Genocide is intentional action to destroy a people. I suppose the Russians might have *intended* to invade and destroy a people but got bored after the first couple hundred.
Energy return on energy invested is the basic idea: how many barrels of oil do you get if you invest one barrel of oil in production? Shale oil and tar sands are quite a bit below the world average, which means middle east conventional oil must be considerably above that average.
Now, this is my speculation, but a bit of math with fuel prices would probably support it. Oil is a famously inelastic resource, meaning it doesn't really obey the law of supply and demand; demand is pretty much fixed, because transportation isn't something we can easily use less of. So the Saudi's are printing money, but shale oil and tar sands do okay too, even with their crappy EROI, because the price of oil is inflated due to its necessity for our transportation infrastructure. Because of the price of oil is inflated relative to its intrinsic energy value, almost nobody burns it to make electricity.
By the way, the conclusion of that paper is basically that we're screwed unless we improve our energy production technology. Economic prosperity is closely related to EROI; if you have low EROI you spend all your resources trying to produce energy, not much left over for other things. Fossil fuel EROI is all decreasing towards the danger zone. Hydro power has good EROI, but is mostly exploited. Nukes would work, at least for a while. Wind and solar are okay, at least in ideal locations, but need to improve in order to maintain our current prosperity.
Also, the much maligned aether, a substance that permeates all of space and (a bit later, the medium through which light propagates), is basically how you'd describe quantum fields (from the standard model) and space-time (from relativity).
North Korea isn't building nukes to hit the US. They're building nukes to hit American expeditionary forces attacking North Korea. Maybe with a few left over to lob at Japan and South Korea.
Stop buying the media hype. The North Korean leadership isn't crazy. Ever since Russia built one, nukes are for self defence through deterrence. North Korea has demonstrated the ability to build them, removing invasion from the options the US can consider.
Yes. That's what I said. The best missile defence is one that doesn't work. If the US thinks their missile defence works, they might do something stupid. If they know it doesn't, less chance of stupid. If the Russians or Chinese or whoever don't think it works, MAD is maintained as usual. If they think it does work, for whatever reason, they're less likely to launch, not more.
Good. Missile defences that others think might work are much better for stability than ones that host countries know will work. The defender knows they're vulnerable, so they don't do anything stupid. Potential attackers can't be sure an attack would succeed, so they don't do anything stupid either.
This is a good demonstration of something few people realize: oil, outside of the conventional reserves in the middle east and a few other places, isn't really a viable energy source. American shale oil extraction is energy positive, but not enough to be really worthwhile. Oil's value is mostly as an energy storage mechanism to power transportation.
So it doesn't make sense to burn that oil to power your extraction operations. It does make sense to build a solar facility to power extraction.
In quantum mechanics a vacuum is defined as the lowest possible energy state. The uncertainty principle prohibits zero energy states, so a vacuum always has some base energy, which can be thought of as a sea of virtual particles popping into and out of existence.
Space that includes a strong magnetic field and particles zipping through it isn't really a vacuum, but I think the point of the calculation is that the virtual particles in an actual vacuum can be polarized a little bit by a strong magnetic field and will thus act as a medium with some small but non-unity index of refraction. That will slightly lower the speed of light, causing very high energy charged particles to emit Cherenkov radiation.
That's not entirely true. There's error in our measurement, yes, but there's also error in the theoretical prediction.
Most modern theories require some approximation. QED, for example, posits that any particle interaction will be affected by an essentially infinite sum of contributions. Since these contributions become smaller and smaller, you can add up the first few and get a pretty good prediction... but not perfect.
Still, instead of "most precise theory" a better description might be "most precisely validated theory."
Precision is important because higher energy phenomena can manifest as small anomalies in lower energy interactions. For example, an alternative to building a bigger collider, able to directly generate high energy conditions, is to build specialty colliders and detectors that can more precisely measure lower-energy interactions. This is what the g2 experiment at Fermilab is doing: precision measurements of the magnetic moment of the muon disagree with theoretical predictions, possibly indicating some higher-energy new physics.
Arxiv, which was mentioned by the GP, is funded by public grants and publishes papers for around $6 per, including the cost of developing and maintaining the platform.
The two big AI journals publish their budgets, and they both publish papers for a couple bucks per, half of which is to register a DOI.
The problem with all the open publishing initiatives is that they're aimed directly at the subscription model. That's not the problem. The problem is that scientific publishing, no matter how you pay for it, is currently ridiculously expensive.
Easy enough to get around that. Make it a loan, with an interest rate and payback schedule calculated to yield 15% of the maximum expected salary over two years. Then forgive any overrun if the salary is less than the max.
You'd have to be completely nuts to sign such a thing, but people are stupid.
Kind of. NASA has said they didn't plan an all-female spacewalk, it just happened in the schedule. It then got hyped, probably when the PR department noticed it. I'm sure there was a bunch of discussion at NASA about political impact and whether it was worth preflighting a second medium or sending McClean back out in a large.
To their immense credit, NASA and the astronauts made the right decision. Now that they've experienced the wrath of the twit-verse, they may not the next time.
That's a very common human bias. You're much safer with a computer at the controls. You're *safest* with a computer at the controls and a well-trained human ready to override if necessary. A major part of that training is knowing when not to do so.
I use all three regularly. OS X to Linux is fairly trivial. The GUIs are pretty much all equivalent, although Microsoft likes to screw with it every once in a while and then relent and put it back.
If you're doing any kind of scientific computing, Linux or OS X are both fine, Windows is a big disadvantage. Lots of projects aren't really supported on Windows. If you play games, the opposite is true.
Very true. I'm not sure where the tens of thousands of GPU hours figure comes from either. Google may have trained for that long because they've got the hardware. You can train a decent ImageNet clone on a commodity GPU on your own computer in a day or so.
Yes... they tested with one task on a single dataset. For a technique that (as far as I can gather from the terrible article) evaluates and discards options, that has a high risk of working well only in particular circumstances.
That would be fine, if JavaScript were *used* as a scripting language. Back in the old days, when you just wanted to make a button on your web page change colours or something, cool.
Problem is, Javascript today is used as a programming language, to write real (sizeable) programs that incorporate a decent amount of their own logic.
Slashdot Mirror
Curves are easy to quantify, just count the number of zero second derivatives, or maybe give the spherical harmonic spectrum.
Rolls are curves after all.
Of course they say that. It builds up their reputation in the press. See the beginning of my second paragraph: "stop buying the media hype."
If you've only got a few bombs and you actually want to nuke the US, missiles are probably the worst way to do it.
Genocide is intentional action to destroy a people. I suppose the Russians might have *intended* to invade and destroy a people but got bored after the first couple hundred.
Try starting here: https://www.sciencedirect.com/...
Energy return on energy invested is the basic idea: how many barrels of oil do you get if you invest one barrel of oil in production? Shale oil and tar sands are quite a bit below the world average, which means middle east conventional oil must be considerably above that average.
Now, this is my speculation, but a bit of math with fuel prices would probably support it. Oil is a famously inelastic resource, meaning it doesn't really obey the law of supply and demand; demand is pretty much fixed, because transportation isn't something we can easily use less of. So the Saudi's are printing money, but shale oil and tar sands do okay too, even with their crappy EROI, because the price of oil is inflated due to its necessity for our transportation infrastructure. Because of the price of oil is inflated relative to its intrinsic energy value, almost nobody burns it to make electricity.
By the way, the conclusion of that paper is basically that we're screwed unless we improve our energy production technology. Economic prosperity is closely related to EROI; if you have low EROI you spend all your resources trying to produce energy, not much left over for other things. Fossil fuel EROI is all decreasing towards the danger zone. Hydro power has good EROI, but is mostly exploited. Nukes would work, at least for a while. Wind and solar are okay, at least in ideal locations, but need to improve in order to maintain our current prosperity.
Yes. In ways they couldn't imagine.
Also, the much maligned aether, a substance that permeates all of space and (a bit later, the medium through which light propagates), is basically how you'd describe quantum fields (from the standard model) and space-time (from relativity).
Ah, "genocide" has been watered down to mean a few hundred casualties. Awesome, another word rendered useless.
North Korea isn't building nukes to hit the US. They're building nukes to hit American expeditionary forces attacking North Korea. Maybe with a few left over to lob at Japan and South Korea.
Stop buying the media hype. The North Korean leadership isn't crazy. Ever since Russia built one, nukes are for self defence through deterrence. North Korea has demonstrated the ability to build them, removing invasion from the options the US can consider.
Yes. That's what I said. The best missile defence is one that doesn't work. If the US thinks their missile defence works, they might do something stupid. If they know it doesn't, less chance of stupid. If the Russians or Chinese or whoever don't think it works, MAD is maintained as usual. If they think it does work, for whatever reason, they're less likely to launch, not more.
Good. Missile defences that others think might work are much better for stability than ones that host countries know will work. The defender knows they're vulnerable, so they don't do anything stupid. Potential attackers can't be sure an attack would succeed, so they don't do anything stupid either.
This is a good demonstration of something few people realize: oil, outside of the conventional reserves in the middle east and a few other places, isn't really a viable energy source. American shale oil extraction is energy positive, but not enough to be really worthwhile. Oil's value is mostly as an energy storage mechanism to power transportation.
So it doesn't make sense to burn that oil to power your extraction operations. It does make sense to build a solar facility to power extraction.
In quantum mechanics a vacuum is defined as the lowest possible energy state. The uncertainty principle prohibits zero energy states, so a vacuum always has some base energy, which can be thought of as a sea of virtual particles popping into and out of existence.
Space that includes a strong magnetic field and particles zipping through it isn't really a vacuum, but I think the point of the calculation is that the virtual particles in an actual vacuum can be polarized a little bit by a strong magnetic field and will thus act as a medium with some small but non-unity index of refraction. That will slightly lower the speed of light, causing very high energy charged particles to emit Cherenkov radiation.
That's not entirely true. There's error in our measurement, yes, but there's also error in the theoretical prediction.
Most modern theories require some approximation. QED, for example, posits that any particle interaction will be affected by an essentially infinite sum of contributions. Since these contributions become smaller and smaller, you can add up the first few and get a pretty good prediction... but not perfect.
Still, instead of "most precise theory" a better description might be "most precisely validated theory."
Precision is important because higher energy phenomena can manifest as small anomalies in lower energy interactions. For example, an alternative to building a bigger collider, able to directly generate high energy conditions, is to build specialty colliders and detectors that can more precisely measure lower-energy interactions. This is what the g2 experiment at Fermilab is doing: precision measurements of the magnetic moment of the muon disagree with theoretical predictions, possibly indicating some higher-energy new physics.
Arxiv, which was mentioned by the GP, is funded by public grants and publishes papers for around $6 per, including the cost of developing and maintaining the platform.
The two big AI journals publish their budgets, and they both publish papers for a couple bucks per, half of which is to register a DOI.
The problem with all the open publishing initiatives is that they're aimed directly at the subscription model. That's not the problem. The problem is that scientific publishing, no matter how you pay for it, is currently ridiculously expensive.
Easy enough to get around that. Make it a loan, with an interest rate and payback schedule calculated to yield 15% of the maximum expected salary over two years. Then forgive any overrun if the salary is less than the max.
You'd have to be completely nuts to sign such a thing, but people are stupid.
Unfortunately it doesnâ(TM)t seem to be possible to do that anymore. The PR people tell you where to go, how to stand, and you better smile.
Kind of. NASA has said they didn't plan an all-female spacewalk, it just happened in the schedule. It then got hyped, probably when the PR department noticed it. I'm sure there was a bunch of discussion at NASA about political impact and whether it was worth preflighting a second medium or sending McClean back out in a large.
To their immense credit, NASA and the astronauts made the right decision. Now that they've experienced the wrath of the twit-verse, they may not the next time.
Astronauts change size in orbit. McClain has mentioned that she's grown by about two inches.
NASA has a *very* detailed fitting procedure for spacesuits, but final fit choice is made by the astronaut herself.
That's a very common human bias. You're much safer with a computer at the controls. You're *safest* with a computer at the controls and a well-trained human ready to override if necessary. A major part of that training is knowing when not to do so.
I use all three regularly. OS X to Linux is fairly trivial. The GUIs are pretty much all equivalent, although Microsoft likes to screw with it every once in a while and then relent and put it back.
If you're doing any kind of scientific computing, Linux or OS X are both fine, Windows is a big disadvantage. Lots of projects aren't really supported on Windows. If you play games, the opposite is true.
Classic OS had a graphing calculator too, since 1994.
Grapher really is nice though. Unfortunately it's a bit hidden, so not a lot of people know it exists.
Very true. I'm not sure where the tens of thousands of GPU hours figure comes from either. Google may have trained for that long because they've got the hardware. You can train a decent ImageNet clone on a commodity GPU on your own computer in a day or so.
Yes... they tested with one task on a single dataset. For a technique that (as far as I can gather from the terrible article) evaluates and discards options, that has a high risk of working well only in particular circumstances.
That would be fine, if JavaScript were *used* as a scripting language. Back in the old days, when you just wanted to make a button on your web page change colours or something, cool.
Problem is, Javascript today is used as a programming language, to write real (sizeable) programs that incorporate a decent amount of their own logic.
I thought they called that C++. Every C++ program I've ever seen looks like someone started by specifying their own language.
Code barista. I like it.