I suspect that the subpixel algorithm assumes that an LCD has stripes in the order RGB... and, IIRC, the bronze G3 has GRB stripes, meaning that it's setting the wrong subpixels. What they really need is an algorithm that can adapt to this situation... but apparently it's a small enough population of their market that it's not worth the effort, and (AFAIK)/every/ color LCD is RGB order right now.
How the hell did that get a +5 insightful? It's not in the least bit insightful... or rather, if it is, it's heavily nitpicking and silliness. Hmph. +3 would have been better. Mod me back down overrated, please.
Well, in terms of keeping moore's law, clearly the evolutionary technology is preferred. One of the most obvious corollary of moore's law is that progress is continuous; revolutionary directly implies non-continuous, and it seems unlikely that the development of feasible quantum computers would lead to a keeping of moore's law, rather than a breaking of it in one direction or the other.
The thing is, the back of the envelope calculations show that this whole thing is silly, at least in the context of particle physics. (I think it's even more silly in the case of nanotech and biotech, but that's a different issue.) We don't need to look at black holes, or even the sun, to see the kind of reactions we're playing with 'in nature.' In our own atmosphere there are collisions thousands of times more powerful than we can even imagine recreating... the point of particle accelerators isn't to do stuff that doesn't happen in the real world, but rather to reproduce common occurances in a setting where they can be studied. If any of these experiments were going to create a black hole that ate the whole world... the black holes would literally be falling from the sky all around us.
Um, the size of our galaxy is measured in whatever units you'd like, so why not hundreds of light years? However, if that's our unit u, the diameter of the galaxy is on the order of 1000 u, not 1 u.
Also, with a 17" apple display and a Viewsonic VX800 on my desk, I'm seriously considering getting rid of the viewsonic and picking up another apple... they're just much nicer on the eyes.
Do people really wait 30 seconds for mozilla to load, or even 10? I'm on a 600MHz machine here with 128MB of RAM, and Safari loads in about three seconds (two bounces, for mac folk), and even so I tend to leave it open rather than waiting for it to start up. The idea of waiting that long to get to something as basic as a web browser seems as bizarre as waiting for a bash prompt.
Why do you care that your 3GHz processor is designed for, say, 2.4GHz? Answer: You don't. Now, you may care that the stability is lower... but we've already established in our legal system that stability isn't required of anything except cars (lemon laws), and the judge is unfortunately right to say (a) your initial complaint is invalid and (b) the real reason behind your initial complaint is your own problem, should have bought the repair plan.
Honestly, I've never understood PVC potato guns. I always preferred 2" ID stainless steel piping, buffed inside until it shines, and 2" ball bearings. Maybe that's just me.
"How many innocent people would die during the next 10, 20 or more years of Saddam's reign?"
I honestly don't know, but I readily confess that it may be higher than the number of civilians we killed. It also may be lower; I'm really not sure. But (a) our government is in the business of protecting it's citizens. It is not a world government. I won't argue that this means we shouldn't get involved in other issues; I'll simply say that our chosen route was guaranteed to kill Americans, the very people the government is supposed to protect; the alternate route showed no particularly high probability of resulting in American deaths. And (b) guaranteed deaths by an outsider are bad. Possible deaths by a local ruler are bad too, but I honestly believe that we will end up killing a large number of people, both in the fighting and in the reconstruction afterwards, and this is a greater danger to Americans (through increase in world anger) that than the alternative, just as it is a greater danger to Iraqis.
I've done my best to not dodge your question, but the truth is I just don't know, and I'm not sure it matters. Is 1100 deaths ((admittedly biased ref) higher than we might have seen with Saddam? It depends. Saddam was not immortal, and was not actively causing death. You may be right that we saved lives, but we really never can know. What we did do is kill Americans and destabilize one of the most politically stable regions of the world.
I'm amused too, so don't think I'm offended or anything. But if you're really concerned that that's the situation, please read my other postings on this story. As I said... 2 years ago things were different. And no, I'm not working hard enough; I am a student.
(a) Wasn't talking about sparse matrices, was talking about small stuff. I'm doing 3D manipulation, so I'm dealing mostly with 4x4 matrices and 4-vectors.
(b) As far as I know, Mathematica doesn't do sparse lists by default... so your 100x100 array will be 100 100-long lists in mathematica, and either an array or a sparse matrix in Matlab. Matlab is surprisingly well optimized for an interpreted language.
Mathematica is slow, inefficient, and a total joy to work with. Matlab is about 10x faster, and at least 100x faster for matrix stuff (mathematica does matrices as linked lists of linked lists!)... but if I want speed, I'll use Fortran 90, really. Mathematica is just lisp + pretty printing... but it undeniably works, and I've never used a more productive tool in my life.
My basic method is to take a 3D solid (that is, a polygonal 3D model that encloses a volume). I apply a few standard smoothing operations to this model to get a very, very rough shape (in particular, I smooth it until it is fully convex, the first time). I then do simulated annealing based on a handful of hardcoded starting conditions to find a good approximation of that shape -- metropolis might work better for precision, but I decidedly don't want precision now. I then take the model and resmooth it, but one step less. I use the previous foldset as a starting point, and anneal from there. I repeat this for each level of smoothing that was originally needed.
Normally it takes around a million attempts to approximate each smoothing level, although this varies by a factor of at least one hundred, where the swan, for instance, takes about fifty levels of smoothing.
Make sense? Not saying it works wonderfully, but I think it's the correct approach and just needs tuning. Amount of work is, to a first approximation, linear with the geometric complexity of the model, and more or less independent of the number of folds... certainly not exponential in the number of folds!
The problem is scale. A nice origami model is easily 100+ folds. Let's say I ask for a hemisphere. By your approach, we'd generate a maximum number of folds (say, 40), and choose the closest. But that includes, of course, two trillion (2^40 + 2^39 +...) possible models... and Mathematica, as a lisp implementation, isn't memory efficient, so we're talking tens of terabytes. With my model, I'm guaranteed something that looks reasonably good with quite low memory usage (hundreds of megabytes, usually)... it's just not always the best looking. Brute force is almost always an unreasonable approach, and for this problem certainly.
Hmm. We'll have to see. But again... I was doing different things two years ago. I'm an undergrad student, I couldn't exactly justify laying down $2k on a box just as an excuse to spend $500 on mathematica licenses. Again, I'll give BSD a try. If I like it, I'll stick with it. If I don't, and I need to reformat anyway, I'll stick Linux on and see if I like that. I've now been given incentive to change, only time can tell what I'll change to.
Slashdot Mirror
Why? Why should the end user have to agree to anything? Let them use the damn software, don't take up their time translating legal language.
I suspect that the subpixel algorithm assumes that an LCD has stripes in the order RGB... and, IIRC, the bronze G3 has GRB stripes, meaning that it's setting the wrong subpixels. What they really need is an algorithm that can adapt to this situation... but apparently it's a small enough population of their market that it's not worth the effort, and (AFAIK) /every/ color LCD is RGB order right now.
I can just see it. Corporation 1: The new Processor 4e7+. Corporation 2: The new Processor 4f7+! It must be better!
How the hell did that get a +5 insightful? It's not in the least bit insightful... or rather, if it is, it's heavily nitpicking and silliness. Hmph. +3 would have been better. Mod me back down overrated, please.
Well, in terms of keeping moore's law, clearly the evolutionary technology is preferred. One of the most obvious corollary of moore's law is that progress is continuous; revolutionary directly implies non-continuous, and it seems unlikely that the development of feasible quantum computers would lead to a keeping of moore's law, rather than a breaking of it in one direction or the other.
But after a few weeks, you do start to wonder who the really tall guy is.
Heh. Nothing quite like hearing your concealed packet sniffer spin up, I imagine.
"No decision to go ahead with an experiment... should be made unless the general public is satisfied..."
When was the last time the general public was satisfied!?
The thing is, the back of the envelope calculations show that this whole thing is silly, at least in the context of particle physics. (I think it's even more silly in the case of nanotech and biotech, but that's a different issue.) We don't need to look at black holes, or even the sun, to see the kind of reactions we're playing with 'in nature.' In our own atmosphere there are collisions thousands of times more powerful than we can even imagine recreating... the point of particle accelerators isn't to do stuff that doesn't happen in the real world, but rather to reproduce common occurances in a setting where they can be studied. If any of these experiments were going to create a black hole that ate the whole world... the black holes would literally be falling from the sky all around us.
If nothing else, they use a million times more pixels. Apple //c vs. Apple 23" Cinema. My, how we've grown.
Um, the size of our galaxy is measured in whatever units you'd like, so why not hundreds of light years? However, if that's our unit u, the diameter of the galaxy is on the order of 1000 u, not 1 u.
Also, with a 17" apple display and a Viewsonic VX800 on my desk, I'm seriously considering getting rid of the viewsonic and picking up another apple... they're just much nicer on the eyes.
Do people really wait 30 seconds for mozilla to load, or even 10? I'm on a 600MHz machine here with 128MB of RAM, and Safari loads in about three seconds (two bounces, for mac folk), and even so I tend to leave it open rather than waiting for it to start up. The idea of waiting that long to get to something as basic as a web browser seems as bizarre as waiting for a bash prompt.
Why do you care that your 3GHz processor is designed for, say, 2.4GHz? Answer: You don't. Now, you may care that the stability is lower... but we've already established in our legal system that stability isn't required of anything except cars (lemon laws), and the judge is unfortunately right to say (a) your initial complaint is invalid and (b) the real reason behind your initial complaint is your own problem, should have bought the repair plan.
Actually, I meant that as the biggest complement possible to both Mathematica and Lisp.
Honestly, I've never understood PVC potato guns. I always preferred 2" ID stainless steel piping, buffed inside until it shines, and 2" ball bearings. Maybe that's just me.
"How many innocent people would die during the next 10, 20 or more years of Saddam's reign?"
I honestly don't know, but I readily confess that it may be higher than the number of civilians we killed. It also may be lower; I'm really not sure. But (a) our government is in the business of protecting it's citizens. It is not a world government. I won't argue that this means we shouldn't get involved in other issues; I'll simply say that our chosen route was guaranteed to kill Americans, the very people the government is supposed to protect; the alternate route showed no particularly high probability of resulting in American deaths. And (b) guaranteed deaths by an outsider are bad. Possible deaths by a local ruler are bad too, but I honestly believe that we will end up killing a large number of people, both in the fighting and in the reconstruction afterwards, and this is a greater danger to Americans (through increase in world anger) that than the alternative, just as it is a greater danger to Iraqis.
I've done my best to not dodge your question, but the truth is I just don't know, and I'm not sure it matters. Is 1100 deaths ((admittedly biased ref) higher than we might have seen with Saddam? It depends. Saddam was not immortal, and was not actively causing death. You may be right that we saved lives, but we really never can know. What we did do is kill Americans and destabilize one of the most politically stable regions of the world.
I'll take a shot. Where's the simple, direct question?
What's a Q-byte? Smaller letters.
You're not a programmer, are you?
I'm amused too, so don't think I'm offended or anything. But if you're really concerned that that's the situation, please read my other postings on this story. As I said... 2 years ago things were different. And no, I'm not working hard enough; I am a student.
(a) Wasn't talking about sparse matrices, was talking about small stuff. I'm doing 3D manipulation, so I'm dealing mostly with 4x4 matrices and 4-vectors.
(b) As far as I know, Mathematica doesn't do sparse lists by default... so your 100x100 array will be 100 100-long lists in mathematica, and either an array or a sparse matrix in Matlab. Matlab is surprisingly well optimized for an interpreted language.
Mathematica is slow, inefficient, and a total joy to work with. Matlab is about 10x faster, and at least 100x faster for matrix stuff (mathematica does matrices as linked lists of linked lists!)... but if I want speed, I'll use Fortran 90, really. Mathematica is just lisp + pretty printing... but it undeniably works, and I've never used a more productive tool in my life.
My basic method is to take a 3D solid (that is, a polygonal 3D model that encloses a volume). I apply a few standard smoothing operations to this model to get a very, very rough shape (in particular, I smooth it until it is fully convex, the first time). I then do simulated annealing based on a handful of hardcoded starting conditions to find a good approximation of that shape -- metropolis might work better for precision, but I decidedly don't want precision now. I then take the model and resmooth it, but one step less. I use the previous foldset as a starting point, and anneal from there. I repeat this for each level of smoothing that was originally needed.
Normally it takes around a million attempts to approximate each smoothing level, although this varies by a factor of at least one hundred, where the swan, for instance, takes about fifty levels of smoothing.
Make sense? Not saying it works wonderfully, but I think it's the correct approach and just needs tuning. Amount of work is, to a first approximation, linear with the geometric complexity of the model, and more or less independent of the number of folds... certainly not exponential in the number of folds!
The problem is scale. A nice origami model is easily 100+ folds. Let's say I ask for a hemisphere. By your approach, we'd generate a maximum number of folds (say, 40), and choose the closest. But that includes, of course, two trillion (2^40 + 2^39 + ...) possible models... and Mathematica, as a lisp implementation, isn't memory efficient, so we're talking tens of terabytes. With my model, I'm guaranteed something that looks reasonably good with quite low memory usage (hundreds of megabytes, usually)... it's just not always the best looking. Brute force is almost always an unreasonable approach, and for this problem certainly.
Hmm. We'll have to see. But again... I was doing different things two years ago. I'm an undergrad student, I couldn't exactly justify laying down $2k on a box just as an excuse to spend $500 on mathematica licenses. Again, I'll give BSD a try. If I like it, I'll stick with it. If I don't, and I need to reformat anyway, I'll stick Linux on and see if I like that. I've now been given incentive to change, only time can tell what I'll change to.