I've always thought that Tufte isoverblown. I mean, sparklines - duh, its a small graph without axes. The famous "Napeleon's army" chart is really not any more enlightening or informative than just reading a paragraph or two and looking at some tables of numbers.
I'm ok with that. I'm interested in electing a president based on his policies, not his personal opinions. I'd rather have a constitutionalist racist, bigot, and homophobe than somebody who tries to legislate tolerance. How do you know his policies won't be influenced by his personal opinions? Given that he claims his newsletters were written without him knowing what was in them, how do you know he won't appoint people who push a racist, bigoted political agenda and claim he "didn't know" what they were doing?
And who exactly is this tolerance-legislating straw man you put up? As if somehow we have only two choices (1) Crazy Racist Ron Paul, or (2) unnamed candidate who "legislates tolerance".
You can get Barcelona systems from siliconmechanics. The big guys don't have the systems yet. Thanks for the tip! Just over $13K for a 64GB system with 16 cores...not bad.
Outside of giant clusters, is anybody running Barcelona yet either? I have been unable to find any systems available for purchase. Word on the street is January before they are available in quantity to the general public.
>So when we need to see something far away, can anyone explain why we need to make the actual lenses bigger?
larger apertures collect more light, so you can see fainter objects.
an alternate method to get more light (increase signal to noise) is to observe for a long time and 'average' the images, but this method smears out any features that change quicker than your observation time. Not quite right, there are actually two issues: one is the faintness of the object, for which both larger aperture and longer exposure time can help. The other is resolution, for which only larger aperture can help. The theoretical resolving power of a telescope of a given aperture is limited by optical diffraction. There is no getting around that limit, to get more resolution you have to build a bigger telescope.
That answer then raises the question: Can't we make more sensitive equipment instead of increasing the lens size? Generally, the more sensitive the imaging equipment, the more noise creeps into the picture. That's why cameras have to use a flash in really low light instead of just using a more sensitive CCD - it gets to a point where the noise in image overwhelms the signal if you just increase sensitivity. And the longer you expose the image, the worse the noise gets.
As for why the noise is there, you'll have to study thermodynamics a bit to understand that as far as I know.
We don't necessarily have to understand human intelligence in order to duplicate it. Given a fast enough computer one could simulate all the individual cells in the human brain and body with high fidelity, thereby creating a "virtual" human who has human intelligence (and flaws, too).
I can build a dual socket quad-core w/ 1GB ram and a 250GB hd for a lot less than "$3747". Oh wait, I already have. I guess you consider your time to be worth $0 per hour.
We cannot directly compare Apple hardware with Sony hardware, but we can in fact compare the things they re-sell, such as RAM or disks.
And Apple markup tends to be a lot higher on these compared with HP or Sony. Whether it's due to the "milk the fanboys" attitude or all the rigoros testing, I will not say. Actually, you can directly compare Apple hardware with HP, Sony, or Dell hardware since they use many of the same components. For example, a Mac Pro can be directly compared to a number of different Dell or HP servers; the CPU, motherboard, memory, etc. are nearly identical. The Mac Pro cost compares quite favorably. For example, a Dell Poweredge 1900 dual socket quad core Xeon 5365 machine w/ 1 GB ram and 250 GB disk costs $3747 with no OS. The same CPUs, disk, and RAM from apple is $3997 but includes the OS, iLife, etc.
The Dell will sound and feel like an industrial-strength hairdryer. The Apple machine is basically silent.
You do know that upgrading your own memory without paying an Apple-certified technician will void your warranty, right? This one probably qualifies as an urban legend by now.
Apple "recommends" you use their memory, but you can use any memory you want and install it yourself without voiding the warranty; see the standard Apple hardware warranty http://images.apple.com/legal/warranty/docs/cpuwarranty.pdf. Same applies to hard drives, video cards, etc.
The only exclusion is if you damage the machine while installing hardware, or if the stuff you are installing damages the machine somehow.
By comparing the listed prices of upgrades (RAM, hard-drive, bluetooth) and addons at Apple online store vs HP or Sony.
In fact, my comment should properly read "at least 30%" as they tend to quote double the price in many cases (such as $700 vs $350 for 4gb of notebook ram) I hate that I'm even replying to this, but that's just stupid. You can buy ram, hard drives, and bluetooth upgrades for your Mac from many other vendors. I personally wouldn't buy them from HP or Sony either.
Re:I'm surprised that no-one's mentioned Gorillas
on
DOS 5 Upgrade Video
·
· Score: 1
I had an Amiga 1000 back in the late 80's/early 90's... I wanted to learn C, but I couldn't get a C compiler without paying $$$! So I hand-wrote programs on paper, without compiling, in preparation for the day when I would actually have access to a compiler.
The 8-core Mac Pro is about 3 times faster than the "supercomputer" mentioned in the article, for a little more than twice the price (I have an 8 core 16 GB Mac Pro too, used for scientific computing).
A typical 2.33 Ghz, 8 core clovertown machine can do about 50 Gflops on Linpack. You can get one for under $4K with 8 GB of ram. No assembly required, and you don't have to use MPI to parallelize since the code runs on just one machine, not four.
If you look at the wikipedia article on Mobius strips (linked by the poster, actually), they show a fairly simple example of algebraic equations which define a Mobius strip. The article in question here is not about that - its about the physics of forming Mobius strips from other shapes. No one seems to get that.
algebraic equations for Mobius strips are not new
on
Möbius Strip Riddle Solved
·
· Score: 5, Interesting
TFA doesn't say what the poster says it does. The article is really about the physics of actually making Mobius strips out of various materials. The equations which parameterize a mobius strip are not complicated and can take many forms (a good math undergrad should be able to put it together with some help from Mathematica, for example).
Who said the tanks have to have eyes? They could link up with fleets of little helicopters that provide the visuals for them. Much better because they could actually see around corners then too.
What you are missing is the human factor. Predicting baseball is more like predicting the economy than predicting the weather. The latter is difficult only because of the sheer amount of variables involved (but we do understand the underlying principles), while the former is ultimately attempting to predict human behavior (we have no reliable scientific methods for doing that).
no you are wrong, they've had dual quad core systems for quite some time . I visit their page regularly since I buy dell's regularly.
The key is 3 Ghz. Try to find a Dell with two Intel quad core chips running at 3 Ghz. You won't find it. The fastest they offer (as of this morning) is the E3555, which runs at 2.66 Ghz.
Intel doesn't even mention that such a chip is available yet - presumably it will be called the E3565.
Dell has had these in their precision workstations for quite some time.
They've had quad-core Xeons (aka clovertowns), but not at 3 Ghz. As far as I can tell Apple is the first to offer 3 Ghz Clovertown chips in their machines.
I love how these press releases from universities pretend as if no one had ever even contemplated such an idea before the ground-breaking work mentioned in the press release. People have been doing mathematical modeling of tumors for years. Avner Friedman has been doing work since the late 1990's at least (see http://www.math.ohio-state.edu/node/22077, for example).
I am a U.S. native, product of public secondary education. I have always been very, very good at math, have a Ph.D. in Applied Mathematics, and now work as a professional mathematician. As you can probably already guess, I also have always been fairly confident about my mathematics ability (though I certainly don't think I'm better at math than other mathematicians, just better than most people). I also seem to get along well with most people, enjoy social activities, am married to a beautiful woman, and I am tall, blond haired, blue-eyed, and athletic.
Interestingly, most of the mathematicians/scientists I work with are not US natives, but none of them are "geeky looking" or socially awkward. If you want to achieve success professionally in ANY field, you usually have to be good with people and a good writer/communicator, otherwise no one understands your work, no one likes you and you don't get ahead (unless you are a super-freak-genius, in which case nothing else matters).
P.S. The problem with US mathematics education is (a) low expectations placed on students, and (b) low expectations placed on teachers.
Slashdot Mirror
I've always thought that Tufte isoverblown. I mean, sparklines - duh, its a small graph without axes. The famous "Napeleon's army" chart is really not any more enlightening or informative than just reading a paragraph or two and looking at some tables of numbers.
And who exactly is this tolerance-legislating straw man you put up? As if somehow we have only two choices (1) Crazy Racist Ron Paul, or (2) unnamed candidate who "legislates tolerance".
Ron Paul is a racist, bigot, and homophobe. Have you seen this?
http://www.tnr.com/politics/story.html?id=e2f15397-a3c7-4720-ac15-4532a7da84ca
Outside of giant clusters, is anybody running Barcelona yet either? I have been unable to find any systems available for purchase. Word on the street is January before they are available in quantity to the general public.
larger apertures collect more light, so you can see fainter objects.
an alternate method to get more light (increase signal to noise) is to observe for a long time and 'average' the images, but this method smears out any features that change quicker than your observation time. Not quite right, there are actually two issues: one is the faintness of the object, for which both larger aperture and longer exposure time can help. The other is resolution, for which only larger aperture can help. The theoretical resolving power of a telescope of a given aperture is limited by optical diffraction. There is no getting around that limit, to get more resolution you have to build a bigger telescope.
That answer then raises the question: Can't we make more sensitive equipment instead of increasing the lens size? Generally, the more sensitive the imaging equipment, the more noise creeps into the picture. That's why cameras have to use a flash in really low light instead of just using a more sensitive CCD - it gets to a point where the noise in image overwhelms the signal if you just increase sensitivity. And the longer you expose the image, the worse the noise gets.
As for why the noise is there, you'll have to study thermodynamics a bit to understand that as far as I know.
We don't necessarily have to understand human intelligence in order to duplicate it. Given a fast enough computer one could simulate all the individual cells in the human brain and body with high fidelity, thereby creating a "virtual" human who has human intelligence (and flaws, too).
And Apple markup tends to be a lot higher on these compared with HP or Sony. Whether it's due to the "milk the fanboys" attitude or all the rigoros testing, I will not say. Actually, you can directly compare Apple hardware with HP, Sony, or Dell hardware since they use many of the same components. For example, a Mac Pro can be directly compared to a number of different Dell or HP servers; the CPU, motherboard, memory, etc. are nearly identical. The Mac Pro cost compares quite favorably. For example, a Dell Poweredge 1900 dual socket quad core Xeon 5365 machine w/ 1 GB ram and 250 GB disk costs $3747 with no OS. The same CPUs, disk, and RAM from apple is $3997 but includes the OS, iLife, etc.
The Dell will sound and feel like an industrial-strength hairdryer. The Apple machine is basically silent.
Apple "recommends" you use their memory, but you can use any memory you want and install it yourself without voiding the warranty; see the standard Apple hardware warranty http://images.apple.com/legal/warranty/docs/cpuwarranty.pdf. Same applies to hard drives, video cards, etc.
The only exclusion is if you damage the machine while installing hardware, or if the stuff you are installing damages the machine somehow.
In fact, my comment should properly read "at least 30%" as they tend to quote double the price in many cases (such as $700 vs $350 for 4gb of notebook ram) I hate that I'm even replying to this, but that's just stupid. You can buy ram, hard drives, and bluetooth upgrades for your Mac from many other vendors. I personally wouldn't buy them from HP or Sony either.
I had an Amiga 1000 back in the late 80's/early 90's... I wanted to learn C, but I couldn't get a C compiler without paying $$$! So I hand-wrote programs on paper, without compiling, in preparation for the day when I would actually have access to a compiler.
The 8-core Mac Pro is about 3 times faster than the "supercomputer" mentioned in the article, for a little more than twice the price (I have an 8 core 16 GB Mac Pro too, used for scientific computing).
A typical 2.33 Ghz, 8 core clovertown machine can do about 50 Gflops on Linpack. You can get one for under $4K with 8 GB of ram. No assembly required, and you don't have to use MPI to parallelize since the code runs on just one machine, not four.
If you look at the wikipedia article on Mobius strips (linked by the poster, actually), they show a fairly simple example of algebraic equations which define a Mobius strip. The article in question here is not about that - its about the physics of forming Mobius strips from other shapes. No one seems to get that.
TFA doesn't say what the poster says it does. The article is really about the physics of actually making Mobius strips out of various materials. The equations which parameterize a mobius strip are not complicated and can take many forms (a good math undergrad should be able to put it together with some help from Mathematica, for example).
Who said the tanks have to have eyes? They could link up with fleets of little helicopters that provide the visuals for them. Much better because they could actually see around corners then too.
What you are missing is the human factor. Predicting baseball is more like predicting the economy than predicting the weather. The latter is difficult only because of the sheer amount of variables involved (but we do understand the underlying principles), while the former is ultimately attempting to predict human behavior (we have no reliable scientific methods for doing that).
That should read 'E5355' as the fastest one at Dell, ant 'E5365' as the probably part number for the 3 Ghz clovertown.
The key is 3 Ghz. Try to find a Dell with two Intel quad core chips running at 3 Ghz. You won't find it. The fastest they offer (as of this morning) is the E3555, which runs at 2.66 Ghz.
Intel doesn't even mention that such a chip is available yet - presumably it will be called the E3565.
They've had quad-core Xeons (aka clovertowns), but not at 3 Ghz. As far as I can tell Apple is the first to offer 3 Ghz Clovertown chips in their machines.
...you could hit anyone, anywhere, on short notice.
I love how these press releases from universities pretend as if no one had ever even contemplated such an idea before the ground-breaking work mentioned in the press release. People have been doing mathematical modeling of tumors for years. Avner Friedman has been doing work since the late 1990's at least (see http://www.math.ohio-state.edu/node/22077, for example).
I am a U.S. native, product of public secondary education. I have always been very, very good at math, have a Ph.D. in Applied Mathematics, and now work as a professional mathematician. As you can probably already guess, I also have always been fairly confident about my mathematics ability (though I certainly don't think I'm better at math than other mathematicians, just better than most people). I also seem to get along well with most people, enjoy social activities, am married to a beautiful woman, and I am tall, blond haired, blue-eyed, and athletic.
Interestingly, most of the mathematicians/scientists I work with are not US natives, but none of them are "geeky looking" or socially awkward. If you want to achieve success professionally in ANY field, you usually have to be good with people and a good writer/communicator, otherwise no one understands your work, no one likes you and you don't get ahead (unless you are a super-freak-genius, in which case nothing else matters).
P.S. The problem with US mathematics education is (a) low expectations placed on students, and (b) low expectations placed on teachers.