life experiments. A new theory or great paper that could be made in this forum would be worded in with as few sentences as possible then linked to in its References like they normally do in science papers. Gives credit where due and don't have to cover the same ground all over again. And whatever may be collaborated is yours as separate without being the other, so that none of us have the strings that are attached to the other theory attached to us when just working with fractal math/science none have good reason to get all shook up over. Title: Re: A (Fractal) Theory Of Everything? Post by: kram1032 on January 07, 2010, 09:34:40 PM I fear not lol
My current graphics card doesn't even support openCL ^^ However a combination of openCL and openGL could not only increase the performance by, like, an infinite times, but at the same time could give an environment which is graphically more meaningful to a, uh.... average person :) I just saw some demonstrations of OpenCL and this kind of calculation must be about perfect for the advantages. Those particles by now must have learned quite a bit as they don't hit the walls that often anymore and start to form groups here :) Right now I'm using 7 Particles (none stationary) and a world size of 11. Current Memory-count is just over 4000 but I think it didn't save the memory count from yesterday. I just let it ran without drawing during surfing and dining and stuff and just checked draw from time to time to see if something changed. Pure randomness by now is definitely gone :) Maybe with some optimization you could do massiv simulations with a million particles or so - the theoretical mechanics behind it don't look too complex. There where more complicated particle systems made with reasonablish render times. Some youtube-vids with interactive particles on Cuda or OpenCL can be found :) A rather simple thing like this, not even requiring complex formulas could hit amazing performance that way. - However I dunno which Graphics Cards they used on those YouTube vids... Maybe two cards at once or all the possible craziness... (Too bad I can't use either, using a 2-year-old ATI^^) What would be interesting too would be some kind of learning BOIDS with, say, 5 species with different pretador/prey configurations but each species at dirst has to learn how to move nicely like that single robot has to to find his food-sources (the charging platforms) One could be omnivorous, one could be a plant-eater (forgot the nice name xD) and the other three differently sized carnivores "eating" each other with different conditions, for instance the bigger can only be eaten if there are more of the smaller... Plants would basically spawn randomly. However that could easily be way more complex as you'd also have to consider dynamic particle numbers when an individuum dies or gets born and when they get born. Basically learning BOIDS with the abillity to reproduce. At first, both for simplicity and actually also closer correlation to the real world, just do that on a toric or if somehow possible spherical 2D world :) (Talking about kinds of worlds: I once looked into how a world on a klein-bottle would look like, rather than a torus. It would be like on a torus, you go out of the world on one side and come back in on the other but the side you'd walk would be mirrored. So if you go out of the world at the left side of the top edge, you would come back in at the right side of the bottom edge. Such a world could also be interesting :) ) I'm not sure if I entirely followed you with all your descriptions but if I got it right, you'd like to kind of copy the DNA's coding and just let it run so intelligent-ish structures start to emerge. Very interesting but looking at how long just such a simple system as the Particles in a Box takes to learn, something like the whole evolution will take for ever even with openCL except if you have a huge computation farm or something ^^ However you could and probably should do some computational optimizations behind the structure of computer-data: In DNA, information is stored in triplets basically and each combination has some double meanings. Those triplets form amino acids which then in one way or an other allow communication in the whole body. DNA by that doesn't really follow a numberic approach but rather a boolean-ish just with a more diffuse system. The double-meanings for instance help to at one hand allow variations which at the other hand don't change anything at all. For computers, such a unit wouldn't be a triplet but an octet - the byte. Interesting would be if you could even find some rules to let it order itself. Find some basic rules which lead to self organizing rules. Let it define on itself, how many codes need to be given twice or even trice and what each code actually means. Seen like that, defining two things (00 and 11) as stop is actually a natural thing as that's just what nature does to ensure greater stabillity. All that is very interesting but the sheer complexity has an attractor at infinty :) Title: Re: A (Fractal) Theory Of Everything? Post by: Gary Gaulin on January 08, 2010, 07:39:33 PM I fear not lol My current graphics card doesn't even support openCL ^^ I have a Celeron CPU so I'm early Pentium. The program speeds things up with the "Draw" checkbox that skips all the graphics to go 20 times or so faster. Even though graphics are slower it's only needed to peek in every once in a while and where it goes much faster needs slowing down to see what the particles are doing. Works good enough for what we now have for hardware, and software. Those particles by now must have learned quite a bit as they don't hit the walls that often anymore and start to form groups here :) Right now I'm using 7 Particles (none stationary) and a world size of 11. Current Memory-count is just over 4000 but I think it didn't save the memory count from yesterday. I just let it ran without drawing during surfing and dining and stuff and just checked draw from time to time to see if something changed. Pure randomness by now is definitely gone :) I left mine on overnight. Used 100 particles with the Inhibit Reversals unchecked, then when I got back to it they were going in tight circles back to start and jittering one step forward then back. They are all perfect solutions to the problem of avoiding walls but keep moving while staying together, achieved 100% success. Maybe with some optimization you could do massiv simulations with a million particles or so - the theoretical mechanics behind it don't look too complex. What needs working on most now is the way the confidence is adjusted, to better approximate the properties of an atom. Otherwise not have anything worth trying a million particles for. What would be interesting too would be some kind of learning BOIDS with, say, 5 species with different pretador/prey configurations but each species at dirst has to learn how to move nicely like that single robot has to to find his food-sources (the charging platforms) One could be omnivorous, one could be a plant-eater (forgot the nice name xD) and the other three differently sized carnivores "eating" each other with different conditions, for instance the bigger can only be eaten if there are more of the smaller... Plants would basically spawn randomly. However that could easily be way more complex as you'd also have to consider dynamic particle numbers when an individuum dies or gets born and when they get born. Basically learning BOIDS with the abillity to reproduce. You have the right idea. And the ability to reproduce would be essential. It's such a part of what keeps our kind of intelligence going I just finished having to write about (PG13) where males/female sexuality comes from (http://www.kcfs.org/forums/viewtopic.php?p=13455#13455) which is important to know to be able to model it properly. Can say that in the beginning, were hermaphrodites. And they're still here. So if they're not in the model then it's not a realistic approximation of the real thing, for when that is trying to be achieved. Once the particles behave like atoms they can be assembled into proteins (made by hydrogen-bonding amino acids together) and ribonucleotides to make prions and self-replicating RNA flocking around certain things without worrying about having to add BOIDS behavior. It might be possible to derive a Confidence +1/-1 number from valence charts and whatever else is on the internet. I also have a Lennard-Jones model that makes them attract and repel properly. Forms the predictable hexagonal close-packing pattern. I'll post the code for it here, in case anyone needs it. Plenty on the internet on it too. The formula is: V = 4 * Epsilon * ((Sigma / R) ^ 12 - (Sigma / R) ^ 6) V is the velocity, same thing as how many pixels to move one way or another. Others adjust size of particle and attraction strength between them. I put it on the screen in 2D with this: Code: Const XYpixels = 512 (Talking about kinds of worlds: I once looked into how a world on a klein-bottle would look like, rather than a torus. It would be like on a torus, you go out of the world on one side and come back in on the other but the side you'd walk would be mirrored. So if you go out of the world at the left side of the top edge, you would come back in at the right side of the bottom edge. Such a world could also be interesting :) ) A physics mind too! I kinda expected that. I'm not sure if I entirely followed you with all your descriptions but if I got it right, you'd like to kind of copy the DNA's coding and just let it run so intelligent-ish structures start to emerge. Very interesting but looking at how long just such a simple system as the Particles in a Box takes to learn, something like the whole evolution will take for ever even with openCL except if you have a huge computation farm or something ^^ Only need to get something like Self-Replicating DNA on the screen and even Jerald Joyce would be here checking that one out, and I'm not kidding either. But to make that we need to make a single ribonucleotide which is not that many atoms/particles. So I keep it down to something that we can do right now with whatever we already code in, by only needing to only make small molecules. However you could and probably should do some computational optimizations behind the structure of computer-data: In DNA, information is stored in triplets basically and each combination has some double meanings. Those triplets form amino acids which then in one way or an other allow communication in the whole body. All that would be due to arrangement of atoms in molecule. Would happen on its own without needing to be put in, just by having the behavior of atoms in the particles well enough approximated. All that is very interesting but the sheer complexity has an attractor at infinty :) I think I know what you are saying about the attractor, maybe not. But I look at it as just needing to get a single +1/-1 number to adjust confidence by from a subroutine that should be possible by compiling valence and other chemical data. Not have to care when it is solid liquid or gas just the basic dynamics built into the bot so it gets the Lennard-Jones attraction (throttle of its X,Y,Z Motors) get set in the right directions, does not have to be precise value just right direction. Otherwise have the problems you mentioned where there is so much calculating needing to be done it's beyond what we could model on our PC. Here, once the memory has been trained the bots only read that to find out what to do in response (direction to set motors) so can go fast enough to see something happen. Even small molecules behaving like the real thing would be amazing. Anything having to do with heat and pressure is easy. Should be able to set the temperature scroll by finding what level water molecules become ice. And heat vibration is like the old Rock Em' Sock Em Robot's that moved on vibrating table, need that but its easy. I know that we do have to account for Valence. But that's easy too, here's the numbers: Code: 'Number Element Valence Radii too: Code: 'From: http://environmentalchemistry.com/yogi/periodic/atomicradius.html And Electron Configuration: Code:
Names are good to have too: Code: 1 Hydrogen H Now that all the data files that I found are here, you can see what there is to design a bot from. Let me know where you know of something missing that needs to be added. If there is then it doesn't seem like there's much more needing to be accounted for. Most everything else like density seems calculable from that. Might be easiest to start by designing is by the way memory Addressing would have to look. Data stored at each Address location is just the X,Y,Z motor settings so the rest after that point in the circuit is easy. Going the other way from Addressing is Sensory to neighbors in a 3D array system that does not need to calculate far away particles. Later can tile interactions, like to make a vesicle one tile is repeated around the sphere that from there holds itself roughly in that shape depending on heat and what it is attracted to or repelled from in the various directions. There can sum up atom on up behavior of phospholipids and such in one bot. But first need the atom bots that make summing up molecule bot behavior easy. I have also connected one memory system so that Sensory addresses Data, that is Sensory for the next where it's in a sense summed up so that only necessary detail be sent on down the line to where eventually Data goes straight (with added entropy amount) to X,Y,Z motors (move pixel that amount on screen). I like to keep things as simple as possible, so that what I propose is not expensive or hard to code. What makes it scientifically novel is the way it works, the confidence/behavior/intelligence algorithm and theory it is part of. Hard part is explaining what makes it all relatively straightforward. And explaining my sordid scientific past that includes still entertaining that taboo theory but I cannot help myself, it's too useful for helping to make predictions that lead to science that is not written anywhere yet. Besides, even university level intelligence science took a hammering by just having the "I" word in it. But in that area evolutionary theory that now exists can really only say that things change over time and along with it intelligence evolves and that's honestly about it, and I know from having been one that tried to explain origin of intelligence with existing theory with 0 success convincing someone of that. They wanted to know where love and emotions plus consciousness came from and are for, not be told "natural selection did it" for the 250'th time. ET simply cannot go there. Needed a theory that can predict a starting state of hermaphrodite which is somewhat contrary to what the prevailing wisdom saw as the starting point for reproduction on up to our level which means their models are boring and incomplete. And my wife was laughing to tears reading where males and females come from in a link above, so it obviously helps make science fun which is important due to getting people to read something scientific is not easy either. I'm glad there are exceptions here. In edit (for anyone trying this) I have to add that later on energy levels will need to be added. Increasing the energy level of oxygen should shift the virtual orbits of electrons to bond to form O3 ozone, in the unenergized O2 environment. Title: Re: A (Fractal) Theory Of Everything? Post by: kram1032 on January 11, 2010, 10:50:10 PM hmmm.... Do you think the "atoms" would "learn" faster if you split the things to be learned into different memory files, running different simmulations for each with different things being considered, rather than throw the "full problem" on them?
It would be cool if you could get rid of the grid so the movements could become "smooth" :) Title: Re: A (Fractal) Theory Of Everything? Post by: Gary Gaulin on January 12, 2010, 01:23:14 PM hmmm.... Do you think the "atoms" would "learn" faster if you split the things to be learned into different memory files, running different simmulations for each with different things being considered, rather than throw the "full problem" on them? Good thinking! I thought about that too. From what I have seen from their behavior it might take a little longer to train them one problem at a time (but never confirmed it). An analogy for why, would be learning to jump across to the other side of something by going full speed at it. It would not take long to learn how to do that. Then we could add the next problem where if they go too fast then they bash into the wall just on the other side. Where each has its own memory (not needed in this model since identical particles have identical behaviors) they would all crash into the wall from being so used to there not being one there. Same thing as pulling a chair from under someone just before they sit down which even causes injury it's such a "surprise". They are not used to that happening because chairs do not normally move. Where they did they would be more careful and "keep an eye on it" while sitting. As a result all the time training them to be very successful at one task would make them immediately terrible at another that has changing parameters, in which case they become like lemmings following each other off a cliff. Might as well try to pull the chair from under them while first learning to use one so that they know it can happen when a prankster is behind them, so are prepared for it then do the right thing every time. Dividing the problem is though very important to how memory is constructed. For example there is the ring of neurons that gives awareness of where something went that it cannot see, which adds a very simple memory circuit to the one that does the actual thinking which greatly increases navigation success. In that case the full problem only needed to be divided at the bot design level. It would be cool if you could get rid of the grid so the movements could become "smooth" :) I agree. And it's not that hard to do. Just have to use a different world-array system where their velocity is a variable derived from motor acceleration and direction stored as precise x,y,z points in space (as opposed to one of the possible squares or cubes in the world to step to). This is great because the world can then be as large as you want and motion would be smooth, but the drawbacks are slower speed caused by having to calculate all movements and positions in physical space instead of an instant step in one of the possible directions. Also, the properties of atoms include integer value "harmonics" that double like *2, *4, *8 or triple *1, *3, which suggests that a grid might be able to do a better job than it appears. And you might have noticed that when the particles become unstable they fly off in one or more directions with a repeating wave with a frequency that is here in the shape of triangular, square, or combination that produces a "wavelength" and the particle is acting as both a particle and a wave (which as you may know is still a big mystery of science to fully understand). Even though the model has low granularity, that might be all it needs. So instead of starting with something that has been complicated by an algorithm that adds its properties to those of the algorithm that produces intelligent behavior I started with the least complex method. From what I can see it accounts for mass and other properties that could also be calculated into it but doing that properly would require knowing what is already being expressed by the particles. Guess you can say that in this case it's best to learn to walk before running or else all that would be expected are stumbling bots. Title: Re: A (Fractal) Theory Of Everything? Post by: kram1032 on January 12, 2010, 02:52:03 PM I see :)
I have yet an other idea: Right now, your particles totally rely on the confidence-system as some kind of fixed (hardcoded) base of intelligence. But what if you would add the confidence-system inside the learning recursion? Do some kind of general version where the "optimal confidence behaviour" would be learned on the fly during trying to learn the "rules of the game" So, not only learn to behave intelligent but also learn to actually learn :) If you look at all the confidence systems you tried so far, do you see some kind of generic pattern which could be generalized to a learnable method? Confidence-driven confidence behaviour? :) Title: Re: A (Fractal) Theory Of Everything? Post by: LesPaul on January 12, 2010, 10:47:03 PM Gary, your initial post sounds rather similar to concepts seen frequently in "genetic algorithms" and "genetic programming." These might be really interesting to you if you're looking for further research.
As a very high-level introduction, genetic algorithms are computer programs that find their solution by random guessing, but refined by a "fitness function." This sounds very much like what you're describing. There are many techniques in genetic algorithms that I think you would find very useful -- things like "crossover" and "mutation" and "selectivity" play very important roles in helping the population of random guesses approach the best solution. To take a step further, you come to "genetic programming." In a genetic algorithm, the actual program is fixed (written by a human) and it uses the method described above to find the solution to a problem, for example, the best shape of an antenna. In genetic programming, the program itself is not fixed and is actually evolved over time. So, little bits of code like a for-loop or an if-else statement make up the population and they are randomly smashed together. Here, the "fitness function" is simply the measure of how close the program comes to solving the problem you're trying to solve. You might think that just random pieces of code would take billions of years to actually do anything useful, but it is surprising how effective this technique can be. For example, John Koza (http://www.genetic-programming.org (http://www.genetic-programming.org)) has created a sort of home-brew supercomputer that runs genetic programs all day long, and it has produced results substantial enough that his machine has been granted a patent! I don't mean that the machine itself is patented, I mean that the machine actually invented something that was worthy of a patent! It was the first non-human to ever be granted a patent. [edit: fixed "John Koza" typo] Title: Re: A (Fractal) Theory Of Everything? Post by: Gary Gaulin on January 13, 2010, 09:22:41 AM Hi LesPaul, thanks for mentioning all that!
Gary, your initial post sounds rather similar to concepts seen frequently in "genetic algorithms" and "genetic programming." These might be really interesting to you if you're looking for further research. Yes, it is similar to a genetic algorithm. As a very high-level introduction, genetic algorithms are computer programs that find their solution by random guessing, but refined by a "fitness function." This sounds very much like what you're describing. There are many techniques in genetic algorithms that I think you would find very useful -- things like "crossover" and "mutation" and "selectivity" play very important roles in helping the population of random guesses approach the best solution. In this model "random guess" is the most rudimentary guess mechanism, with it being better (but not always necessary) to include "good guess" that can be added a number of ways. Same idea though. Finding the best solution would be necessary for a GA too so in that sense they are the same but it's "confidence driven" which is entirely internal as opposed to "natural selection" or "fitness" that is external. It is either able to survive (possibly reproduce) in the environment or it is not. Modeling a reproducing animal could be accomplished by each gene being a small bot responding to local control molecules and conditions, possibly many genes combined into one memory that acts as one unit of exchange during crossover. Where it needs one, another memory would be it's brain. Since there is university level evidence that the centrosome is also one of these mechanisms it would be used to add complex IR sensing (and what else they sense) guided migration behavior to the cell(s). After you have critters wandering around on the screen you can then be the natural selector by zapping out the slow boring ones. I doubt they would need that to speciate. We're here essentially reducing variation in the population needed for new species to emerge. Zapping out the slow ones would eliminate all hope of there ever being snails, starfish, and other slow creatures in the environment. To take a step further, you come to "genetic programming." In a genetic algorithm, the actual program is fixed (written by a human) and it uses the method described above to find the solution to a problem, for example, the best shape of an antenna. In genetic programming, the program itself is not fixed and is actually evolved over time. So, little bits of code like a for-loop or an if-else statement make up the population and they are randomly smashed together. Here, the "fitness function" is simply the measure of how close the program comes to solving the problem you're trying to solve. In this algorithm there are no loops being tried in new combinations, it is always just a simple Input/Output memory like a RAM chip, neural network, genome. You might think that just random pieces of code would take billions of years to actually do anything useful, but it is surprising how effective this technique can be. For example, John Koza (http://www.genetic-programming.org (http://www.genetic-programming.org)) has created a sort of home-brew supercomputer that runs genetic programs all day long, and it has produced results substantial enough that his machine has been granted a patent! I don't mean that the machine itself is patented, I mean that the machine actually invented something that was worthy of a patent! It was the first non-human to ever be granted a patent. I have to agree that GA's are excellent at designing new inventions. Better fan blades on up. Problem though is they are not inherently confidence driven which is useful in modeling "emergent" biological processes that work the same at each level, which is a new area of science. It is easy enough to adapt a GA to work the same way but then it's being used for a new purpose that has no natural selection or fitness it is how confident the entity in in its actions. On the other side of the equation I don't think that kind of model would be very useful in industry where they want a fan-blade, not a virtual intelligent amoeba sliming around on the screen. Each has its application it's best suited for and required vocabulary. It's great you noticed the similarities. Most completely miss that, even where they claim to be experts in how they work. To also help answer Krams questions this section of the theory helps boil things down to the confidence driven mechanism that is in us that we can feel. Theory predicts that's exactly how our intelligence works too. Same thing but at a greater complexity level. Impossible to make a prediction like this with GA theory. Quote Multicellular Intelligence Multicellular intelligence as produced by a brain operates with clock cycles that can be detected from the outside by tuning to waves with an electroencephalograph (EEG) machine to observe brain-waves. In humans, lifetime learning is shown by academic test scores and personal accomplishments. REQUIREMENT #1 of 4 - SOMETHING TO CONTROL Human intelligence controls a body with muscles. Each muscle has a connection from brain to muscle to control when it will flex, and a second connection back to brain from muscle to sense that it is indeed flexing as commanded. This is seen in the circuit of the computer model by the two bit (opposing muscles) motor on/off Action data (on right) connecting to motors, while on the other side (on left) are two bits added to Sensors to sense that motors are not stalled against a wall or stuck. External to the human body are feedback circuits such as a thermostat to control room temperature. The intelligent cause that created the intelligent design is the human intelligence that ultimately controls the room temperature, not the thermostat that only makes it easier for its human intelligent designer to control it with an electrical feedback system of that design. When it fails to perform its mechanical function it gets mercilessly ripped out of the wall to be replaced by a new one. That right there shows where the intelligence controlling the room temperature actually is (and it is not the mechanical thermostat), and where the thermostat that is a product of human intelligent design came from. Academics control using a reward system by "degrees" and "credentials" which often prevents employment to those who did not "make the grade" even where there are self-learners who have more knowledge and experience from learning and experience while growing up. Human intelligence is here again controlling something for it's own benefit. In this case learning, and knowledge itself. REQUIREMENT #2 of 4 - ADDRESSABLE MEMORY Brain neurons store memories with synapic junctions (synapse) between their many neural fibers that grow from them to other neurons. Addressing is through tree-like networks that route signals to proper places. At first a synapse has a small space between them forming a non-permanent connection between neurons which makes them a "short term memory". With repetition/practice the synapse is permanently fused together by glue-like molecules to become part of a "long term memory". REQUIREMENT #3 of 4 - FEEDBACK TO GAUGE FAILURE AND SUCCESS Confidence changing feedback which gauges failure and success is something we can consciously "feel". At our level we are consciously rewarded by "success" and feel punished by "failure". For that reason games and sports are very popular to achieve the euphoria that accompanies success. By being able to "feel for others" we can share in the success or failure of another intelligence simply by watching them. We therefore have heroes who succeed and villains who fail us. In human culture this is well expressed as winning over the 1970's "pinball machine" that preceeded home video games and personal computers. The pinball machine had to be fed quarters which in turn kept many teenagers out of spending money, which in turn helped make the impossible dream of being able to control the game for endless replays the ultimate success for many of that generation. In the musical movie by the Who named "Tommy" is the song/scene "Tommy and the Pinball Machine" where a "deaf dumb and blind kid" that can only do one thing at all (get endless free games) first discovers a pinball machine that was luring him to wander off. After finding his "calling" the world "lights up" around him by that superhuman success of beating a machine that always eventually won having been achieved. In reality the ability to fully control a pinball machine would not make it function without being plugged into an electrical outlet or can light up the air around them but our human intelligence abstracts things this way because of how the intelligence mechanism inherently seeks to as much as possible control other things and how increased confidence in reward for being able to do so feels good. And after spending much money attempting to control a pinball machine, human intelligence would then spend more money to see a confidence building movie that feels good by showing what that ultimate superhuman control over a controlling machine would look like when abstracted through art. Our intelligence here understands a reality by relating to something that in reality could not possibly happen. What is in human culture is here useful for explaining what is producing human intelligence by these expressions through the images that exist in its art. Low confidence of repeated failure or being held down by others attempting to control us produces an imprisoning "bad feeling" that we will work very hard to "get free" from. This is expressed by the Tommy movie song "I'm Free" where he breaks through to the other side of the mirror he once endlessly stared into. The running through the world inviting all to join him then through scenes that resemble going back in time to a primordial planet is here an expression of confidence level suddenly greatly increasing. Humans have such a need to fill memory with knowledge many feel incomplete especially when it comes to the "big questions" like where we came from and in time will go. Some may seek knowledge from history and/or religion. Scientists may try to answer that by searching for new knowledge scientifically, driven to keep taking their intellect and science to new levels. In fact, that powerful need for knowledge is why this theory exists. In human culture the search for knowledge is often expressed as climbing a mountain for the light of knowledge and wisdom as in the movie Tommy (The Who) - See Me, Feel Me - Listening to You (1975) where after following Tommy because of all wanting what he has tragedy forces aside the controlling of a game what once seemed important and Tommy must now run through the flames then on that long journey to the knowledge that all their lives they suffered to discover. REQUIREMENT #4 of 4 - ABILITY TO TAKE A GUESS Externally human intelligence can take a random guess aided by flipping a coin or draw from deck of cards. A good guess is based on success and failure of actions in memory for similar environmental conditions being sensed by their "senses". Title: Re: A (Fractal) Theory Of Everything? Post by: kram1032 on January 13, 2010, 04:36:45 PM I wonder, what a nice combination of confidence driven and genetic programming could do...
For a full model, you might need both... :) Btw: I don't see the problem with your statement that the industry wants blades and not amoebiae: After all, you want the 2nd and thus, genetic programming would be valid for that too. :) Though, if I understood your answer correctly, you say "no" to confidence driven confidence ^^ Title: Re: A (Fractal) Theory Of Everything? Post by: Gary Gaulin on January 14, 2010, 02:33:53 AM I wonder, what a nice combination of confidence driven and genetic programming could do... For a full model, you might need both... :) That might be an excellent way to show the similarities and the differences in the techniques. Help answer questions that you have that are hard to explain without such a model. I know I would have to read every word of that one too! In this case need someone into GA's like I'm into the Intelligence Generator loop/circuit, which would abbreviate to IG which for some reason sounds so funny it has me laughing! My first thought was use the GA to provide a variety of critters like the Intelligence Generator and Detector that has the six sided ring of neurons for a sense of awareness of what is around itself: (http://3.bp.blogspot.com/_Gyn0-PcXrJY/SSL0TNsuKaI/AAAAAAAAAGM/CVtoX5lePVY/s400/Screen.jpg) http://intelligencegenerator.blogspot.com/ It is definitely worth trying to hybridize the two. Good number of educators would want to know about that. The next step after that would be to eliminate the GA from the model, by directly modeling the 4500 gene genome of E.coli K-12 along with all known epigenetic controls. Here's more stats: http://www.ecocyc.org/ECOLI/organism-summary?object=ECOLI Here's excellent info on its "cell sensing" circuit: http://regulondb.ccg.unam.mx/CellSensing.jsp Each gene would be a bot. The gene's regulators would be it's environmental inputs. Instead of producing movement of motors/muscles action data would produce RNA that would sent on a given pathway like to the golgi to be used to replicate protein molecule bots. Not have to right away include all cellular processes, can start with just some of the interesting sensing circuits that are known. But without some form of rudimentary intelligence the best I would expect from adding flagella is random motion. Else be stuck in one direction always on or off. http://www.cellsalive.com/animabug.htm There is excellent info at the start of this paper describing their "swarming" behavior on an agar surface. It also goes into detail about how some of it might work. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1797309/ http://www.rowland.harvard.edu/labs/bacteria/movies_swarmecoli.html Can start simple with a single memory circuit that has its confidence level set by what E.coli would do when it reaches the end of the raft and in between, and in response to chemical environment swimming through where here each possible condition like 0=dry, 1=agar/gel, 2=slippery, 3=liquid (two bits) is added to the memory addressing, so it knows what it is in. From there it can be made to replicate. After that we better know what is needed to produce variation, but that's here more of a molecular reaction from the behavior of the gene-bots having changed. Are then ready to look again at what is here a "learning curve" of the combined bots in the cell and cell types that have emerged at that point in time: Cambrian Explosion The computer model displays detailed graphs of electronic intelligence that are available on the screen. Molecular intelligence leaves behind fossils and for survivors of time their genetic memories can be phylogenetically read. It's therefore possible to compare the two datasets, with it here predicted that the near-exponential growth curve of the electronic intelligence also exists in fossil and phylogenetic data. Trilobites (extinct arthropods) are known to have had well developed crystal lens compound eyes of modern insects. This level of complexity was reached approximately half way through the sudden proliferation of multicellular organisms some 530 million years ago. Trilobite morphology alone requires us to account for a genomic mechanism capable of this rate of information increase. We will here graph the information increase of beta class intelligence then look for correlation with fossil and phylogenetic evidence to determine whether this could be the cause. The intelligence will be kept busy going from feeder to feeder while we monitor memory (information) increase with time. With its relatively static yet challenging environment this would represent typical information increase in an unexploted niche. A blue line shows number of highest confidence 3 memories. This is a measure of the most successful responses to environment, a measure of overall "fitness". The middle two lines show number of lesser confidence 1 and 2 memories, a measure of relative uncertainty. The black line shows total memory, a measure of relative genome size. (http://3.bp.blogspot.com/_Gyn0-PcXrJY/SdAGOdSPiFI/AAAAAAAAAdU/dU3szvo9Fds/s400/MemoryGrowth.GIF) There is at first a very rapid information increase. The number of confidence 3 memories in comarison to total can suggest a relatively large genome size in comparison to overall usefulness. We can now predict at least two major events. The first when molecular intelligence emerged which rapidly proliferated cells. Then a second when eukaryote cellular intelligence emerged to rapidly proliferate multicellular organisms. This two event possibility is shown with dotted line in the phylogenetic data shown below. The very left of the graph shows the first event where the simple prokaryotes appeared virtually at the same time. This is the first DNA molecular intelligence. It can exist as a rudimentary cell or in combination as an organelle for cells that fully achieve the next emergent level of cellular intelligence. (http://1.bp.blogspot.com/_Gyn0-PcXrJY/SdoL_eEnUsI/AAAAAAAAAdk/6_7AEmKhPHQ/s400/CellTypes.JPG) http://www.biomedcentral.com/1471-2148/4/2 The middle event would be expected where molecular intelligence has passed the threshold to become cellular intelligence where the cell itself meets all 4 requirements of any intelligence. There is then social/migration behavior of amoeba (also stem cells) and excellent hunting skills of unicellular "single celled animals" the protozoans. Catastrophic environmental events would alter the static environment causing extinctions followed by quick recover that retakes control of lost niches. The fossil evidence therefore shows the relative information increase below. (http://4.bp.blogspot.com/_Gyn0-PcXrJY/SfgmuRxAklI/AAAAAAAAAd0/pLfzagLA728/s400/JoeMeertTimeline2.jpg) http://www.clas.ufl.edu/users/jmeert/timeline2.jpg We can conclude that rapid proliferation of multicellular biodiversity is possible where a genome crosses the threshold to meet all four requirements of the next possible emergent level of intelligence. This is by no means the only explanation for the Cambrian Explosion. There would first have to be an oxygen source for animals to exploit, before their emergence would have been possible. Here we show how information increase typical of at least beta class intelligence reasonably accounts for such a rapid proliferation of biodiversity. This again shows how there is a property of intelligence that makes it the kind of thing where once it gets started "Watch out!" because it relatively quickly takes control of everything it can figure out how to take control of, as here likely also happened in the Cambrian Explosion. Anyone want to try throwing all together over the weekend? Or at least part of it? :D Title: Re: A (Fractal) Theory Of Everything? Post by: kram1032 on January 14, 2010, 08:14:41 PM Too bad I'm not good at coding at all ^^
You know what? In nature, C-type intelligence developed to B-type intelligence which got to the next type of intelligence (A) by evolutionary processes. It would be interesting to write a confidence driven GA type thing which's actual goal is to get more and more intelligent for aribitary systems so it basically progressively becomes A-intelligent from maybe a C-intelligence base, even. Set the starting conditions for C-intelligence but give a tiny bit of information so it goes into the direction of getting more intelligent. Maybe, a generic memory system would be enough for that without any actual confidence system. In the end, the ideal confidence system would start to form on its own.... - it has to solve the following problems: -finding a way to use memories which at first get used at random. Just a fitness function leads to actual first success. - extending the found solution (which still willl continue to be approximated better and better) to allow generic intelligence - developing new kinds of collective intelligence systems (interspecies communication where species could be a single type of cell) as well as specialized split evolution for parts of cell-groups (shaping bigger bodies which are made from single cells of certain kind, like muscle-fibres as oppsed to nerve-cells) - allow for even manipulating the fittness-function in some nice way which means, there must be a basic fittness-function which tries to extend intelligence and the abillity to solve problems and sub-fittness functions which can be shaped at the species' "own will", so to say, so new problems can be found and solved. (This might actually be a requirement for the one before) It might be hard to give the whole system a direction in a world that starts off as nothing. So I suggest that the system has to learn "turing completeness" and some way of during-runtime-new-problem-input which would be comparable with a catastrophy or new ecological niches to be found in order to give the evolution a new thing to solve. That is not supposed to stop the previous quests but rather just to add a new situation so on one hand the algorithm could become even more intelligent (due to the main fittness-curve) and on the other hand, new species could be derived from already existing ones which approximate problems nicely. So a few things you could throw in to start with would be your atomic simulation or, to start simple, "find an algorithm as simple as possible for addition" and as soon as that's solved, go further to "find one for multiplication" and so on. That way you could rediscover basic maths and soon (those basic cases probably will be solved in a satisfying way rather quickly) more complex problems like, say, an algorithm for the gamma function or for rotating a given point in a vectorfield around another point in the same field (or do so with the whole field). Also an interesting application would be to somehow generalize the dimensions the cells could act in. So, at first they only have a single motor and thus can only move back and forth. At the same time, they can "feel" in the 1D-directions and that's all for now. They first have to figure out how to come to dimension 2 by evolutionary adding motors and organs which provide both information and movement along the new directions. Just let it learn about the existence of higher dimensions and maybe solve the problem of the best Close-packing of spheres in 4D and higher looks like more or less on the fly. One important thing would be that already concluded things can be applied to new topics or combined to problems yet to be found. In case of unicellular organisms, that usually works even over the borders of a species which means that two not really related species sometimes share genomes and lead to a new kind of organism which might (or might not) be way "fitter" than any earlier organism. In case of multicellular organisms, that's a bit harder as the single cells are often highly specialized. Usually, additional information in that case would lead to illnesses or, if the corresponding parts are deactivated (by splicing for instance), nothing but immunity agains the corresponding type of illness. Anyways, to actually allow to share information even between two kinds of "results of certain problems" so, under circumstances, to either find the solution of a totally unrelated problem - maybe unknowingly a problem which actually (in reallity) only is discovered in two or so years from now - or to merge two found subproblems to the solution of the actual problem will be important as well. Just got an idea how confidence could fit into that in a different way. The site posted above mentioned that to find a solution for problems with contrary goals, one often has to set priorities to the fittness function. Rather than desciding this beforehand, just do it confidence drivenly. The factors will slowly converge against the best values which lead to the best/fastest intelligence increase. :) Also never totally kill any solutions, just maybe lower their confidence so recursion-priorities can be set. "Better" solutions get more attention in order to allow for faster progress but "not so good ones" might just hold that tiny bit that by the above mentioned exchange of information could be the key to not only a better but possibly even the BEST solution :) (However, in that case, confidence from 1-3 probably isn't enough as the system most likely will try to solve more problems over time...) Puh... I hope, all that isn't too random ^^ Title: Re: A (Fractal) Theory Of Everything? Post by: Gary Gaulin on January 15, 2010, 08:14:48 PM Kram, there is no doubt that you are way beyond average knowledge in robotics and the properties of intelligence! And discovered how applications for what I have been describing are so numerous it's hard to decide where to begin. Where there is just one starting point, the project is easy to get started. Where there are hundreds or more that would all probably work it can become overwhelming.
I proposed a "theory of everything" since that helps narrow down a field of possiblities that gets wider and more complicated to code as complexity (emergent level) of what is modeled increases. Modeling atoms would be starting point 0 where all else above it is theoretically emergent from that alone. Can then save fully trained atom-bot memories, that sum up to molecule-bot memories that only need one per molecule type instead of one for each molecule in the model. Can then sum up each molecule to make organelles. Then sum up each organelle to make cells. Then sum cell to make multicellular organisms. Even where it's still probably impossible to fit a human into a PC it would still be the most awesome computer model ever. Since you do not code your own programs yet it would be possible to add to the particle-bot program a way to experiment with their behavior. Possibly a window that comes on the screen when you push a command button to enter and save the conditional If..Then.. statements the program will use. Can also have it load the last memory you saved so that it does not have to start from scratch each time. Also be nice to have three types of memories/entities so there can be the elementary particles; electrons, neutrons and protons. Once the behavior is similar to real atoms we can see what is emergent from them by adding a way to sum up entity types to share the same memory. Title: Re: A (Fractal) Theory Of Everything? Post by: kram1032 on January 16, 2010, 10:39:19 PM That would be like adding new goals over time.
You pause the program to add new conditions and go on :) That sounds nice :) Title: Re: A (Fractal) Theory Of Everything? Post by: Gary Gaulin on January 17, 2010, 12:55:47 AM I'm not rushing into recoding the program, but after thinking about a way for others to experiment with it I got to thinking about the free MASM32 compiler:
http://www.masm32.com/ Since it directly codes the instructions that the CPU itself runs on, assembler language is the fastest executing code there is. And there are not a large number of commands to learn. Or is it something that becomes obsolete, it's a free download because at the CPU level what worked years ago to code it with still works as well now. Should be able to write in multi-core features, with codes that would be in the CPU documentation for them that Intel should have online. But no rush there, first need to get something running. I downloaded then reinstalled it without too much difficulty except for the MacAfee Anti-Virus having to be turned off for 1/2 hour for the reason they explained in the install program then had to delete the MASM32 folder it made but was not allowed to install files into. Looked like it's using memory like crazy but was just unpacking example programs. Once finished there is information on what to do next to get started that I printed out. Also makes an icon on your desktop for an editor that has an "Introduction To Assembler" in it that is not very long. Briefly explains how the CPU works so you know what the instructions in the example programs are doing. There are a good number of example ".asm" programs (that compile to ".exe" files) that make slider controls, listboxes, zoomable bitmap grid, message boxes, and all else needed to build a better bot or even fast fractal software. Just skip "qexit.exe" unless you want to find out the hard way (like I did) that it turns off your computer. The "tutorial" examples and maybe others seem to be for DOS mode but most worked fine with windows. It does not seem worthwhile to recode any of the programs that I now have, so I made another attempt at using the Schrodinger's Equation that years ago I had drawing orbits but the theory behind each variable was relatively mysterious. I'm now thinking that it might be better to organize waves that are each somehow one step amounting to a complete trip around an orbital. After there is one electron in each subshell they must pair up oppositely and opposite orbital direction to cancel out each others wave (remain stable) or else like forces will repel sending the wave flying out of the atom. And there are photons to account for that are part of Shrodinger's Equation. Details of their trajectory are here: http://physics.bu.edu/~duffy/py106/PhotoelectricEffect.html That brought me back to the physics "Particle In A Box" equation, which is the simplest Schrodinger Equation that there is. Here's a good lecture on it in case you or anyone needs it. http://www.youtube.com/watch?v=f-ibK6TMf7k And here's a real challenging lecture on the Schrodinger's Equation (jumpy image but still good info) that starts off by explaining the wave-like properties of the elementary particles, that are hard to account for in a model: http://www.youtube.com/watch?v=2ejyr-E7q2M&feature=related Also have a short animation showing some of the things atoms do together. http://www.youtube.com/watch?v=DrGRTtdJgoo&feature=related It looks like there is no easy way out of the math of this problem. But it's looking easier all the time so that's a good sign! Title: Re: A (Fractal) Theory Of Everything? Post by: David Makin on January 17, 2010, 03:03:21 PM Hi all, those of you interested in this thread will definitely be interested in the end section of this BBC program:
http://www.fractalforums.com/fractal-related-links/chaosplusfractalsbbc-tv-limited-availability/msg11706/#msg11706 (http://www.fractalforums.com/fractal-related-links/chaosplusfractalsbbc-tv-limited-availability/msg11706/#msg11706) Title: Re: A (Fractal) Theory Of Everything? Post by: Gary Gaulin on January 17, 2010, 03:43:19 PM And just in case the math is still not making sense of the problem here are several relatively short "Common Sense Quantum Mechanics" videos to explain an intuitive way to visualize the concepts for making models and such that uses pins/arrows going through wires to account for polarization of light. Very low budget but he does an excellent job of explaining things:
http://www.youtube.com/watch?v=SYzFC-nxy9I Another on polarization but does not demonstrate polarization: http://www.youtube.com/watch?v=0udtI6FTMa0 Then there is the one for the Schrodinger's Equation explained as spinning arrows. http://www.youtube.com/watch?v=JmEMVJYbTu8 Also Planck's Constant http://www.youtube.com/watch?v=LPQS9cF3C7s Wave travel can be broken down to a number of degrees at a time so that we can see it. But a real particle that is also like a wave would seem to be a bot that is also a self-powered rotating needle that turns 360 degrees per one Planck's Constant energy unit step. With there being no need for motors, all data and addressing of the self-learning memory system would then sense and control direction of travel around what we might visualize as curved space/time around atomic nuclei. Title: Re: A (Fractal) Theory Of Everything? Post by: Gary Gaulin on January 17, 2010, 04:03:53 PM Hi all, those of you interested in this thread will definitely be interested in the end section of this BBC program: Hi David, thanks for the video! I just played it, was a big help seeing how the tree fractal fit into things. Helped show how fractals and the physics are related. And with all the physics video's I posted that was a good break from the heavy math that is in that science. Title: Re: A (Fractal) Theory Of Everything? Post by: kram1032 on January 17, 2010, 08:55:27 PM So, you're going to code a new botparticle-style program but this time with the schrödinger-equation? :)
Title: Re: A (Fractal) Theory Of Everything? Post by: David Makin on January 17, 2010, 09:12:50 PM So, you're going to code a new botparticle-style program but this time with the schrödinger-equation? :) You can't possibly do that - the program will only exist when you're viewing it :D Title: Re: A (Fractal) Theory Of Everything? Post by: kram1032 on January 17, 2010, 10:23:44 PM haha, yeah xD
Well, just let it run for a year, look at it and pure quantum randomness will have done its job xD Instead of particles, you'll see clouds of particle-finding-probabilities :) Thanks for the video-link, david :) The last thing perfectly drew the relations of fractals and intelligence and besides featured a very nice application :) Title: Re: A (Fractal) Theory Of Everything? Post by: Timeroot on January 17, 2010, 10:24:51 PM So, you're going to code a new botparticle-style program but this time with the schrödinger-equation? :) You can't possibly do that - the program will only exist when you're viewing it :D Title: Re: A (Fractal) Theory Of Everything? Post by: Gary Gaulin on January 18, 2010, 04:50:03 PM So, you're going to code a new botparticle-style program but this time with the schrödinger-equation? :) I have to make sure to also work on what might help pay growing bills but that would certainly help achieve realistic particle behavior, in the least amount of time. But thankfully studying how the Schrodinger Equation works established a needed X,Y with rotation coordinate axis system for the CrossCut algorithm. I found out that it has the same set of rotations that are needed for that algorithm even the "phase" angle I was trying to explain that does the exact same thing which here makes the spinning (while cutting a spiral) crosscuts the spinning needles! Not following a tried and proven system that works exactly one that already has names for each thing led to confusion while reinventing that wheel which led to a mess that one way or another had to be straightened out before I wasted even more time renaming variables and guessing which way things should rotate with the rest of the system. Another case of "learning how they do it" ending up saving me time in the long run, and I think I'm almost there as far as truly understanding it all. What really speeded things up is that Arjen (common sense quantum physicist) wrote me back and included this link to something they wrote that's very very good: http://en.wikiversity.org/wiki/Making_sense_of_quantum_mechanics/Principles_of_Quantum_Mechanics I'm now "taking notes" and am putting them into a shell program (where the existing crosscut and particle-bot code later gets put inside) where I can test the new math system to make sure it's all working before recoding other software to work that way. Will give me a chance to try out what Arjen explained without neglecting work-work that I am also committed to finishing in the least amount of time. Problem is though, that work connected to fractals that connected to Schrodinger's equation that connects to the particle bots for displaying that behavior that connects back to the CrossCut program that needs the exact same math or it will remain a programmers nightmare to properly structure. Might even say that in this case the CrossCut algorithm not having things the way they are in math/physics makes it somewhat amateurish. So there is no easy way out, no matter what I do. With it now being much easier to apply the math to the particle-bot problem than it is to the CrossCut algorithm I will first try to figure out how Schrodinger's equation relates to that. From what I can see it is simply a matter of adjusting confidence upward by 1 when a particle moves to where there is a higher probability of finding it, else confidence is reduced by one. Schrodinger's looks well suited for controlling something like that until it is trained, but seems very poorly suited for directly modeling from due to there being so much number crunching required to keep the model moving and probabilities a fuzzy logic that as the name suggests is hard to get a clear picture from. It's too early to say whether the particle-bot idea would work but where it's hooked in correctly it should do something interesting. I can just as easily (or even more easily) test the math by as soon as possible adding Schrodinger's to the existing particle-program where electrons leave a trail behind that can be made as long as we want and other things. Would otherwise need some way to see it from all angles and in action, no sense in adding that to the shell program when what might show them even better and faster is already there. Only drawback is in needing to figure out an easy way to get something that works right away so it does not end up stretching on for days then weeks. One idea I had is to make a small enough 3D world environment where the X,Y,Z of each particle relative to nuclei is simply and easily added to addressing. Now only need a sense of everyhing else The behavior gained from Schrodinger's equation should make the lines (sort of speak) on the gridlike environment fall on Fourier Transform intervals (a curve is already in it) but with a program it's much easier to draw a straight line from point to point. Also, there is a "State Vector" (energy E, position r, momentum p) that seems to belong in addressing. And it will need a way to sense other particles, but will better know what is needed when I finish figuring out all of the new information from Arjen. They sure are making a hard concept easy to figure out, and I sure appreciate the help. So, you're going to code a new botparticle-style program but this time with the schrödinger-equation? :) You can't possibly do that - the program will only exist when you're viewing it :DWith what I had to endure like that (but for real) in other forums I better not even get started on a tangent in that direction! The wrong "i" word in the wrong place really can get one henpecked over nothing for mentioning it. Which is why I'm glad that there are forums like this one where what I say can be understood, or for the most part anyway. :D Title: Re: A (Fractal) Theory Of Everything? Post by: kram1032 on January 18, 2010, 08:03:59 PM If I got you right, you're first trying to do the 2D-Version and then the 3D-one but the 3D-one will be EASIER than the 2D-one? O.o
That would be more than rare xD You said, you need an all-side view... Coulnd't you just add 6 (smaller) static views of the rotating cube and let the 7th be the rotating cube as always? :) Maybe give each cube-face a different colour, so you can easily see on the spinning one, which side of the fixed ones is where :) But I'd highly recommend you to learn OpenGL, get some OpenCL-enabled graphics card (so, any modern nVidia, afaik - no need for it to be too expensive), and learn that too :) As soon as you know those two, you'll surely benefit on both your single programs (for vastly decreased calculation times) and on your job's side (if you're a programmer, at least...) Title: Re: A (Fractal) Theory Of Everything? Post by: Gary Gaulin on January 20, 2010, 06:42:45 AM If I got you right, you're first trying to do the 2D-Version and then the 3D-one but the 3D-one will be EASIER than the 2D-one? O.o Actually there are probably a couple hundred different ways to use Schrodinger's equation, depending on variables that are given and what is to be drawn. Can also be implemented using matrix math. One method first calculates the time-independent equation then uses that result to calculate time-dependant units. Going from 2D to 3D is easy. Getting else just right is the hard part. And in this case the computer model already has a coordinate system like in the equation, just needs the rest of Schrodinger's added to it. You said, you need an all-side view... Coulnd't you just add 6 (smaller) static views of the rotating cube and let the 7th be the rotating cube as always? :) Maybe give each cube-face a different colour, so you can easily see on the spinning one, which side of the fixed ones is where :) Usually when I need a quick view from 3 sides I make 3 pictureboxes then draw (x,y), (x,z), (y,z) but in this case it's already set to go with spherical shells. Only had to include electron x,y,z in addressing, like I mentioned trying out. Worked great. Links below are for a download. (http://2.bp.blogspot.com/_Gyn0-PcXrJY/S1aGpxLOY-I/AAAAAAAAAuw/N17rp8de5JM/s320/Sphere.JPG) http://selflearningbots.blogspot.com/ http://sites.google.com/site/intelligenceprograms/Home/AtomSphere.zip?attredirects=0&d=1 Here is the code. The math that produces the sphere is between the two ".####" lines that mark where it is: Code: '>>>>>>>>>>>>>>>>>>>>>>>> Main Intelligence Generation Loop <<<<<<<<<<<<<<<<<<<<<<< Title: Re: A (Fractal) Theory Of Everything? Post by: kram1032 on January 22, 2010, 09:06:40 PM Whoa, this seems to be a lot faster :D
Very nice :) This looks good and seems to work :) So this is using Schrödinger's Equations? They learn the correct behaviour more or less instantly! So, you should add more rules. A sphere is the easiest and only exactly found solution of the Schrödinger Equations, after all... Even in the biggest world-size with maximum amount of electrons, one gets 64 fps. That's a LOT faster than the same stuff in the previous version :) Things to add: protons and neutrons which need a virtual electron and a proton, quantum numbers (main, secondary, magnetic and spin) so electrons don't just all go into a single spherical orbit... :) Title: Re: A (Fractal) Theory Of Everything? Post by: kram1032 on January 27, 2010, 06:38:59 PM Still alive, Gary?
Title: Re: A (Fractal) Theory Of Everything? Post by: Gary Gaulin on January 28, 2010, 08:43:09 AM Still alive, Gary? Yes, I'm still here! :D I have an (almost ready to go) 3D Lennard Jones to use as the starting point to model the motion dynamics of atom nuclei. And (surprise!) you can select how many protons and neutrons, which should in turn determine how many electrons can be held in its orbitals. Instead of the behavior being trained into memory it memorizes the conformal change in the close-backing geometry (but not yet showing that motion in a second screen). From what I can see vibration alone produces internal orbital motion from one semi-stable geometry to the next, or in other words random vibration is here producing nonrandom motion. Where there are four particles (inert helium) the geometry is too stable, so there is only one geometry. Will need to finish the program to see what happens with other inert element geometries, then see what happens when they are given "spin". This one required some additional study. Am also experimenting with Java. But the popular NetBeans IDE was a real stinker and nothing but problems giving me warnings of a bug in Java slowing it down, and was so slow it was totally useless and had to be uninstalled. It could not even recognize the type of Java program being loading, or find what were supposed to be standard libraries. There is also a simple 3D physics IDE called "Processing" that did work much better but for now I want to find the current Java standard. Let me know where you know of one. I'll upload what I have as soon as I get the code finished up. Or at least get back in a day or so to show how it's going. I might stay with Visual Basic for this project, but an applet would be nice too. Title: Re: A (Fractal) Theory Of Everything? Post by: Gary Gaulin on January 28, 2010, 06:28:19 PM I finally had some success with the "Eclipse IDE for Java EE Developers":
http://www.eclipse.org/downloads/ And found some great starting models to get used to the code, like Hydrogen Atom Orbitals: http://www.falstad.com/qmatom/ And Molecule Orbitals! http://www.falstad.com/qmmo/index.html Unfortunately the 3D Vector Fields that I most wanted to experiment with had unrecognized code like "#ifdef" and "#define" which still remains a mystery. So as usual, the install is still missing something!!! ??? http://www.falstad.com/vector3d/ In case you want to try it. For the http://www.falstad.com/qmatom/ download the Zip archive of this applet. on your hard drive where you can easily find it later. In very upper left of Eclipse click: File > New > Project Double click "Java Project" Enter a name for the project like "HydrogenAtom" then "Finish" since either the default JRE (jre6) that it was set with or J2SE-1.4 used in their tutorial both worked fine. When the treeview icon shows up in the package explorer window in upper left of screen you expand the nodes where the |+| is and the node "src" should be empty so click on that to select. In very upper left now click: File > Import Under "General" select "Archive File" then double click (or click "Next") Now use the "Browse" command to find the zip file that was downloaded, which fills the two boxes below with info on what's in it. Click "Finish" The treeview icon for "src" in the package explorer window should now have "AtomViewer.java" to double click on to bring up it's code. Should then finally be able to "Run" the program. Will be yellow warnings all over it that I next need to figure out, but at least they do not crash the program. Title: Re: A (Fractal) Theory Of Everything? Post by: kram1032 on January 28, 2010, 07:13:24 PM nice stuff :D
processing is actually what I use for my images. :) However, it's suboptimal for that purpose... But with particle-like simulations I've seen a lot of nice stuff done with that :) So it'll most likely work for this too ;) But eclipse is good, too, I guess :) Title: Re: A (Fractal) Theory Of Everything? Post by: Gary Gaulin on January 29, 2010, 12:00:09 PM I can see why you like Processing. It is simple to use and is ready to go with sample applications for working with fractals! Here's the link in case anyone else wants to try it:
http://processing.org/download/index.html I would say it's half-way between Visual Basic and Java. For your application it's the best of both worlds. Do have to admit though that the Visual Basic IDE is faster loading and able to compile and have a program running on the screen almost instantly. For some reason all Java type compilers are slow. The Processing IDE was faster though. There was also a good link in the reference material for Processing to the Sun-Java website that explains the very basics of the language, needed to get started. I read this one carefully: http://java.sun.com/docs/books/tutorial/java/concepts/index.html It's hard to say goodbye to a standard that your proficient at. But with it looking like VB.Net will never catch on it looks like I have no choice but to stay current. And I'm not getting out of this one with Processing alone, especially with having to study the existing physics applets. Thankfully the more I use it the more I'm liking the Eclipse IDE even though it's slow. A faster computer would help that, so I guess I can dream on about one day affording it. Which reminds me, I have a 20x40 foot area loaded with dinosaur footprints in brownstone if you need some! The area should be in a museum but they can't afford this one in part because of the size of display and no matter how green it is this kind of science is not valued/supported anymore. The tracksite horizon would last a long time as a patio or something. But getting back on topic! I did find another interesting applet that would compile with just minor warnings, most I already fixed: http://www2.biglobe.ne.jp/~norimari/science/JavaApp/e-SolarSys.html I like that one because it's very simple. Also seems to have a good rotational system to apply to the CrossCut Fractal. But I'll need to study it some more to know whether it would work or not, which requires getting more used to Java. Eclipse does make that easy by having information on each command automatically come on screen, like Visual Basic had to some degree but was not as informative. The hard part was getting started, especially with most of the source code for good applets not being compatible with the newest IDE's. Title: Re: A (Fractal) Theory Of Everything? Post by: Gary Gaulin on January 30, 2010, 09:23:01 AM Sorry for rushing through half a thought. Being from Massachusetts I have to love seeing things like Processing come from MIT students/professors.
Like all sample programs that come with a new IDE they are simple little programs to help get us started and can be added to, so we don't mind the limitations. If we need more (and are willing to program ourselves) then at some point have to get beyond the limits of a language made to make things easy like Processing and either add to that software package or learn a language like Java. But at high resolution I would expect Java to be speed limited which one either has to live with or learn a language like Assembler (MASM32 or MASM64) that can go full speed straight to the CPU. Or where it will work (like in molecular dynamics) get a parallel-video card and code in what they recommend, one I looked at uses CUDA. All languages are suboptimal in one thing or another. CUDA only works with the right hardware installed. Assembler does not compile applets. And the internet is littered with Java source code that does not run. I just tried to find a fractal applet and the first one I downloaded had so many errors it could not even begin to run it. Another did not have a needed library in the program or anywhere I could find on the webpage. There might be a good one that will work with a 2010 compiler, somewhere. But I gave up looking for one! With all said: These days we have to be happy to get anything running at all. Title: Re: A (Fractal) Theory Of Everything? Post by: kram1032 on January 30, 2010, 11:17:03 AM That app is really nice too :D
I wonder how it would look like with all the planets and their moons in correct scale (with fake-optical radius, so they're actually still visible at that a small screen^^) Title: Re: A (Fractal) Theory Of Everything? Post by: Gary Gaulin on January 30, 2010, 02:24:12 PM That app is really nice too :D I wonder how it would look like with all the planets and their moons in correct scale (with fake-optical radius, so they're actually still visible at that a small screen^^) I think it would look great where there was a button for each planet that will fly you there. I saw a listing for one that at least had life size ones but tried again but none there. I did find one where the comets can be taken out of the picture and the planets look like our inner solar system. http://www.arachnoid.com/dark_energy/space_applet.html Also found its source code and was able to get it error-free, ran great. Only had to copy/paste the code they have on another page into the "src" folder icon that appears after creating a new program in Eclipse, that's like an annoying nanny that has to have everything just perfect. But I did shut off its spell checker that was finding errors in the comments even. Gee, I just got going in Java and am already debugging all the code on the internet! :D Title: Re: A (Fractal) Theory Of Everything? Post by: kram1032 on January 30, 2010, 06:24:07 PM xD
Nice :) Now, go, learn form that and do it better xD j/k Title: Re: A (Fractal) Theory Of Everything? Post by: Nahee_Enterprises on January 30, 2010, 06:36:27 PM Like all sample programs that come with a new IDE they are simple little programs to help get us started and can be added to, so we don't mind the limitations. ....or learn a language like Java. .....or learn a language like Assembler..... All languages are suboptimal in one thing or another. ..... Another did not have a needed library in the program or anywhere.... Such discussions always take me back to the old "How To Shoot Yourself In The Foot In Any Programming Language" jokes that may be found all over the Internet. Here is but one of those: http://www.toodarkpark.net/computers/humor/shoot-self-in-foot.html (http://www.toodarkpark.net/computers/humor/shoot-self-in-foot.html) Title: Re: A (Fractal) Theory Of Everything? Post by: Gary Gaulin on January 31, 2010, 01:10:13 AM You're right! I could have even just posted that instead. And can see I forgot to mention that going full speed straight to the CPU with the Assembly language assembler is like saving time on a long trip by putting the gas pedal to the floor and not letting it up until you get there. The slightest error in aiming causes an immediate and sometimes spectacular crashing of the entire system. But there is no faster way to get there, so sometimes we just have to strap ourselves in real tight then hope for the best. Quote Assembly You try to shoot yourself in the foot only to discover that you must first invent the gun, the bullet, the trigger, and your foot. You crash the OS and overwrite the root disk. The system administrator arrives and shoots you in the foot. After a moment of contemplation, the system administrator shoots himself in the foot and then hops around the room rapidly shooting at everyone in sight. By the time you've written the gun, you are dead, and don't have to worry about shooting your feet. Alternatively, you shoot and miss, but don't notice. Using only 7 bytes of code, you blow off your entire leg in only 2 CPU clock ticks. MS-DOS You finally find the gun, but you can't find the file with the bullets for the life of you. You shoot yourself in the foot, but you can unshoot yourself with add-on software. Lisp You shoot yourself in the appendage which holds the gun with which you shoot yourself in the appendage which holds the gun with which you shoot yourself in the appendage which holds the gun with which you shoot... You attempt to shoot yourself in the foot, but the gun jams on a stray parenthesis. PostScript foot bullets 6 locate loadgun aimgun shoot showpage Visual Basic You'll only appear to have shot yourself in the foot, but you'll have so much fun doing it you won't care. You do a Google search on how to shoot yourself in the foot using Visual Basic. You find seventeen completely different ways to do it, none of which are properly structured. You paste the first example into the IDE and compile. It brushes your teeth. .NET You can now shoot yourself in the foot with any of fourteen weapons, ranging from an antique medieval crossbow to a laser-guided Destructo-Beam. However, all these weapons must be manufactured by Microsoft and you must pay Microsoft royalties every time you shoot yourself in the foot. Java You write a program to shoot yourself in the foot and put it on the Internet. People all over the world shoot themselves in the foot, and everyone leaves your website hobbling and cursing. You amputate your foot at the ankle with a fourteen-pound hacksaw, but you can do it on any platform. JavaScript You find that Microsoft and Sun have released incompatible class libraries both implementing Gun objects. You then find that although there are plenty of Foot objects implemented in the past in many other languages, you cannot get access to one. But, seeing as JavaScript is so cool, you don't care and go around shooting anything else you can find. Although they missed Turbo Basic, Power Basic and a few others I have experience with that actually does sum it all up very well. xD Nice :) Now, go, learn form that and do it better xD j/k Yes, I have done enough hobbling and cursing from websites and finally have something to shoot myself in the foot with using Eclipse which is like a scatter-gun that thus far made little yellow and red holes in all programs it has so far successfully loaded. But it comes with a little nanny-nurse to help patch some of them up, which is better than not having one so it's at least a little less painful that way. |