Without logs, it seems it would be harder to track down network abuse (i.e. crackers). So you trade privacy for some protection from assholes. To me, that's a fair tradeoff, but what happens when the German courts demand that an ISP assist in some investigation and they can't because they've deleted certain logs (as the SAME courts told them they have to do)?
Seems like it puts the ISP between a very uncomfortable rock and a hard place.
Um.. Thermonuclear weapons are hard to make. REALLY hard. Fission-only weapons are pretty simple, but you have to avoid accidentally irradiating and killing yourself while building them. But a hydrogen bomb is not something that could be thrown together by anybody without very specific, deep knowledge of certain parts of physics, and a LOT of money and equipment. And the necessary knowledge is voluntarily kept secret by the few humans on the planet who actually know enough to build a working thermonuclear device. Kip Thorne's book "Black Holes And Time Warps" spends a few chapters discussing in detail the history of the development of thermonuclear weapons in both the US and the Soviet Union. I suggest reading it.
In fact, the first "hydrogen bomb" created by the United States was actually more like a BUILDING than a bomb. Nothing that could possibly be dropped by an airplane or otherwise delivered to an actual enemy. Although it did completely destroy the island of Enewetak it was situated on. Even if some rogue nation (or to be even more ridiculous some terrorist group) managed to build a device that could set off a thermonuclear explosion, they'd never be able to pack it into, say, a suitcase or deliver it in some neato-looking warhead, without a LOT of testing, further spending, and time invested. The fact that every technologically advanced society on earth would instantly detect any such test and pinpoint its location makes the whole idea even more far fetched.
The risk of stolen thermonuclear warheads is far far greater than any wild possibility of some third party manufacturing working hydrogen bombs.
Nontrivial yes, but it's arguable if the novelty rises to the level of something patentable. Compression algorithms based on energy-compressing transformations are nothing new, and JPEG was merely the first format to standardize a particular algorithm. The real hard work of JPEG was banging out a spec and agreeing on particulars like the exact layout of blocks, the exact values of quantization coefficients, etc. Very hard work I agree, but the basic idea which makes the whole thing work -- the idea of applying a DCT to concentrate image energy into bands and aggressively quantizing those bands -- was around before JPEG.
I've tried to find the actual patent online, but USPTO seems to be down at the moment and apparently my Google-fu is not good enough. So I can only guess as to the contents of the patent. But most compression-related patents apply to very particular aspects of the compression process, such as the particular method used for arithmetic encoding of a particular value, or some such. I pretty much guarantee that the patent does NOT cover the basic idea of energy-compressing transformations, which are the real meat of how JPEG does what it does.
Don't get too over-awed by the math. JPEG is clever but it's not exactly General Relativity.
A crash is one thing, but a re-install to fix it? I have my doubts, but if anyone can pull it off, it's Microsoft!
Not that hard to imagine, really. A filesystem driver bug that blows away critical tables in the filesystem could put you out of commission pretty quick. (I have no idea what the bug is but filesystem corruption is the most likely thing I can think of.)
Drug pushers destroy lives for their own profit, and they have some pretty devastating, instantly addictable weapons in their arsenal that they use to draw young people, particulary girls, into their net.
And it does not occur to you that drug pushers (which in the above statement you explicitly blame) COULD NOT EXIST if there were legal markets for these drugs?
Could underage/irresponsible people still get hold of these drugs if they were legal and controlled? Of course. But look at alcohol. There is a LEGAL market for alcohol, and as a result there is NO BLACK MARKET. If an underage kid wants to get booze, he/she has to go into a legally-licensed liquor store and try to weasel it. The HUGE difference between this scenario and the kid getting booze from some black market "dealer" is that the dealer is probably a low life and potentially a gun-toting psychotic, whereas if you piss off a liquor store clerk the worst that happens is you get kicked out of the store or picked up by the cops and brought home to daddy. Let's see the cops rescue you from a coked up freak in a little shack out in the woods 15 miles from the nearest house.
What's the real danger of cocaine? The scene. The dealers, the freaks, the criminals you have to hang around in order to get it. If a person wants coke, they have to descend into this dangerous environment whether they want to or not.
And as for little teenie bopper wanting to do coke in the first place? That's not cocaine's fault, it's her parents' fault.
Device drivers must, at some level, have a kernel component; because nothing in userland is allowed to talk to I/O ports.
Hrm? On Linux, we've had ioperm(2) since... I don't know, but probably forever. A userland process (with the appropriate capabilities, either root or CAP_SYS_RAWIO) can request the kernel to open access to specific ports. This doesn't require the process to run in Ring-0, and on x86 at least it doesn't even require the kernel to trap I/O requests -- they pass straight on through to the device. The CPU does it.
There's no need to put ANY device-specific code into the kernel as long as there is a reasonable way (within the chosen security model) to allow individual processes the access to hardware resources that are required. And I think any reasonably flexible architecture would allow that.
DMA arbitration obviously has to be done by the kernel because synchronization of requests from multiple processes would be impossible without it. But again, nothing about the kernel DMA interface has to be device-specific.
Yeah, I get it... But you don't have to use the language perfectly to be comprehensible. There's a big difference between a grammatical/spelling mistake here and there, and completely butchering the language. I think the sentence you quote is better formed than a lot of sentences I hear from native speakers.
Anyway, replace "capable" with "able" and the sentence is perfect. Or, replace "to understand" with "of understanding" and again, perfect. We're talking about nuances of the language here, not terrible errors.
The second paragraph in the comment you link to is false as it is written there. That's all there is to it. A newtonian model can be completely exact even if the time-steps are finite. As long as you choose proper circular coordinates.
What I should have said, instead of "the simplest method possible," was "the most immediately obvious method." To me, using cartesian coordinates is the most immediately obvious. Anyway, I'm interested in first order simulations in polar coordinates -- how well does this translate to N-body?
What the fuck are you talking about? Of COURSE N-body simulations are possible. Closed solutions are (probably) not. Do you think nobody simulates gravity? "Whoops, we can't get a closed form solution, guess we're fucked!" Right.
Re:lack of gravitational pull??
on
The Sun Had Sisters
·
· Score: 2, Informative
You're both wrong, of course. The order of discretization has nothinig to do with this, the naive choice of coordinates does.
This "naive" method is PRECISELY what we are discussing. Look at my comment. Notice my usage of the word "naive." Notice that you are not following the topic.
This thread is not about the impossibility of an energy-conserving first order method. It is SPECIFICALLY about the naive cartesian Euler method, which is what I presumed the OP implemented.
But hey, you get to show off your mad skillz... Fine.
I mean translational kinetic energy, fergawdsake. Good grief, must we wallow in Dictionopolis definitioneering when there's interesting astrophysics to ponder?
There is a huge translational kinetic energy involved -- the translation of each individual particle in the neutron star as it rotates. The body viewed as a whole has no translational energy. The sum of the translational energies of all the particles is equal to the rotational K.E. of the body. This energy can't just disappear when the body breaks up -- the remnants are hurled in all directions. Because the body had a total translational momentum of zero, the distribution of the hurled fragments must be equal in all directions in order for the total momentum to stay the same. But stuff's moving. Really fast.
And the exact definitions of all these quantities are VITAL to understanding correctly what's going on...
I've had an idea for a while for an analog computer which computes orbits.
First step, acquire a star. Second step, acquire a planet and place it in orbit around the star. Record what happens. See, it's an analog computer that calculates orbits!
Aye, but where's the supernova remnant itself? The rapidly-rotating neutron star with the nasty high-energy pulsar radiation? It was at the center of the explosion, so it had an initial kinetic energy of nearly zero.
A massive, rapidly rotating object has a kinetic energy of nearly zero? What professor did you hear this from, he/she needs a beatdown.
A few things that commonly confuse people: 1) Kinetic energy is NOT conserved. TOTAL ENERGY is. 2) Momentum and kinetic energy are NOT THE SAME THING. 3) A total momentum of zero does NOT mean that nothing is moving.
for even a simple two-body system using the inverse-square law, the orbiting object will spiral outward due to accumulated discretization error.
Yep, and it's easy to see why without even doing any math. Assuming the initial conditions are set up for a circular orbit, the body's initial velocity is at 90 degrees to the vector to the "sun". In the first timestep, the body will move only along this perpendicular direction, which means it ends up further from the "sun" than when it started (which is wrong already, because for a perfectly circular orbit the radius does not change). This will happen no matter how small you make the timestep.
Write a program using standard Newtonian physics that takes as its input a cluster of 'stars' of various masses. Start the program. After a while there is nothing left because all og the stars havesling shoted themselfs off to never never land.
There are two reasons this happens. First, even in a BOUND gravitational system (potential energy plus kinetic energy is a negative quantity), individual bodies can still be ejected as long as the total energy remains the same. However, in the real universe there will always be at least two bodies which remain bound in orbit. In a simulation this is often not the case, because naive (by that I mean simple, not stupid) methods are not energy-conserving.
Imagine the simplest gravitational simulation possible, where at each timestep the total forces on all the bodies are computed and accelerations determined. The acceleration is then applied over the timestep to change the velocity of each body, and then the velocity is (again) applied over the timestep to compute the updated positions. The problem is, the timesteps are not infinitesimals. This means that a perfect balance between potential and kinetic energy is not achieved, and numerical errors result. Usually, kinetic energy wins, and all bodies are eventually ejected from the simulation.
So the claim is that hundreds, maybe thousands, of sun-like stars were in close proximity to each other, but they didn't generate enough gravity to stay in the same neighborhood? How does that make any kind of sense?
Allow me to introduce my good friend, Kinetic Energy.
Ok great! Now only the honest kids will be prevented from signing up to most forums.
What does "honest" mean? Teaching my kids honesty would be one of my top priorities, but I'd also tell them to feel free to click whatever the hell they want on the web. Being "honest" when clicking a radio button on a web site is not one of my top moral priorities.
Oh noes! I lied to a computer program! It must feel so... hurt.
Slashdot Mirror
Are you kidding? Because that's pretty damn funny (and very sad) if true.
Without logs, it seems it would be harder to track down network abuse (i.e. crackers). So you trade privacy for some protection from assholes. To me, that's a fair tradeoff, but what happens when the German courts demand that an ISP assist in some investigation and they can't because they've deleted certain logs (as the SAME courts told them they have to do)?
Seems like it puts the ISP between a very uncomfortable rock and a hard place.
Um.. Thermonuclear weapons are hard to make. REALLY hard. Fission-only weapons are pretty simple, but you have to avoid accidentally irradiating and killing yourself while building them. But a hydrogen bomb is not something that could be thrown together by anybody without very specific, deep knowledge of certain parts of physics, and a LOT of money and equipment. And the necessary knowledge is voluntarily kept secret by the few humans on the planet who actually know enough to build a working thermonuclear device. Kip Thorne's book "Black Holes And Time Warps" spends a few chapters discussing in detail the history of the development of thermonuclear weapons in both the US and the Soviet Union. I suggest reading it.
In fact, the first "hydrogen bomb" created by the United States was actually more like a BUILDING than a bomb. Nothing that could possibly be dropped by an airplane or otherwise delivered to an actual enemy. Although it did completely destroy the island of Enewetak it was situated on. Even if some rogue nation (or to be even more ridiculous some terrorist group) managed to build a device that could set off a thermonuclear explosion, they'd never be able to pack it into, say, a suitcase or deliver it in some neato-looking warhead, without a LOT of testing, further spending, and time invested. The fact that every technologically advanced society on earth would instantly detect any such test and pinpoint its location makes the whole idea even more far fetched.
The risk of stolen thermonuclear warheads is far far greater than any wild possibility of some third party manufacturing working hydrogen bombs.
Nontrivial yes, but it's arguable if the novelty rises to the level of something patentable. Compression algorithms based on energy-compressing transformations are nothing new, and JPEG was merely the first format to standardize a particular algorithm. The real hard work of JPEG was banging out a spec and agreeing on particulars like the exact layout of blocks, the exact values of quantization coefficients, etc. Very hard work I agree, but the basic idea which makes the whole thing work -- the idea of applying a DCT to concentrate image energy into bands and aggressively quantizing those bands -- was around before JPEG.
I've tried to find the actual patent online, but USPTO seems to be down at the moment and apparently my Google-fu is not good enough. So I can only guess as to the contents of the patent. But most compression-related patents apply to very particular aspects of the compression process, such as the particular method used for arithmetic encoding of a particular value, or some such. I pretty much guarantee that the patent does NOT cover the basic idea of energy-compressing transformations, which are the real meat of how JPEG does what it does.
Don't get too over-awed by the math. JPEG is clever but it's not exactly General Relativity.
Can some law enforcement agency use the GPS data to prove that you did something illegal for example?
If you are planning to do something illegal (or you're just paranoid), why would you carry a phone with you that tracks your every movement?
Same here, version 10.3.9.
A crash is one thing, but a re-install to fix it? I have my doubts, but if anyone can pull it off, it's Microsoft!
Not that hard to imagine, really. A filesystem driver bug that blows away critical tables in the filesystem could put you out of commission pretty quick. (I have no idea what the bug is but filesystem corruption is the most likely thing I can think of.)
Install a gigantic solar system / battery storage.
Great idea, it would only cost about $16 million dollars...Drug pushers destroy lives for their own profit, and they have some pretty devastating, instantly addictable weapons in their arsenal that they use to draw young people, particulary girls, into their net.
And it does not occur to you that drug pushers (which in the above statement you explicitly blame) COULD NOT EXIST if there were legal markets for these drugs?
Could underage/irresponsible people still get hold of these drugs if they were legal and controlled? Of course. But look at alcohol. There is a LEGAL market for alcohol, and as a result there is NO BLACK MARKET. If an underage kid wants to get booze, he/she has to go into a legally-licensed liquor store and try to weasel it. The HUGE difference between this scenario and the kid getting booze from some black market "dealer" is that the dealer is probably a low life and potentially a gun-toting psychotic, whereas if you piss off a liquor store clerk the worst that happens is you get kicked out of the store or picked up by the cops and brought home to daddy. Let's see the cops rescue you from a coked up freak in a little shack out in the woods 15 miles from the nearest house.
What's the real danger of cocaine? The scene. The dealers, the freaks, the criminals you have to hang around in order to get it. If a person wants coke, they have to descend into this dangerous environment whether they want to or not.
And as for little teenie bopper wanting to do coke in the first place? That's not cocaine's fault, it's her parents' fault.
Device drivers must, at some level, have a kernel component; because nothing in userland is allowed to talk to I/O ports.
Hrm? On Linux, we've had ioperm(2) since... I don't know, but probably forever. A userland process (with the appropriate capabilities, either root or CAP_SYS_RAWIO) can request the kernel to open access to specific ports. This doesn't require the process to run in Ring-0, and on x86 at least it doesn't even require the kernel to trap I/O requests -- they pass straight on through to the device. The CPU does it.
There's no need to put ANY device-specific code into the kernel as long as there is a reasonable way (within the chosen security model) to allow individual processes the access to hardware resources that are required. And I think any reasonably flexible architecture would allow that.
DMA arbitration obviously has to be done by the kernel because synchronization of requests from multiple processes would be impossible without it. But again, nothing about the kernel DMA interface has to be device-specific.
I hadn't even noticed those errors, which is a testament to their irrelevance.
If the bids are secret, it's not really bidding anymore. It's just the usual business of "pick the lowest price."
Yeah, I get it... But you don't have to use the language perfectly to be comprehensible. There's a big difference between a grammatical/spelling mistake here and there, and completely butchering the language. I think the sentence you quote is better formed than a lot of sentences I hear from native speakers.
Anyway, replace "capable" with "able" and the sentence is perfect. Or, replace "to understand" with "of understanding" and again, perfect. We're talking about nuances of the language here, not terrible errors.
I can't find any other word for it.
Hah! So hilarious... Truly, I am speechless.The second paragraph in the comment you link to is false as it is written there. That's all there is to it. A newtonian model can be completely exact even if the time-steps are finite. As long as you choose proper circular coordinates.
What I should have said, instead of "the simplest method possible," was "the most immediately obvious method." To me, using cartesian coordinates is the most immediately obvious. Anyway, I'm interested in first order simulations in polar coordinates -- how well does this translate to N-body?
What the fuck are you talking about? Of COURSE N-body simulations are possible. Closed solutions are (probably) not. Do you think nobody simulates gravity? "Whoops, we can't get a closed form solution, guess we're fucked!" Right.
You're both wrong, of course. The order of discretization has nothinig to do with this, the naive choice of coordinates does.
This "naive" method is PRECISELY what we are discussing. Look at my comment. Notice my usage of the word "naive." Notice that you are not following the topic.
This thread is not about the impossibility of an energy-conserving first order method. It is SPECIFICALLY about the naive cartesian Euler method, which is what I presumed the OP implemented.
But hey, you get to show off your mad skillz... Fine.
I mean translational kinetic energy, fergawdsake. Good grief, must we wallow in Dictionopolis definitioneering when there's interesting astrophysics to ponder?
There is a huge translational kinetic energy involved -- the translation of each individual particle in the neutron star as it rotates. The body viewed as a whole has no translational energy. The sum of the translational energies of all the particles is equal to the rotational K.E. of the body. This energy can't just disappear when the body breaks up -- the remnants are hurled in all directions. Because the body had a total translational momentum of zero, the distribution of the hurled fragments must be equal in all directions in order for the total momentum to stay the same. But stuff's moving. Really fast.
And the exact definitions of all these quantities are VITAL to understanding correctly what's going on...
I've had an idea for a while for an analog computer which computes orbits.
First step, acquire a star. Second step, acquire a planet and place it in orbit around the star. Record what happens. See, it's an analog computer that calculates orbits!
Aye, but where's the supernova remnant itself? The rapidly-rotating neutron star with the nasty high-energy pulsar radiation? It was at the center of the explosion, so it had an initial kinetic energy of nearly zero.
A massive, rapidly rotating object has a kinetic energy of nearly zero? What professor did you hear this from, he/she needs a beatdown.
A few things that commonly confuse people: 1) Kinetic energy is NOT conserved. TOTAL ENERGY is. 2) Momentum and kinetic energy are NOT THE SAME THING. 3) A total momentum of zero does NOT mean that nothing is moving.
for even a simple two-body system using the inverse-square law, the orbiting object will spiral outward due to accumulated discretization error.
Yep, and it's easy to see why without even doing any math. Assuming the initial conditions are set up for a circular orbit, the body's initial velocity is at 90 degrees to the vector to the "sun". In the first timestep, the body will move only along this perpendicular direction, which means it ends up further from the "sun" than when it started (which is wrong already, because for a perfectly circular orbit the radius does not change). This will happen no matter how small you make the timestep.
Write a program using standard Newtonian physics that takes as its input a cluster of 'stars' of various masses. Start the program. After a while there is nothing left because all og the stars havesling shoted themselfs off to never never land.
There are two reasons this happens. First, even in a BOUND gravitational system (potential energy plus kinetic energy is a negative quantity), individual bodies can still be ejected as long as the total energy remains the same. However, in the real universe there will always be at least two bodies which remain bound in orbit. In a simulation this is often not the case, because naive (by that I mean simple, not stupid) methods are not energy-conserving.
Imagine the simplest gravitational simulation possible, where at each timestep the total forces on all the bodies are computed and accelerations determined. The acceleration is then applied over the timestep to change the velocity of each body, and then the velocity is (again) applied over the timestep to compute the updated positions. The problem is, the timesteps are not infinitesimals. This means that a perfect balance between potential and kinetic energy is not achieved, and numerical errors result. Usually, kinetic energy wins, and all bodies are eventually ejected from the simulation.
So the claim is that hundreds, maybe thousands, of sun-like stars were in close proximity to each other, but they didn't generate enough gravity to stay in the same neighborhood? How does that make any kind of sense?
Allow me to introduce my good friend, Kinetic Energy.
Ok great! Now only the honest kids will be prevented from signing up to most forums.
What does "honest" mean? Teaching my kids honesty would be one of my top priorities, but I'd also tell them to feel free to click whatever the hell they want on the web. Being "honest" when clicking a radio button on a web site is not one of my top moral priorities.
Oh noes! I lied to a computer program! It must feel so... hurt.
If parents raise their children in a halfway decent manner, having them exposed to some awful sites will cause revulsion but not harm.
Ahh, the sight of the naked female form. Yep, that should definitely inspire revulsion.
I've got an idea that will solve this once and for all. From here on out, all women must wear garments covering their bodies from head to toe.