However small Linux sales were, if they stop Linux sales, they will lose business. If they made a profit on Linux sales, they even lose profit by cutting their Linux offer. They will also lose investments they made.
No, this is not necessarily the case. Building Linux laptops has an opportunity cost. If the opportunity cost of building one Linux laptops is building two Window laptops, then building Linux laptops might be causing them lost profit from the Windows laptops they could have been building in that time.
There's a dated poster that hangs in a fairly well traveled hallway in the CoRE Building at Rutgers where Teoma was developed. It is an enlarged copy of an article theorizing that Teoma would provide significant competition to Google. It's somewhat funny to look at since it was written in the forgotten era of Google being popular, but not dominant. In fact, here's the article. My favorite quote: "Google has reached its maturity." Maybe this is Teoma's second chance at attacking Google.
as is the case when citizens are so poorly educated as to be unable to parse a word like "nuclear" properly
Did you even bother to read the article? The article cites one possible reason for the mispronunciation as a behavioral problem rather than an educational one. I've found that certain words I learned younger when I was a child were learned incorrectly and I pronounced them wrong. Maybe it was because I was reading beyond my level and had no reinforcement from others as to how they should be pronounced; maybe I'm mildly dyslexic. I find it a stretch to liken the mispronunciation of one single word to the use of ebonics. When you define everything in language as either correct or incorrect, you're left saying that one of American English and British English are wrong. The words vary, but we don't scream that they devolved.
When education trends towards a "get them through the system" mode rather than a "fail them if they cannot perform" mode.
Why you expect the educational system to properly educate children when all the money is stripped away from education and given to deadbeats and welfare recipients is beyond me. Furthermore, why you expect the government to do any of these things properly is a rather interesting concept as well.
In fact, Bush's usage is so common that it appears in at least one dictionary. Merriam-Webster's, by far the most liberal dictionary, includes the pronunciation, though with a note identifying it as "a pronunciation variant that occurs in educated speech but that is considered by some to be questionable or unacceptable." A 1961 Merriam-Webster's edition was the first to include "nucular"; the editors received so many indignant letters that they added a usage note in the 1983 version, pointing out its "widespread use among educated speakers including scientists, lawyers, professors, congressmen, U.S. cabinet members, and at least one U.S. president and one vice president." They even noted its prominence among "British and Canadian speakers." taken from http://www.slate.com/id/2071155/.
First, we should probably clarify something. The output of the FFT is the output of a DFT (Discrete Fourier Transform) for all intents and purposes. The FFT is a fast way of computing those values by using the fact that when you actually compute a DFT, you're recalculating a lot of the same stuff over and over due to the fact that the Fourier matrix is a circulant matrix. Now that that's out of the way, it should be said that the DCT is a specific case of the DFT when the signal is real and even. See my explanation for that here. No, you probably wouldn't have any advantage using a generic DFT instead of a DCT. First of all, you don't care about complex signals for audio and images and second of all, I forget my second point;-)
A Fourier transform can be used on any finite-energy signal (if you integrate over the whole signal from -infinity to infinity in time, the value is not infinite).
which basically means all real-life signals because you can make any real-life signal have compact support, I suppose.
Yes, a DCT is a DFT. It's the equivalent in a Fourier series to when you take a function that's periodic on [0, Pi], and then mirror the function across the y-axis to make it even and periodic on [-Pi, Pi] and then take the Fourier series of it. Since the function you just constructed is even, all the sin(nx) coefficients in the Fourier series drop out and you're left with cosines.
Even though I was taught the DFT, I would not have understand the explanation without first the introduction to Fourier series. The basic idea is that if you have an inner product space, or what we can simply call a bunch of functions that we can compute inner products with... take some f(x) and take its inner product with every sin(nx) and cos(nx). Normalize these values and use them as coefficients for each sin(nx) and cos(nx). The neat thing is that if my f(x) = sin(x) and I take its inner product with every sin and cos, the only one that has a non zero is sin(1x), the function itself. So the Fourier series of sin(x) is sin(x). Basicaly, you're projecting f(x) onto a bunch of trig polynomials. It works because the set of sin(nx) and cos(nx) are orthogonal with the exception of each one not be orthogonal to itself. You can do the same with any orthonormal basis.
Next, look up the Number Theoretic Transform. It's a Fourier transform but instead of using the field of complex numbers, you're using integers, so its significance to someone who's studied crypto would probably be high.
In the TOP500, it looks like ethernet is not yet an "outsider." Perhaps in the "top 100."
It depends on what you're doing. If your job is highly parallel, Ethernet is fine. But what happens when every CPU needs access to every other CPUs results in "real time?" Well, low latency is then a must. 1 ms latency is potentially millions of wasted cycles.
Hah, thanks, but I can't get too excited about that. If I did, it would be acknowledging that there was something novel about my comment. But there isn't. Admitedly, when I first saw Fourier series, I thought it was the most interesting math I had yet encountered. Now, that was in no small part due to this guy and his excellent teaching and explanation of Fourier Series. This is really no different than when it was proved by Fourier that any function (for some reasonable definition of any) could be represented as a sum of trig polynomials. It was a *cool* idea. Once you learn that, the DFT is cool, but not as earth shaking. To me, second only to Fourier series in terms of the "wow" factor when learning were wavelets and the associated Discrete Wavelet Transform and Filterbanks. But, one fact remains. Once you learn the DFT or the DCT or the mDCT or the DCT-II or furthermore the DWT, the first most *obvious* question is "Hey, what happens when we start removing frequency components?" And that's what all the lossy algorithms amount to; take a transform, set entries to zero, do some entropy encoding. How this is patentable is beyond me. In fact, this compression scheme is built in to learning Fourier series. Even though the sum in a Fourier series is to infinity, you obviously can't calculate or graph that with infinite precision. But you quickly realize that you only need a few terms. Aside from Gibbs' effect, a few Fourier terms represent a function really well.
You are missing the difference between Initial Key Exchange and Shared Secret
No, I don't think so. You say the article mentions a shared secret. It only mentions it in the context of using Diffie-Hellman to generate it. Once more, the shared secret is a secret generated by the Diffie-Hellman algorithm.
Shared Secret encryption under the right circumstances will be relatively impervious to MitM attacks, since the attacker would need to know both the protocol and the key to start attacking the encryption (which would, if done right, change on a "pseudo random" basis).
You're not hiding the protocol. All cryptanalysis is based on the fact that the attacker knows every detail of the algorithm(s) involved.
The "initial key exchange" is performed before any communications are attempted (at setup time). The article specifically mentions "shared secret"
How do you think you establish a shared secret? Here's how Diffie Hellman works in short.
Consider some generator g for some group in Z* mod p, where p is prime and g and p are public knowledge. Bob calculates g^a (mod p), where a is a random integer, sends it to Alice. Alice calculates g^b (mod p), where b is a random integer, sends it to Bob. Bob calculates (g^b)^a (mod p) and Alice calculates (g^a)^b (mod p). These two are the same, and g^(ab) (mod p) is the shared secret!
This is secure because calculating g^(ab) (mod p) is difficult knowing g^a (mod p) and g^b (mod p) only. This is an instantiation of the discrete logarithm problem.
Now if there were just a handful of these cell phones being used, the NSA could (probably) handle that and decrypt them.
It's unlikely they could. Assuming the key exchange works properly, and assuming they're using a known good algorithm (such as Rjindael aka AES), the NSA has no shot. Assume they use AES. Default is 128 bits and 10 rounds. Then the following little blurb from Apple's website applies: AES gives you 3.4 x 10^38 possible 128-bit keys. In comparison, the Digital Encryption Standard (DES) keys are a mere 56 bits long, which means there are approximately 7.2 x 10^16 possible DES keys. Thus, there are on the order of 10^21 times more possible AES 128-bit keys than DES 56-bit keys. Assuming that one could build a machine that could recover a DES key in a second, it would take that machine approximately 149 trillion years to crack a 128-bit AES key.
(To put that into perspective, the universe is believed to be less than 20 billion years old.)
Now, that assumes you can crack a DES key in a second. The fastest successful crack by Deep Crack was just shy of 24 hours, or, 86400 seconds.
The top-level ISPs charge a lot for moving traffic across the country.
Actually, no they don't. For two reasons, mainly. One, big ISPs enter into BGP peering agreements in which no money changes hands. This is the whole "moving traffic across the country." That reason is somewhat circumstantial, however, as it depends on who your ISP is and whom they cater to. Two, bandwidth is much cheaper in bulk. The cost of upstream bandwidth is simply less than the cost of providing bandwidth to users. (The difference of the two is a key to profit.)
The 10 line function and the Haiku are simpler. Also, more elegant and better.
I disagree with the idea the cramming the most amount of logic into the least amount of lines is elegant. Brian Kernighan said it best, "Debugging is twice as hard as writing the code in the first place. Therefore, if you write the code as cleverly as possible, you are, by definition, not smart enough to debug it."
Engineers represent the best analogy to IT. There are Chem Engineers, Mechanical, Civil, etc. A multitude of very technical and highly trained professionals.
Yes, because most IT workers are both highly trained and professional. Please. Professional Engineers actually had to endure hard curriculum and difficult testing over the course of their academic career in addition to being able to prove themselves to other engineers and get recommendations from them. That's the opposite of most "IT" workers who have but a high school education and some training from DeVry or some MCSE course. Comparing IT workers on the whole to engineers is really doing a disservice to anyone who is an engineer or studying to be one.
E3 creates a complete media blackout for anything else.
You're forgetting this is Apple. Anything at E3 including Nintendo is far far smaller than an announcement from Apple. There's plenty of people interested in Apple's announcement that don't know what E3 is, let alone WHEN it is. I, for instance, had no clue it's coming up soon and frankly I couldn't care less.
There's a problem. If two people get the same song from iTunes, the songs sound the same, but the files are not identical. All the downloads are watermarked and encrypted using a set of keys unique to a user and on top of that, the purchaser's user id is inside the file. Every person needs to get a unique file from Apple. While this presents a very interesting secret sharing problem, it is by no means easy, and not as simple as including bit torrent.
chill dude, as an engineer i know exactly what you're talking about but there's no need to get so touchy about it...:)
It's not really about being touchy. It's just annoying that articles aren't fact checked before being linked to. I really should just chill since slashdot is what it always has been and will never change, yet I still post. Joke's on me;-)
Slashdot Mirror
However small Linux sales were, if they stop Linux sales, they will lose business. If they made a profit on Linux sales, they even lose profit by cutting their Linux offer. They will also lose investments they made.
No, this is not necessarily the case. Building Linux laptops has an opportunity cost. If the opportunity cost of building one Linux laptops is building two Window laptops, then building Linux laptops might be causing them lost profit from the Windows laptops they could have been building in that time.
There's a dated poster that hangs in a fairly well traveled hallway in the CoRE Building at Rutgers where Teoma was developed. It is an enlarged copy of an article theorizing that Teoma would provide significant competition to Google. It's somewhat funny to look at since it was written in the forgotten era of Google being popular, but not dominant. In fact, here's the article. My favorite quote: "Google has reached its maturity." Maybe this is Teoma's second chance at attacking Google.
When you define everything in language as either correct or incorrect, you're left saying that one of American English and British English are wrong.
should be
When you define everything in language as either correct or incorrect, you're left saying that one of American English and British English is wrong.
as is the case when citizens are so poorly educated as to be unable to parse a word like "nuclear" properly
Did you even bother to read the article? The article cites one possible reason for the mispronunciation as a behavioral problem rather than an educational one. I've found that certain words I learned younger when I was a child were learned incorrectly and I pronounced them wrong. Maybe it was because I was reading beyond my level and had no reinforcement from others as to how they should be pronounced; maybe I'm mildly dyslexic. I find it a stretch to liken the mispronunciation of one single word to the use of ebonics. When you define everything in language as either correct or incorrect, you're left saying that one of American English and British English are wrong. The words vary, but we don't scream that they devolved.
When education trends towards a "get them through the system" mode rather than a "fail them if they cannot perform" mode.
Why you expect the educational system to properly educate children when all the money is stripped away from education and given to deadbeats and welfare recipients is beyond me. Furthermore, why you expect the government to do any of these things properly is a rather interesting concept as well.
In fact, Bush's usage is so common that it appears in at least one dictionary. Merriam-Webster's, by far the most liberal dictionary, includes the pronunciation, though with a note identifying it as "a pronunciation variant that occurs in educated speech but that is considered by some to be questionable or unacceptable." A 1961 Merriam-Webster's edition was the first to include "nucular"; the editors received so many indignant letters that they added a usage note in the 1983 version, pointing out its "widespread use among educated speakers including scientists, lawyers, professors, congressmen, U.S. cabinet members, and at least one U.S. president and one vice president." They even noted its prominence among "British and Canadian speakers." taken from http://www.slate.com/id/2071155/.
Language evolves. Deal with it.
First, we should probably clarify something. The output of the FFT is the output of a DFT (Discrete Fourier Transform) for all intents and purposes. The FFT is a fast way of computing those values by using the fact that when you actually compute a DFT, you're recalculating a lot of the same stuff over and over due to the fact that the Fourier matrix is a circulant matrix. Now that that's out of the way, it should be said that the DCT is a specific case of the DFT when the signal is real and even. See my explanation for that here. No, you probably wouldn't have any advantage using a generic DFT instead of a DCT. First of all, you don't care about complex signals for audio and images and second of all, I forget my second point ;-)
A Fourier transform can be used on any finite-energy signal (if you integrate over the whole signal from -infinity to infinity in time, the value is not infinite).
which basically means all real-life signals because you can make any real-life signal have compact support, I suppose.
Yes, a DCT is a DFT. It's the equivalent in a Fourier series to when you take a function that's periodic on [0, Pi], and then mirror the function across the y-axis to make it even and periodic on [-Pi, Pi] and then take the Fourier series of it. Since the function you just constructed is even, all the sin(nx) coefficients in the Fourier series drop out and you're left with cosines.
Even though I was taught the DFT, I would not have understand the explanation without first the introduction to Fourier series. The basic idea is that if you have an inner product space, or what we can simply call a bunch of functions that we can compute inner products with... take some f(x) and take its inner product with every sin(nx) and cos(nx). Normalize these values and use them as coefficients for each sin(nx) and cos(nx). The neat thing is that if my f(x) = sin(x) and I take its inner product with every sin and cos, the only one that has a non zero is sin(1x), the function itself. So the Fourier series of sin(x) is sin(x). Basicaly, you're projecting f(x) onto a bunch of trig polynomials. It works because the set of sin(nx) and cos(nx) are orthogonal with the exception of each one not be orthogonal to itself. You can do the same with any orthonormal basis.
Next, look up the Number Theoretic Transform. It's a Fourier transform but instead of using the field of complex numbers, you're using integers, so its significance to someone who's studied crypto would probably be high.
In the TOP500, it looks like ethernet is not yet an "outsider." Perhaps in the "top 100."
It depends on what you're doing. If your job is highly parallel, Ethernet is fine. But what happens when every CPU needs access to every other CPUs results in "real time?" Well, low latency is then a must. 1 ms latency is potentially millions of wasted cycles.
Hah, thanks, but I can't get too excited about that. If I did, it would be acknowledging that there was something novel about my comment. But there isn't. Admitedly, when I first saw Fourier series, I thought it was the most interesting math I had yet encountered. Now, that was in no small part due to this guy and his excellent teaching and explanation of Fourier Series. This is really no different than when it was proved by Fourier that any function (for some reasonable definition of any) could be represented as a sum of trig polynomials. It was a *cool* idea. Once you learn that, the DFT is cool, but not as earth shaking. To me, second only to Fourier series in terms of the "wow" factor when learning were wavelets and the associated Discrete Wavelet Transform and Filterbanks. But, one fact remains. Once you learn the DFT or the DCT or the mDCT or the DCT-II or furthermore the DWT, the first most *obvious* question is "Hey, what happens when we start removing frequency components?" And that's what all the lossy algorithms amount to; take a transform, set entries to zero, do some entropy encoding. How this is patentable is beyond me. In fact, this compression scheme is built in to learning Fourier series. Even though the sum in a Fourier series is to infinity, you obviously can't calculate or graph that with infinite precision. But you quickly realize that you only need a few terms. Aside from Gibbs' effect, a few Fourier terms represent a function really well.
It's so silly. The only one who should be able to file suit is Jean Baptiste Joseph Fourier. After all, the JPEG standard is a DCT is a DFT.
You are missing the difference between Initial Key Exchange and Shared Secret
No, I don't think so. You say the article mentions a shared secret. It only mentions it in the context of using Diffie-Hellman to generate it. Once more, the shared secret is a secret generated by the Diffie-Hellman algorithm.
Shared Secret encryption under the right circumstances will be relatively impervious to MitM attacks, since the attacker would need to know both the protocol and the key to start attacking the encryption (which would, if done right, change on a "pseudo random" basis).
You're not hiding the protocol. All cryptanalysis is based on the fact that the attacker knows every detail of the algorithm(s) involved.
The "initial key exchange" is performed before any communications are attempted (at setup time). The article specifically mentions "shared secret"
How do you think you establish a shared secret? Here's how Diffie Hellman works in short.
Consider some generator g for some group in Z* mod p, where p is prime and g and p are public knowledge.
Bob calculates g^a (mod p), where a is a random integer, sends it to Alice.
Alice calculates g^b (mod p), where b is a random integer, sends it to Bob.
Bob calculates (g^b)^a (mod p) and Alice calculates (g^a)^b (mod p).
These two are the same, and g^(ab) (mod p) is the shared secret!
This is secure because calculating g^(ab) (mod p) is difficult knowing g^a (mod p) and g^b (mod p) only. This is an instantiation of the discrete logarithm problem.
Now if there were just a handful of these cell phones being used, the NSA could (probably) handle that and decrypt them.
It's unlikely they could. Assuming the key exchange works properly, and assuming they're using a known good algorithm (such as Rjindael aka AES), the NSA has no shot. Assume they use AES. Default is 128 bits and 10 rounds. Then the following little blurb from Apple's website applies:
AES gives you 3.4 x 10^38 possible 128-bit keys. In comparison, the Digital Encryption Standard (DES) keys are a mere 56 bits long, which means there are approximately 7.2 x 10^16 possible DES keys. Thus, there are on the order of 10^21 times more possible AES 128-bit keys than DES 56-bit keys. Assuming that one could build a machine that could recover a DES key in a second, it would take that machine approximately 149 trillion years to crack a 128-bit AES key.
(To put that into perspective, the universe is believed to be less than 20 billion years old.)
Now, that assumes you can crack a DES key in a second. The fastest successful crack by Deep Crack was just shy of 24 hours, or, 86400 seconds.
...I say just make bits illegal and get it over with.
As Claude Shannon rolls in his grave...
Have you ever *tried* accessing the engine on a newer car?
Yes. It's really not difficult. Not much has changed. Does the plastic cover scare you?
The top-level ISPs charge a lot for moving traffic across the country.
Actually, no they don't. For two reasons, mainly. One, big ISPs enter into BGP peering agreements in which no money changes hands. This is the whole "moving traffic across the country." That reason is somewhat circumstantial, however, as it depends on who your ISP is and whom they cater to. Two, bandwidth is much cheaper in bulk. The cost of upstream bandwidth is simply less than the cost of providing bandwidth to users. (The difference of the two is a key to profit.)
The 10 line function and the Haiku are simpler. Also, more elegant and better.
I disagree with the idea the cramming the most amount of logic into the least amount of lines is elegant. Brian Kernighan said it best, "Debugging is twice as hard as writing the code in the first place. Therefore, if you write the code as cleverly as possible, you are, by definition, not smart enough to debug it."
Engineers represent the best analogy to IT. There are Chem Engineers, Mechanical, Civil, etc. A multitude of very technical and highly trained professionals.
Yes, because most IT workers are both highly trained and professional. Please. Professional Engineers actually had to endure hard curriculum and difficult testing over the course of their academic career in addition to being able to prove themselves to other engineers and get recommendations from them. That's the opposite of most "IT" workers who have but a high school education and some training from DeVry or some MCSE course. Comparing IT workers on the whole to engineers is really doing a disservice to anyone who is an engineer or studying to be one.
E3 creates a complete media blackout for anything else.
You're forgetting this is Apple. Anything at E3 including Nintendo is far far smaller than an announcement from Apple. There's plenty of people interested in Apple's announcement that don't know what E3 is, let alone WHEN it is. I, for instance, had no clue it's coming up soon and frankly I couldn't care less.
It's hard to tell, but perhaps he meant that it's not enough to have the code compile without errors. It needs to compile without warnings.
There's a problem. If two people get the same song from iTunes, the songs sound the same, but the files are not identical. All the downloads are watermarked and encrypted using a set of keys unique to a user and on top of that, the purchaser's user id is inside the file. Every person needs to get a unique file from Apple. While this presents a very interesting secret sharing problem, it is by no means easy, and not as simple as including bit torrent.
chill dude, as an engineer i know exactly what you're talking about but there's no need to get so touchy about it... :)
;-)
It's not really about being touchy. It's just annoying that articles aren't fact checked before being linked to. I really should just chill since slashdot is what it always has been and will never change, yet I still post. Joke's on me