Sure you can derive 'conservation of energy' like most conservation laws from the appropriate symmetry of the dynamical laws.
And yes the mathematics is right.
But still, that doesn't meant that the expression of "conservation of energy" that we intuitively know and use for all conventional situations necessarily applies. (conservation of energy in GR is not so obvious either)
If you have different physics you have different conservation laws. Bashing new physics because of derivations of old physics isn't fair.
After all, for a 19th century scientist, superconductivity itself would appear to be a nearly impossible and preposterous. But nowadays, we can just say it's a macroscopic quantum state which is the lowest state so it can stay there indefinitely. How would you explain to somebody why there is no resistance, or even assuming zero resistance, why, knowing there are particle charge carriers, the energy doesn't radiate away and drain a superconducting coil? It is truly baffling.
This is way out of my field (i'm a nonlinear dynamicist) but the fact that superconductors are basically the only truly physically accessable and remarkably MACROSCOPIC quantum system, and the fact that we don't know squat about quantum gravitation says to me that this line of research is not _a priori_ bullshit, because we may be looking at the first experimentally accessible coupling of quantum gravity.
This is not like cold fusion. In cold fusion I was willing to forget about normal requirements to get fusion, but there were MASSIVE problems that could simply not be overcome, like ANY high-energy nuclear reaction has to make ionizing radiation in rough numbers proportional to the reaction rate. No matter what form of radiation you assume there would have been activation of the metallic substrate in the CF apparatus and resultant soft-x rays. People with state-of-the-art x-ray detectors (including an expert in legitimate muon-catalyzed cold fusion) saw NOTHING beyond background when the expected rate was at least 8 orders of magnitude larger.
Indeed the relation of the solar cycle to climate is very important, but the most recent analyses of data show t at solar forcing cannot account for the dominant fraction of the warming trend.
It's still mostly our fault.
But in any case it doesn't really matter: if changes in the Sun's dynamics are making the Earth hotter, it will suck just that much more when we add insult to injury.
Why is C++ multiple inheritance bad? Among many other reasons, because C++ has NO DISAMBIGUATION UPON INHERITANCE.
Eiffel does.
That makes an enormous difference in making multiple inheritance clean and painless. Why is it that so many people say "use delegation, not MI"??? Because delegation is a painful and manual way of, in effect, specifying how to disambiguate potentially conflicting features.
With GOOD multiple inheritance, you don't need to do that. You just use delegation when delegation really is the best solution.
Notice that Java still lacks dismbiguation, and it IS still a problem, but as Java only has MI of interfaces, the problem just happens to occur less often.
I disagree with the notion that the Java language designers were that brilliant. Compared to any other new OO language designed around the same time, Java is pretty damn poor. It would get maybe a C+:) in a programming language class, not a B or A.
Only if you are used to poking yourself in the eye with pointed sticks (C++) or poisoned pointed sticks (MFC under C++) would you think that Java is all that great.
There are other boneheaded decision in the Java language design as well.
I always thought "well the language is lame, but at least we have all these great API's". I personally never used them. Now people who do are saying that the APIs suck but the langauge is OK.
Oh boy. Is there anything actually good about Java???
I guess the bytecode verifier and security model is fairly decently "high tech" and well designed. That's about it, as far as I see.
The discussion of the "symmetry problem"---i.e. how to guarantee uniform compression of the fusion fuel---is a common meme in popular discussion of fusion.
What's interesting is that much of this confusion has to be intentional misinformation.
Why? Because it was the solution to the symmetry problem which was the key to the H-bomb. The complete discussion of the concept remained officially classified for decades after the H-bomb was invented. In contrast, the basic physics and outline of the engineering behind the A-bomb was made public in the 1940's already.
Some details had leaked out in one way or another by various 'exposes' but nothing really got it right. I believe it was finally declassified in about 1994 or so---I remember the article in the New York Times. By that point, intertial confinement fusion civilian researchers had already figured it out.
The trick is 'radiation driven implosion'.
The soft x-rays from a fission explosion are diverted down to a hollow tube (hohlraum {hollow room} in the German used by the early researchers). There is in fact a barrier, usually of tungsten or some other heavy material directly in between the fission primary and the fusion secondary. This is to prevent the explosion of the fission material itself (the stuff) from getting to the fusion part and ruining it. You just want the photons, and if you block the shrapnel, the photons get there first.
The primary is basically many spherical layers---the secondary is a long cylinder with a few coincentric layers, with a gap near the outermost layer for the photons to flow into.
This form a 'photon gas'---but this fluid equilibrates at the speed of light and not the speed of sound, like normal fluids. Which means that the density and energy of the photons gets very uniform quickly. Although you don't normally feel it in every day life, photons carry momentum and can exert a pressure.
The A-bomb uses chemical explosives to precisely compress the fission fuel, but that isn't powerful and even and fast enough to work with fusion. The description in the article, putting the fusion fuel in the center of the bomb, is mostly incorrect. There is a little bit of fusion fuel in the middle of the a-bomb section but it does not make the really big BANG of the true H-bomb, it is just a booster to make more neutrons to make more fission from a given amount of plutonium or uranium.
Back to the photon fluid. This photon gas has enough energy to very very evenly and violently compress the fusion fuel. This is basically a 3-layered hot dog. The outermost layer is a heavy 'tamper' of lead or U-238 or something else. The middle layer is the fusion fuel. Then, there is a rod in the center with fission fuel like U-235 or Pu-239. The photon gas ablates the material on the tamper---the pressure from the photons and momentum from the ablation compress the fusion tube very severely. At the same time, it is magically arranged for just enough neutrons from the primary fission explosion to start a chain reaction in the fission stuff inside the fusion fuel, known as the 'spark plug'. Thus the fusion fuel is compressed from the outside by the photon gas and ablation and squeezed from the inside from the second fission explosion. This is how much work you need to do to get fusion. The fission material, normally very light, is compressed to a density near that of a white-dwarf star, where it reaches quantum mechanical Fermi degeneracy. That is really really really compressed. Not quite a neutron star (that is even more dense, being a giant atomic nucleus) but still thousands of times more dense than any conventional material.
Then, Boom.
The symmetrical lasers in the Livermore fusion experiments do NOT directly shine on the fusion pellet. That would not make sufficiently even compression. They shine on an outer metallic sphere which ionizes and releases X-rays, and these X-rays equilibrate inside the metallic shell (even though it's now totally vaporized it's still heavy compared to photons) and this compresses the fusion fuel.
This is why the intertially confined fusion program is always more classified than the magnetic fusion, because it is in large measure H-bomb technology. The "national ignition facility" being built for the DOE in the United States is yet another really big ICF project. Unfortuantely for political reasons it is being designed entirely from the point of view of military bomb research and not for any energy research. I think that's a shocking waste but there's a gazillion dollars to test the most obscure things about nukes when we already have thousands of bombs that work way way too well for our planet's and species' good.
The lasers are very power inefficient, but better for getting clean data for bomb research. Ion beams, and probably this Sandia machine are much more power efficient and the likely technology for a power plant, but less is known about them because it doesn't fit in the military application as well.
In any case, it is almost certain that this Sandia machine described in the article uses "indirect drive" {as the x-ray compression method is known} instead of "direct drive" and that is how they plan to work on the symmetry problem. So they probably do have an idea what to do about it, contrary to what the article says. It is still a very challenging problem, but not hope is lost.
The best information on the web about Big Bombs is Carey Sublette's archive. Just about everybody who Really Knows what's going on {and I'm not one of them} has said that the technical details are accurate. In fact, some of the information about the W88 warhead that the Chinese supposedly stole can be found on this website, and the Chinese showed this in their defense.
High energy weapons archive
I am a physicist, but not a nuclear physicist or involved in fusion research in any way whatsoever.
Background: I am a physicist who works in chaotic time series analysis. There are some colleagues in my institute who work on various theoretical aspects of synchronization and information processing with ''realistic'' neurons, i.e. ones which employ the right kinds of time-dynamics. ---------------------------------------------- Firstly, as has been pointed out, the apparent small size of the 'training set' makes the recognition task easier and the apparent results seem better than they are. But...that does not erase the actual accomplishment however. The tasks: It comes down to different statistical concepts.
If you have two hypotheses e.g. A and B, corresponding to 'two words' which were said, then it is easy to build systems which can recognize signals corresponding to A and those corresponding to B embedded in lots of noise. Basically you measure the likelihood ratio p(B)/p(A) using some sort of estimators that you've trained to light up with either A or B. If you gave me the data, I could do this with a number of different semi-conventional numerical techniques on a digital computer. I've seen similar things presented at conferences a few years ago---recognition of specific chaotic waveforms (specifically dolphin and whale song) embedded in lots of noise.
This is known as a "simple hypothesis test".
The more general circumstance, however is that the alternative is not A vs B, but A vs a huge multitude of other possibilities. This task is much more difficult, and correponds to the actual large-vocabulary speech recognition task. Now it becomes much more difficult to set a reliable threshold which will come on only when A is actually present, and not when A is absent. There is a tradeoff of false negative and false positive errors depending on your choice of threshold.
There is no possible way that this thing can recognize 50,000 words. There are only 30 connections, there is fundamentally not enough information processing power intrinsically in there.
What you would do is to have all sorts of these subunits lighting up their own 'word finder lights', and the result of *those* (i.e. the p(A) detectors) would then be inputs into higher level semantic networks of perhaps a similar type. These networks or hidden markov models or whatever are the ones that know which sorts of words follow other sorts of words, and thus let you get better recognition than the individual word finders themselves.
So, what is the accomplishement of this paper??
That they've apparently found an extremely efficient and well-performing low-level subunit using this time-domain information. From our own experimental observations (not on speech but on real live neurons from recently-living animals) this is very important. The fact that it is only 30 connections might mean that it is quite feasible to put 10 or 20 thousands of these subunits on a single chip, running in hardware. Given the factor of a thousand speed increase of electronics over neurons if you could time-division multi-plex different recognizers (blue sky dreaming here!) you could have that much many more of them during the milliseconds to seconds of audio-frequency processing time that we speak at.
If you notice, Professor Berger said that no other speaker-independent system outperformed humans, even in small test bases. Presumably that means in the small Bayesian post-hoc sorts of likelihood test regimes taht I described before. And in addition, it appears that this is not a simulation but that they built it on an actual physical computer chip, another very substantial advance.
My colleagues are going to ask the authors for the actual paper. The title and press release may be overblown, but this smells like real science and a significant advance here to me.
Take home message: even small groups of good neurons can do interesting and useful things. With the right architecture, a small group of neurons can outperform conventional "neuroid networks" of hundreds or thousands of nodes linked by linear transformations of sigmoidal basis functions. We may just be beginning to crack real-AI.
We see major body functions of lower animals being regulated by say ten neurons. Real neurons are much smarter than you think.:)
If small groups of neurons can do this, it makes you appreciate what a hundred billion might be able to do.
Slashdot Mirror
whoomp! there it is
Sure you can derive 'conservation of energy' like most conservation laws from the appropriate symmetry of the dynamical laws.
And yes the mathematics is right.
But still, that doesn't meant that the expression of "conservation of energy" that we intuitively know and use for all conventional situations necessarily applies. (conservation of energy in GR is not so obvious either)
If you have different physics you have different conservation laws. Bashing new physics because of derivations of old physics isn't fair.
After all, for a 19th century scientist, superconductivity itself would appear to be a nearly impossible and preposterous. But nowadays, we can just say it's a macroscopic quantum state which is the lowest state so it can stay there indefinitely. How would you explain to somebody why there is no resistance, or even assuming zero resistance, why, knowing there are particle charge carriers, the energy doesn't radiate away and drain a superconducting coil? It is truly baffling.
This is way out of my field (i'm a nonlinear dynamicist) but the fact that superconductors are basically the only truly physically accessable and remarkably MACROSCOPIC quantum system, and the fact that we don't know squat about quantum gravitation says to me that this line of research is not _a priori_ bullshit, because we may be looking at the first experimentally accessible coupling of quantum gravity.
This is not like cold fusion. In cold fusion I was willing to forget about normal requirements to get fusion, but there were MASSIVE problems that could simply not be overcome, like ANY high-energy nuclear reaction has to make ionizing radiation in rough numbers proportional to the reaction rate. No matter what form of radiation you assume there would have been activation of the metallic substrate in the CF apparatus and resultant soft-x rays. People with state-of-the-art x-ray detectors (including an expert in legitimate muon-catalyzed cold fusion) saw NOTHING beyond background when the expected rate was at least 8 orders of magnitude larger.
Indeed the relation of the solar cycle to climate is very important, but the most recent analyses of data show t at solar forcing cannot account for the dominant fraction of the warming trend.
It's still mostly our fault.
But in any case it doesn't really matter: if changes in the Sun's dynamics are making the Earth hotter, it will suck just that much more when we add insult to injury.
Yes, another Eiffel fan here. (Sather too).
:) in a programming language class, not a B or A.
Why is C++ multiple inheritance bad? Among many other reasons, because C++ has NO DISAMBIGUATION UPON INHERITANCE.
Eiffel does.
That makes an enormous difference in making multiple inheritance clean and painless. Why is it that so many people say "use delegation, not MI"??? Because delegation is a painful and manual way of, in effect, specifying how to disambiguate potentially conflicting features.
With GOOD multiple inheritance, you don't need to do that. You just use delegation when delegation really is the best solution.
Notice that Java still lacks dismbiguation, and it IS still a problem, but as Java only has MI of interfaces, the problem just happens to occur less often.
I disagree with the notion that the Java language designers were that brilliant. Compared to any other new OO language designed around the same time, Java is pretty damn poor. It would get maybe a C+
Only if you are used to poking yourself in the eye with pointed sticks (C++) or poisoned pointed sticks (MFC under C++) would you think that Java is all that great.
There are other boneheaded decision in the Java language design as well.
I always thought "well the language is lame, but at least we have all these great API's". I personally never used them. Now people who do are saying that the APIs suck but the langauge is OK.
Oh boy. Is there anything actually good about Java???
I guess the bytecode verifier and security model is fairly decently "high tech" and well designed. That's about it, as far as I see.
The discussion of the "symmetry problem"---i.e. how to guarantee uniform compression of the fusion fuel---is a common meme in popular discussion of fusion.
What's interesting is that much of this confusion has to be intentional misinformation.
Why? Because it was the solution to the symmetry problem which was the key to the H-bomb. The complete discussion of the concept remained officially classified for decades after the H-bomb was invented. In contrast, the basic physics and outline of the engineering behind the A-bomb was made public in the 1940's already.
Some details had leaked out in one way or another by various 'exposes' but nothing really got it right. I believe it was finally declassified in about 1994 or so---I remember the article in the New York Times. By that point, intertial confinement fusion civilian researchers had already figured it out.
The trick is 'radiation driven implosion'.
The soft x-rays from a fission explosion are diverted down to a hollow tube (hohlraum {hollow room} in the German used by the early researchers). There is in fact a barrier, usually
of tungsten or some other heavy material directly in between the fission primary and the fusion secondary. This is to prevent the explosion of the fission material itself (the stuff) from getting to the fusion part and ruining it. You just want the photons, and if you block the shrapnel, the photons get there first.
The primary is basically many spherical layers---the secondary is a long cylinder with a few coincentric layers, with a gap near the outermost layer for the photons to flow into.
This form a 'photon gas'---but this fluid equilibrates at the speed of light and not the speed of sound, like normal fluids. Which means that the density and energy of the photons gets very uniform quickly. Although you don't normally feel it in every day life, photons carry momentum and can exert a pressure.
The A-bomb uses chemical explosives to precisely compress the fission fuel, but that isn't powerful and even and fast enough to work with fusion. The description in the article, putting the fusion fuel in the center of the bomb, is mostly incorrect. There is a little bit of fusion fuel in the middle of the a-bomb section but it does not make the really big BANG of the true H-bomb, it is just a booster to make more neutrons to make more fission from a given amount of plutonium or uranium.
Back to the photon fluid. This photon gas has enough energy to very very evenly and violently compress the fusion fuel. This is basically a 3-layered hot dog. The outermost layer is a heavy 'tamper' of lead or U-238 or something else. The middle layer is the fusion fuel. Then, there is a rod in the center with fission fuel like U-235 or Pu-239. The photon gas ablates the material on the tamper---the pressure from the photons and momentum from the ablation compress the fusion tube very severely. At the same time, it is magically arranged for just enough neutrons from the primary fission explosion to start a chain reaction in the fission stuff inside the fusion fuel, known as the 'spark plug'. Thus the fusion fuel is compressed from the outside by the photon gas and ablation and squeezed from the inside from the second fission explosion. This is how much work you need to do to get fusion. The fission material, normally very light, is compressed to a density near that of a white-dwarf star, where it reaches quantum mechanical Fermi degeneracy. That is really really really compressed. Not quite a neutron star (that is even more dense, being a giant atomic nucleus) but still thousands of times more dense than any conventional material.
Then, Boom.
The symmetrical lasers in the Livermore fusion experiments do NOT directly shine on the fusion pellet. That would not make sufficiently even compression. They shine on an outer metallic sphere which ionizes and releases X-rays, and these X-rays equilibrate inside the metallic shell (even though it's now totally vaporized it's still heavy compared to photons) and this compresses the fusion fuel.
This is why the intertially confined fusion program is always more classified than the magnetic fusion, because it is in large measure H-bomb technology. The "national ignition facility" being built for the DOE in the United States is yet another really big ICF project. Unfortuantely for political reasons it is being designed entirely from the point of view of military bomb research and not for any energy research. I think that's a shocking waste but there's a gazillion dollars to test the most obscure things about nukes when we already have thousands of bombs that work way way too well for our planet's and species' good.
The lasers are very power inefficient, but better for getting clean data for bomb research. Ion beams, and probably this Sandia machine are much more power efficient and the likely technology for a power plant, but less is known about them because it doesn't fit in the military application as well.
In any case, it is almost certain that this Sandia machine described in the article uses "indirect drive" {as the x-ray compression method is known} instead of "direct drive" and that is how they plan to work on the symmetry problem. So they probably do have an idea what to do about it, contrary to what the article says. It is still a very challenging problem, but not hope is lost.
The best information on the web about Big Bombs is Carey Sublette's archive. Just about everybody who Really Knows what's going on {and I'm not one of them} has said that the technical details are accurate. In fact, some of the information about the W88 warhead that the Chinese supposedly stole can be found on this website, and the Chinese showed this in their defense.
High energy weapons archive
I am a physicist, but not a nuclear physicist or involved in fusion research in any way whatsoever.
If you have two hypotheses e.g. A and B, corresponding to 'two words' which were said, then it is easy to build systems which can recognize signals corresponding to A and those corresponding to B embedded in lots of noise. Basically you measure the likelihood ratio p(B)/p(A) using some sort of estimators that you've trained to light up with either A or B. If you gave me the data, I could do this with a number of different semi-conventional numerical techniques on a digital computer. I've seen similar things presented at conferences a few years ago---recognition of specific chaotic waveforms (specifically dolphin and whale song) embedded in lots of noise.
This is known as a "simple hypothesis test".
The more general circumstance, however is that the alternative is not A vs B, but A vs a huge multitude of other possibilities. This task is much more difficult, and correponds to the actual large-vocabulary speech recognition task. Now it becomes much more difficult to set a reliable threshold which will come on only when A is actually present, and not when A is absent. There is a tradeoff of false negative and false positive errors depending on your choice of threshold.
There is no possible way that this thing can recognize 50,000 words. There are only 30 connections, there is fundamentally not enough information processing power intrinsically in there.
What you would do is to have all sorts of these subunits lighting up their own 'word finder lights', and the result of *those* (i.e. the p(A) detectors) would then be inputs into higher level semantic networks of perhaps a similar type. These networks or hidden markov models or whatever are the ones that know which sorts of words follow other sorts of words, and thus let you get better recognition than the individual word finders themselves.
So, what is the accomplishement of this paper??
That they've apparently found an extremely efficient and well-performing low-level subunit using this time-domain information. From our own experimental observations (not on speech but on real live neurons from recently-living animals) this is very important. The fact that it is only 30 connections might mean that it is quite feasible to put 10 or 20 thousands of these subunits on a single chip, running in hardware. Given the factor of a thousand speed increase of electronics over neurons if you could time-division multi-plex different recognizers (blue sky dreaming here!) you could have that much many more of them during the milliseconds to seconds of audio-frequency processing time that we speak at.
If you notice, Professor Berger said that no other speaker-independent system outperformed humans, even in small test bases. Presumably that means in the small Bayesian post-hoc sorts of likelihood test regimes taht I described before. And in addition, it appears that this is not a simulation but that they built it on an actual physical computer chip, another very substantial advance.
My colleagues are going to ask the authors for the actual paper. The title and press release may be overblown, but this smells like real science and a significant advance here to me.
Take home message: even small groups of good neurons can do interesting and useful things. With the right architecture, a small group of neurons can outperform conventional "neuroid networks" of hundreds or thousands of nodes linked by linear transformations of sigmoidal basis functions. We may just be beginning to crack real-AI.
We see major body functions of lower animals being regulated by say ten neurons. Real neurons are much smarter than you think. :)
If small groups of neurons can do this, it makes you appreciate what a hundred billion might be able to do.