There is a well-known story by Larry Niven called "Hole Man", but it takes place on Mars. The device is an alien gravity wave communicator. The explorers find out that it contains a black hole, which is used as a murder weapon. This was before Hawkings radiation was known : the black hole should have been very bright but was invisible in the short story. The story ends with the head scientist, who commited the murder, musing that Mars will probably disappear in a few years (but the Sun is safe).
The BH is mostly radiating photons, not much massed particles. Photons are created from the vacuum around the BH horizon, and in pairs. As you know antiphotons are the same as photons, it doesn't matter which falls in the BH.
You can fall a lengthy discussion of this here for instance. (I don't generally advocate physicsforum unless you want to waste a lot of time in the details of specifics bits of physics endlessly debated).
However, this may answer your question regarding particle-antiparticle. It is much harder and much more unlikely to produce particles other than photons (photons can be produced of any energy. Particles require very precise energies to be created). BH may also produce other form of energy such as neutrinos/antineutrinos but those are massed and still much more unlikely than photon pairs.
Most importantly, the Hawkings radiation (HR) is just an explanation mechanism for the fact that BH have mass and temperature. If a BH has a temperature then it must radiate by very basic thermodynamics. HR just provides a plausible mechanism for the radiation.
If a massed particle-antiparticle pair is produced near a BH, either can fall in, not just the antiparticle if the BH is made of matter, or the other way around. As stated earlier this event is unlikely, even nigh on impossible near large BH, which polarise the vacuum very little near the horizon.
Conversely, a very tiny BH should be able to polarize the vacuum enough to produce any particle. This is right before the moment it explodes though, in a display you don't want to be near:-)
Heuristics are human-written rules that are about right most of the time, or totally correct a little of the time, depending on the context. They are useful but do not define AI.
A* is a reformulation of Branch-and-Bound found in constrained optimisation (think Integer Programming). Both A* and IP use heuristics a lot.
Complex Bayesian models did not originate in AI but in statistics and optimisation (Metropolis algorithm for instance).
To me AI is a lot about applying algorithms developed elsewhere, with few exceptions. Not that this is uninteresting or not clever, but AI does not have a track record of supplying applied mathematicians with useful algorithms, rather the other way around.
According to Marxist doctrine, communism starts with a dictature of the proletariat (the workers). Then it gradually moves onto paradise once everybody is educated.
Unfortunately the very first step is never correctly executed, never leading onto the paradise bit. The dictature they do have though.
I wouldn't say that, really. For instance calculus was invented to solve a series of physics problems, most prominently point dynamics.
The relationship between maths and physics (all of science really) goes both way. If you read the article, you'll see that the original astrophysics conjectures provided an instance of maximal solution that the mathematicians hadn't found.
If you read the history of general relativity, you'll find that Einstein did a lot of significant work with David Hilbert, the leading mathematician of the early 20th century, on tensor theory. However it was Einstein who came up with the correct formulation of the gravity theory thanks to his physical insight (coming from Mach's principle). It's a fascinating story.
From TFA it's not very clear what this simulation achieved. It was code that already existed and as far as I understand it, it was used to validate some simulation models of low-level biological vision.
However his simulation did not necessarily achieve computer vision in the usual sense, i.e: shape recognition, image segmentation, 3D vision, etc. This is the more cognitive aspect of the visual processus, which at present requires a much higher level of understanding of the vision process that we do not posess.
FYI the whole brain has already been simulated, see the work of Dr izhikevich. It took several months to simulate about 1 second of brain activity.
However this experiment did not simulate thought, just vast amounts of simulated neurons firing together. The simulated brain exhibited large-scale electrical behaviours of the type seen in EEG plots, but this is about it.
This experiment sounds very similar. I'm not all that excited yet.
Multithreading a single-threaded program on even two cores *optimally* is NP-hard, but one can come up with any number of efficiently implemented sub-optimal heuristics (e.g: ignore all cores but one;-)
Yes, but he was still (very) useful to the allies after the war, so he didn't hang. I'm not defending him, he is just the product of his time.
Without WWII, WVB would probably have just been an academic gentleman interested in rocketry. He would probably have caused not a single death and not even become a footnote to history, like most of us.
Conversely you will find many people outside the US that think many recent US presidents should be tried as war criminials.
In the real world morals don't count for too much if you have the connections, the power, or something no one else has. Is that news to you?
V1 and V2 rockets were fair weapons in war. War is ugly. Do you think the firebombing of Dresden and Tokyo were fun happenings?
The slave labor is more of a problem, although it wasn't his idea and claimed he didn't like it but could do nothing to change it. I don't buy that myself.
Von Braun is controversial but helped build the Saturn V booster.
Academic journal are still, or even increasingly important in many areas of science, even though stuff that is published in them is essentially known before it is printed.
Before an article makes it into a journal, its content is usually published online as a report (either through the home institution or on arxiv), then at one or two conferences (where papers are reviewed too, but more quickly). By the time it gets reviewed for a journal, competent reviewers usually have heard of some of its content, which is good.
The journal paper however usually contains more data, more details, more discussion and better results than the previous incarnations. It has also been scrutinised and criticised a whole lot more. It has probably been revised completely at least once. This is a very different "product" than the initial report or conference stuff.
Nowadays the whole review process is online and often double-blind.
If a journal article has taken 2 years to be published it was probably because the authors didn't do a very good job of writing the first version of the article. The whole idea is to make the article's material into a reference.
Most researchers will then look up the article through web interfaces such as the IEEE's, the ACM's, the web of science, etc.
Scientists go through this trouble (they are both the authors, the reviewers and the editor -- not all at one of course) because it is worth it. No one has found a better system. After it is published, a good article will get cited often, and so the meritocratic aspect of science doesn't stop at publication.
In addition, the value of individual scientists is estimated through their paper output: the number of papers published, how often they are cited. Scientist have a strong incentive to publish quality new research, which is as it should be.
Eventually, if the stuff is good, it ends up as a book chapter, or even a whole book.
This is for image analysis and computer vision stuff BTW. It may vary significantly in other areas.
So no, journals are not obsolete. Since they are now easily indexed and searchable, they have become even more valuable and valued. What has changed is that institution have been able to bargain prices down, since paper issues are rarely used now and so the cost of running a journal has gone down. Journals that have an easy and relatively cheap subscription model have been able to get more mindshare, their "impact factor" have gone up, and their value as well. For instance, it is perfectly possible for an individual to subscribe to the IEEE and get some or most of its online library access at a reasonable price. This was unthinkable only a few years ago.
Of course replacing a surgeon with a reliable fully automated robot would be great.
However your description of surgery is not correct. Surgery is difficult, minutious and different for ever patient. Great surgeons must be able to plan ahead, direct a team and control all the details of a surgery procedure as it happens, as well as improvising with a cool head for hours on end if things go wrong.
It's the exact opposite of rote procedure. Especially now with recent advances in real-time non-invasive imaging and haptic instruments procedures change all the time.
So if no one wants to upgrade to a newer, competitive version of Windows, this means computer users will be stuck with the same old viruses, users-as-administrators, drab 32-bit world for years to come?
this is true, but the killer is Win64. To get over the memory hurdle one has to go 64-bit and 8GB+. Unfortunately the Win64 driver and applicaiton situation is bad and is likely to remain that way for long.
Meanwhile Linux has done its 64-bit aggiornamento more than a decade ago and all is well now.
Of course Friedman is very smart, but he is a bit of a one-trick pony.
I'm not so sure capitalism and education mix all that well. At the top end where college competition is the norm, it has simply become very expensive to get a college education and I think this is not a good thing. Fewer and fewer Americans are interested in getting educated: they are interested in learning a well-paying trade in as short as possible timeframe to limit their education debt. This might be why so few Americans know what is happening outside their borders, with end results such as Iraq.
If primary and secondary school start working under that model, what about the very common cases of difficult children? How is the gov. going to check whether minimum standards have actually been taught? Competitive exams every year? What happens when a large conglomerate becomes the de-facto monopoly and start lobbying in Congress to lower the education standards so that it lowers their costs? We've all seen in other contexts that de-facto monopoly can be hard to prove and act upon. Who wants Microsoft-like education quality standards?
This could get real ugly. The US is welcome to try, but I'm not optimistic.
A good statistically valid experiment will get you much more accurate results than all the modeling in the world. It's the difference between a weather simulator and going outside.
Interesting analogy. I seem to observe that weather models commonly allow for 2-5 day relatively accurate weather predictions.
Going outside usually doesn't help you predict tomorrow's weather too well.
The ARM design dates from 1983, so it's now 25 years old. The R is for RISC, so it's not that novel.
Efficiency of an architecture has very little to do with its instruction set. In the case of modern x86 chips they are emulated in microcode anyway.
ARM consumes less essentially because it runs at a few hundred MHz or less, not GHz, and has features for running in embedded environment with not much memory.
Slashdot Mirror
Thanks for the thanks!
This is about making non-cristalline, amorphous metals, but TFA does mention that the result is NOT transparent (rather shiny black).
There is a well-known story by Larry Niven called "Hole Man", but it takes place on Mars. The device is an alien gravity wave communicator. The explorers find out that it contains a black hole, which is used as a murder weapon. This was before Hawkings radiation was known : the black hole should have been very bright but was invisible in the short story. The story ends with the head scientist, who commited the murder, musing that Mars will probably disappear in a few years (but the Sun is safe).
The BH is mostly radiating photons, not much massed particles. Photons are created from the vacuum around the BH horizon, and in pairs. As you know antiphotons are the same as photons, it doesn't matter which falls in the BH.
You can fall a lengthy discussion of this here for instance. (I don't generally advocate physicsforum unless you want to waste a lot of time in the details of specifics bits of physics endlessly debated).
However, this may answer your question regarding particle-antiparticle. It is much harder and much more unlikely to produce particles other than photons (photons can be produced of any energy. Particles require very precise energies to be created). BH may also produce other form of energy such as neutrinos/antineutrinos but those are massed and still much more unlikely than photon pairs.
Most importantly, the Hawkings radiation (HR) is just an explanation mechanism for the fact that BH have mass and temperature. If a BH has a temperature then it must radiate by very basic thermodynamics. HR just provides a plausible mechanism for the radiation.
If a massed particle-antiparticle pair is produced near a BH, either can fall in, not just the antiparticle if the BH is made of matter, or the other way around. As stated earlier this event is unlikely, even nigh on impossible near large BH, which polarise the vacuum very little near the horizon.
Conversely, a very tiny BH should be able to polarize the vacuum enough to produce any particle. This is right before the moment it explodes though, in a display you don't want to be near :-)
Heuristics are human-written rules that are about right most of the time, or totally correct a little of the time, depending on the context. They are useful but do not define AI.
A* is a reformulation of Branch-and-Bound found in constrained optimisation (think Integer Programming). Both A* and IP use heuristics a lot.
Complex Bayesian models did not originate in AI but in statistics and optimisation (Metropolis algorithm for instance).
To me AI is a lot about applying algorithms developed elsewhere, with few exceptions. Not that this is uninteresting or not clever, but AI does not have a track record of supplying applied mathematicians with useful algorithms, rather the other way around.
According to Marxist doctrine, communism starts with a dictature of the proletariat (the workers). Then it gradually moves onto paradise once everybody is educated.
Unfortunately the very first step is never correctly executed, never leading onto the paradise bit. The dictature they do have though.
I wouldn't say that, really. For instance calculus was invented to solve a series of physics problems, most prominently point dynamics.
The relationship between maths and physics (all of science really) goes both way. If you read the article, you'll see that the original astrophysics conjectures provided an instance of maximal solution that the mathematicians hadn't found.
If you read the history of general relativity, you'll find that Einstein did a lot of significant work with David Hilbert, the leading mathematician of the early 20th century, on tensor theory. However it was Einstein who came up with the correct formulation of the gravity theory thanks to his physical insight (coming from Mach's principle). It's a fascinating story.
From TFA it's not very clear what this simulation achieved. It was code that already existed and as far as I understand it, it was used to validate some simulation models of low-level biological vision.
However his simulation did not necessarily achieve computer vision in the usual sense, i.e: shape recognition, image segmentation, 3D vision, etc. This is the more cognitive aspect of the visual processus, which at present requires a much higher level of understanding of the vision process that we do not posess.
FYI the whole brain has already been simulated, see the work of Dr izhikevich. It took several months to simulate about 1 second of brain activity.
However this experiment did not simulate thought, just vast amounts of simulated neurons firing together. The simulated brain exhibited large-scale electrical behaviours of the type seen in EEG plots, but this is about it.
This experiment sounds very similar. I'm not all that excited yet.
and a 32-bit memory model. It still crashed though.
Multithreading a single-threaded program on even two cores *optimally* is NP-hard, but one can come up with any number of efficiently implemented sub-optimal heuristics (e.g: ignore all cores but one ;-)
That is one of the greatest comment on Slashdot ever.
Yes, but he was still (very) useful to the allies after the war, so he didn't hang. I'm not defending him, he is just the product of his time.
Without WWII, WVB would probably have just been an academic gentleman interested in rocketry. He would probably have caused not a single death and not even become a footnote to history, like most of us.
Conversely you will find many people outside the US that think many recent US presidents should be tried as war criminials.
In the real world morals don't count for too much if you have the connections, the power, or something no one else has. Is that news to you?
V1 and V2 rockets were fair weapons in war. War is ugly. Do you think the firebombing of Dresden and Tokyo were fun happenings?
The slave labor is more of a problem, although it wasn't his idea and claimed he didn't like it but could do nothing to change it. I don't buy that myself.
Von Braun is controversial but helped build the Saturn V booster.
Academic journal are still, or even increasingly important in many areas of science, even though stuff that is published in them is essentially known before it is printed.
Before an article makes it into a journal, its content is usually published online as a report (either through the home institution or on arxiv), then at one or two conferences (where papers are reviewed too, but more quickly). By the time it gets reviewed for a journal, competent reviewers usually have heard of some of its content, which is good.
The journal paper however usually contains more data, more details, more discussion and better results than the previous incarnations. It has also been scrutinised and criticised a whole lot more. It has probably been revised completely at least once. This is a very different "product" than the initial report or conference stuff.
Nowadays the whole review process is online and often double-blind.
If a journal article has taken 2 years to be published it was probably because the authors didn't do a very good job of writing the first version of the article. The whole idea is to make the article's material into a reference.
Most researchers will then look up the article through web interfaces such as the IEEE's, the ACM's, the web of science, etc.
Scientists go through this trouble (they are both the authors, the reviewers and the editor -- not all at one of course) because it is worth it. No one has found a better system. After it is published, a good article will get cited often, and so the meritocratic aspect of science doesn't stop at publication.
In addition, the value of individual scientists is estimated through their paper output: the number of papers published, how often they are cited. Scientist have a strong incentive to publish quality new research, which is as it should be.
Eventually, if the stuff is good, it ends up as a book chapter, or even a whole book.
This is for image analysis and computer vision stuff BTW. It may vary significantly in other areas.
So no, journals are not obsolete. Since they are now easily indexed and searchable, they have become even more valuable and valued. What has changed is that institution have been able to bargain prices down, since paper issues are rarely used now and so the cost of running a journal has gone down. Journals that have an easy and relatively cheap subscription model have been able to get more mindshare, their "impact factor" have gone up, and their value as well. For instance, it is perfectly possible for an individual to subscribe to the IEEE and get some or most of its online library access at a reasonable price. This was unthinkable only a few years ago.
Of course replacing a surgeon with a reliable fully automated robot would be great.
However your description of surgery is not correct. Surgery is difficult, minutious and different for ever patient. Great surgeons must be able to plan ahead, direct a team and control all the details of a surgery procedure as it happens, as well as improvising with a cool head for hours on end if things go wrong.
It's the exact opposite of rote procedure. Especially now with recent advances in real-time non-invasive imaging and haptic instruments procedures change all the time.
Textmate doesn't. The web page says version 1.5.7 requires 10.4.2
:-)
Textmate is overrated anyway
So if no one wants to upgrade to a newer, competitive version of Windows, this means computer users will be stuck with the same old viruses, users-as-administrators, drab 32-bit world for years to come?
This doesn't sound like progress much.
this is true, but the killer is Win64. To get over the memory hurdle one has to go 64-bit and 8GB+. Unfortunately the Win64 driver and applicaiton situation is bad and is likely to remain that way for long.
Meanwhile Linux has done its 64-bit aggiornamento more than a decade ago and all is well now.
Of course Friedman is very smart, but he is a bit of a one-trick pony.
I'm not so sure capitalism and education mix all that well. At the top end where college competition is the norm, it has simply become very expensive to get a college education and I think this is not a good thing. Fewer and fewer Americans are interested in getting educated: they are interested in learning a well-paying trade in as short as possible timeframe to limit their education debt. This might be why so few Americans know what is happening outside their borders, with end results such as Iraq.
If primary and secondary school start working under that model, what about the very common cases of difficult children? How is the gov. going to check whether minimum standards have actually been taught? Competitive exams every year? What happens when a large conglomerate becomes the de-facto monopoly and start lobbying in Congress to lower the education standards so that it lowers their costs? We've all seen in other contexts that de-facto monopoly can be hard to prove and act upon. Who wants Microsoft-like education quality standards?
This could get real ugly. The US is welcome to try, but I'm not optimistic.
Interesting analogy. I seem to observe that weather models commonly allow for 2-5 day relatively accurate weather predictions.
Going outside usually doesn't help you predict tomorrow's weather too well.
So models have their places.
The ARM design dates from 1983, so it's now 25 years old. The R is for RISC, so it's not that novel.
Efficiency of an architecture has very little to do with its instruction set. In the case of modern x86 chips they are emulated in microcode anyway.
ARM consumes less essentially because it runs at a few hundred MHz or less, not GHz, and has features for running in embedded environment with not much memory.
Obviously this is not twice the adress space but 2^32 times the adress space, i.e roughly 4 billion.
So the managerial bit is done, right? Come on, physicists, get a move on away from the continuous universe, slashdotters are bored.
Thanks for your input and vision, chief!
The last level is quite long, as well.
Best. comment. evar.