I would suggest a modification to "linear"
on
SHA-1 Broken
·
· Score: 1
How did you do a linear search? Did you just try one number after another until you reached your target number?
I could be wrong, but I think the problem is that your linear search is not searching the space uniformly; it favors lower numbers (or higher numbers) more than the opposite, despite the fact that the numbers are chosen uniformly. Further, the probability that a random search will hit a low number is higher than the probability that the linear search will reach a high number for any sufficiently large set. In other words, the RNG has an advantage probabilistically.
Instead of a linear search, try subdividing the space each time, which guarantees a roughly uniform search of the space. So, for example, if you were guessing the range of 0-10, you'd guess in this order (I subdivided and rounded up or down, alternating): 5 0 10 2 7 1 4 9 3 8 6
There are other ways of doing uniform searches. Pick one and see if your results aren't better.
Scientists could customize the system to target any protein on the surface of a cell, Chen said. He and his colleagues have seen success with about a dozen different molecules, including brain and other blood cells, he said.
They tested it on melanoma. Nowhere in the article does it say that the method only works for melanoma. Also, they imply that it could work on other kinds of cancer.
Ah...how do you decide that something is truly random? The only thing I can think of that has randomness quite apart from observations is mathematically defined random variables.
We have determined that, for example, particle decay is random because it appears to be true. The theories of quantum mechanics are based upon this observation. If we found a particle for which this did not appear to happen for long stretches of time, we would probably revise our theories, not assume that it was a truly random event so we couldn't use it as evidence.
My example wasn't a metaphor. It was an example of what happens if you ignore evidence of a trend.
You have to remember to use their claim "Within close temporal proximity of major world events there will be a spike in the graph that corelates with the number and strength of emotional response of the people involved," not "all spikes coincide with world events." The kicker is that they don't have a real good way to test for "emotional response of people involved."
The response to images makes sense - people would learn a response. If they really want to show the something significant, they'd have to show that people can either anticipate a correct strong response all of the time (by also showing images that would invoke no response), or show that they would invoke a strong response the first time after a series of no-responses.
The other part doesn't jive, though. The theory that this group of devices predicts disasters does not preclude the idea that it also produces false positives - or even that it also picks up something else of significance that has not been identified.
Still, I question how they go about producing these random numbers. That could be the culprit.
Oh, and as far as the straight line thing, and the curve - they're obviously talking about aggregated data. Unlike the "law of averages" as applied to a single number, the probability of getting a large number of the same values over and over can be calculated. It is very important to remember that what has happened in the past should have some weight in predicting what will happen in the future.
What if all scientists took your approach to science?
"Oh look, the apple fell from the tree, and I think it fell at the same accelaration as the last object I saw fall. I wonder if all objects fall with the same acceleration? Too bad I can't learn anything from that, since what happens in the past has no bearing at all on the future. I'm gonna go get some pie."
It is quite easy for someone versed in probability to calculate (and I'm hoping they have) the likelihood of occurance of the anomalies they have witnessed. And if they've gotten a significant result (as in - this possibility that has occured 4 times this month should only happen on average once every hundred years...), then it might be worth looking into. Of course, maybe they're just fooling themselves, or being fooled by someone else. That's an awful lot of highly educated people to not realize that an anomoly is actually normal.
There's this interesting distinction in my areas (Computer Science and Computer Engineering). All the women in my graduate program were going into cognitive science or databases. None went into anything else.
Not that I blame them; I didn't go into any of those things either. It's strange, though, because those are the only two areas offered that don't have any hard math.
You have a very valid point. It just doesn't have anything to do with the actual article (and presumably nothing to do with the problem).
The article is stating that hate mongering groups are starting their own little clubs on google. They want to be flooded by inane and pointless racist remarks. Hasn't this mostly been the way of such groups? Exist in exclusive forms for a long time where the ugliness cannot be seen, and come out only when there is sufficient political power to destroy the object of obsession.
What you are describing is the convolution of a signal with a rect function (I invite you to go over to mathworld and look at the mathematical definition of convolution).
This isn't a very good approach. Even the existing methods will produce less artifacts.
It does give you a bit of insight, though: existing methods work well when you have sound that is predictable, just as yours does - that is the same pitch over the interval of an entire sample (where multiple samples make up a signal). They also tend to cause less distortion when they're doing less - so if you only change things a teeny bit, there won't be many artifacts.
As far as building back the 7 seconds, I think they mostly just start before the commercials end and have it all back at once.
...for some definitions of fairly easy
on
MythTV 0.17 Released
·
· Score: 4, Informative
If by "fairly easy" you mean "there are existing algorithms that do this badly," then you're right.
If you mean "sounds exactly the same, only faster" then you're wrong. Considering the quality of these things now, I'm not sure I wouldn't rather just let the pitch raise.
The problem is how to represent pitch. Most of the time, this is done by converting to a frequency domain and doing a shift, or by convolving the signal with a waveform that causes a signal shift (the classical example of this is using a sine wave, as is done for RF encoding). The problem is that this technique is only really good for a signal that doesn't change over time.
In fact, even the best pitch shifters assume that the pitch can be modeled as function of time and are unable to deal with randomly changing pitches very well. Lots of artifacts are still introduced when dealing with an "instrument" as complex as the human voice (on the other hand, they work great for flutes). Of course, if you don't change the pitch very much, you can get away with less artifacts.
"Support for x86, AMD64, PowerPC, UltraSparc, Alpha and MIPS processors"
Actually, there are 13 ports officially supported by Gentoo (and there are also a few side project ports which are still experimental).
Gentoo doesn't consider embedded processors part of the support base. I think the only one that isn't supported actually supported in any way in that list is VAX. It considers embedded processors a separate family of packages. Further, it covers processor families. The sparc port actually covers all Sparcs.
There are various levels of support for the 8000+ packages in Gentoo. Have a look here for the list which specifies the level of support for any given package.
VMware uses a processor emulation layer. In other words, its sandbox contains a virtual processor that interprets the output and converts it into native code. This means that at best you can get 1/3 of native speed with VMWare. Linux can run at close to native speed under Windows (using coLinux or others), under BSD (using userspace linux), or under MacOS (using userspace linux). The way this works is that there are hooks built into linux that basically make it act less like an operating system and more like an app.
VMWare can make anything run on anything else, but it isn't pretty. And unlike those things I mentioned, it's not free either.
Portage has been ported to BSD. Incidentally, there's also a cygwin and a MacOS port. I just call it a Gentoo port to NetBSD because that's what the people who made it call it.
Wow...it runs on almost as many things as Gentoo does.:)
Maybe that's why there's a Gentoo port to BSD.
Seriously, though, at this point, Linux can run under other operating systems without a CPU emulation layer, and it can do just as many strange processors. Not that I'm discounting NetBSD; there are plenty of reasons why NetBSD has an edge.
The thing you're missing (and the one that I'm sure is mentioned elsewhere on this thread, and that the author is hinting at) is that one of the main arguments for a lower Linux TCO than Windows TCO is that it avoids the Microsoft tax, thus saving you a lot up front to offset the possibly higher costs of learning the system.
I guess this isn't usually true in the corporate world.
Perhaps you should take a psychology course or two. Why do you suppose young men kill young women?
It's for pleasure. It's a sensual experience like rape. Remember the serial killer Ted Bundy? Do you know what he claimed set him off? I'll give you a huge hint: porn.
Seeing violence on TV has not been substantially linked to actual violence - especially not serial violence - but sexual awakening before one is ready can create all kinds of sexual deviance - including a homocidal fetish. This is much worse, and only one of the reasons why people should already have a good understanding of who they are before they do something like look at porn.
Lucky you. I graduated in 97. I took two of the three computer classes my school offered before giving up in disgust.
Class 1: Typing. Not on computers - on typewriters.
Class 2: Microsoft Works for dos FOR A YEAR (for all those who don't know, it's less complicated than office. It took all of a week to learn everything we covered in that course). The teacher optimized it for the XTs that we were using so that it would fit on our floppies. That's right - no harddrives on these machines.
Of course, there were computers with win 3.1 in the libraries, but the libraries weren't open after school, we weren't allowed to use disks on them, and teachers didn't schedule regular trips to the library for that purpose. They just sat there doing nothing.
I don't blame the football program, though - that pays for itself. I was in the band, though, and I know for a fact we had more than a few luxuries that could easily have been spent on twenty or so pentiums (for instance, the $2000 stereo system we got my junior year).
Nearest I can figure, they used this logic for giving the band so much money: 90% of AP and gifted students were in our band => a lot of extra federal money given to support these students (not needed in most cases) => spend it on what the AP and gifted students do in case it helps them
Teachers who become more afluent and wise don't become administrators. Politicians become adminstrators, and not because they care, usually - usually it's to get a better political job later. This is the problem, I think.
RTF and doc are just too complicated for them to be used by programs outside of Open Office and Microsoft products.
I would like a simpler format.
I would prefer to allow any program that is capable of printing a layout to export to some document format, and right now the only possibility is pdf and ps, both of which have no WYSIWG editors.
My almamater uses it (Purdue university). My guess is that it's a compromise to suit the business drones. They like closed products for some reason. I think it gives them a warm, fuzzy feeling.
Netscape will always be able to feed on Mozilla as long as such drones exist.
You can only reorder so much. If the order of execution depends upon data - for example, programs that accept data from the users, from a database, from chaotic systems, or from other programs, then you can't predict it at compile time. Of course, you may not believe this, but that doesn't make it any less true. Maybe one day you'll learn this.
As far as "a lot of CISC compilers," I can only assume that you're referring to gcc and Microsoft, because the other two players in optimization, IBM and Intel, have already moved way beyond that.
I'm just looking forward to seeing how totally completely different it's going to be from the actual game. I'd really like to see how they pull this one off.
You know, I hear they've got "Tetris the Playing Card Game" and "PacMan the pen&paper RPG" in the works as well.
It's a RISC core. But the entire CPU is not the core. The CISC instructions are broken down by the CPU. Reread my post. I mentioned this.
All things being equal, sending CISC instructions to the CPU which are then converted into RISC is faster than RISC because the CPU can figure out good ways to rearrange the ops. What you discribed is generally considered CISC because the instruction set actually sent to the CPU is a complex set of instructions, and there are multiple clock ops.
Compile-time optimizations are only really good for pure numerical stuff (of which the most common is graphics, as I said) because pipelining optimizations can be done at compile-time. For everything else, most advantage goes to just-in-time optimization, which modern day CISC does.
You should take a compiler class. They usually teach these things there.
If there were a box moving instruction, CISC would win.
CISC always wins when executing the special complex instructions thanks to pipelining - which, by the way, is much more complicated than RISC pipelining. CISC boxes usually optimize the code to help this along. This is why CISC is winning: it has a lot of the features of a JIT compiler built-in.
Each trip would take 130 seconds, but it'd be able to do 1.5 of them at a time.
The only place that RISC wins is when the instructions must all be sequential, can't be improved upon more during runtime than during compile time, and require roughly the same amount of hardware to do for the entire operation.
Where would that be? Primarily in graphics. However, for your average multiuser system, CISC is a better idea.
Lyx is not capable of reading any latex file - only Lyx produced Latex files are guaranteed to work.
And no, I am not confusing.txt and.tex. Rather, I am thinking about virtually the only way to produce still-editible fancy documents (for free) given the choice between pdf and plaintext is with latex, since.tex is a plaintext format. Reread my post.
Slashdot Mirror
How did you do a linear search? Did you just try one number after another until you reached your target number?
I could be wrong, but I think the problem is that your linear search is not searching the space uniformly; it favors lower numbers (or higher numbers) more than the opposite, despite the fact that the numbers are chosen uniformly. Further, the probability that a random search will hit a low number is higher than the probability that the linear search will reach a high number for any sufficiently large set. In other words, the RNG has an advantage probabilistically.
Instead of a linear search, try subdividing the space each time, which guarantees a roughly uniform search of the space. So, for example, if you were guessing the range of 0-10, you'd guess in this order (I subdivided and rounded up or down, alternating):
5 0 10 2 7 1 4 9 3 8 6
There are other ways of doing uniform searches. Pick one and see if your results aren't better.
Scientists could customize the system to target any protein on the surface of a cell, Chen said. He and his colleagues have seen success with about a dozen different molecules, including brain and other blood cells, he said.
They tested it on melanoma. Nowhere in the article does it say that the method only works for melanoma. Also, they imply that it could work on other kinds of cancer.
Ah...how do you decide that something is truly random? The only thing I can think of that has randomness quite apart from observations is mathematically defined random variables.
We have determined that, for example, particle decay is random because it appears to be true. The theories of quantum mechanics are based upon this observation. If we found a particle for which this did not appear to happen for long stretches of time, we would probably revise our theories, not assume that it was a truly random event so we couldn't use it as evidence.
My example wasn't a metaphor. It was an example of what happens if you ignore evidence of a trend.
You have to remember to use their claim "Within close temporal proximity of major world events there will be a spike in the graph that corelates with the number and strength of emotional response of the people involved," not "all spikes coincide with world events." The kicker is that they don't have a real good way to test for "emotional response of people involved."
Eh...not a very good debunking, IMHO.
The response to images makes sense - people would learn a response. If they really want to show the something significant, they'd have to show that people can either anticipate a correct strong response all of the time (by also showing images that would invoke no response), or show that they would invoke a strong response the first time after a series of no-responses.
The other part doesn't jive, though. The theory that this group of devices predicts disasters does not preclude the idea that it also produces false positives - or even that it also picks up something else of significance that has not been identified.
Still, I question how they go about producing these random numbers. That could be the culprit.
Oh, and as far as the straight line thing, and the curve - they're obviously talking about aggregated data. Unlike the "law of averages" as applied to a single number, the probability of getting a large number of the same values over and over can be calculated. It is very important to remember that what has happened in the past should have some weight in predicting what will happen in the future.
What if all scientists took your approach to science?
"Oh look, the apple fell from the tree, and I think it fell at the same accelaration as the last object I saw fall. I wonder if all objects fall with the same acceleration? Too bad I can't learn anything from that, since what happens in the past has no bearing at all on the future. I'm gonna go get some pie."
It is quite easy for someone versed in probability to calculate (and I'm hoping they have) the likelihood of occurance of the anomalies they have witnessed. And if they've gotten a significant result (as in - this possibility that has occured 4 times this month should only happen on average once every hundred years...), then it might be worth looking into. Of course, maybe they're just fooling themselves, or being fooled by someone else. That's an awful lot of highly educated people to not realize that an anomoly is actually normal.
...which is why so few have an account on somethingawful.com.
It's like they're saying "come here and listen to all the people who paid us money. Your opinion doesn't matter until you do too."
There's this interesting distinction in my areas (Computer Science and Computer Engineering). All the women in my graduate program were going into cognitive science or databases. None went into anything else.
Not that I blame them; I didn't go into any of those things either. It's strange, though, because those are the only two areas offered that don't have any hard math.
You have a very valid point. It just doesn't have anything to do with the actual article (and presumably nothing to do with the problem).
The article is stating that hate mongering groups are starting their own little clubs on google. They want to be flooded by inane and pointless racist remarks. Hasn't this mostly been the way of such groups? Exist in exclusive forms for a long time where the ugliness cannot be seen, and come out only when there is sufficient political power to destroy the object of obsession.
What you are describing is the convolution of a signal with a rect function (I invite you to go over to mathworld and look at the mathematical definition of convolution).
This isn't a very good approach. Even the existing methods will produce less artifacts.
It does give you a bit of insight, though: existing methods work well when you have sound that is predictable, just as yours does - that is the same pitch over the interval of an entire sample (where multiple samples make up a signal). They also tend to cause less distortion when they're doing less - so if you only change things a teeny bit, there won't be many artifacts.
As far as building back the 7 seconds, I think they mostly just start before the commercials end and have it all back at once.
If by "fairly easy" you mean "there are existing algorithms that do this badly," then you're right.
If you mean "sounds exactly the same, only faster" then you're wrong. Considering the quality of these things now, I'm not sure I wouldn't rather just let the pitch raise.
The problem is how to represent pitch. Most of the time, this is done by converting to a frequency domain and doing a shift, or by convolving the signal with a waveform that causes a signal shift (the classical example of this is using a sine wave, as is done for RF encoding). The problem is that this technique is only really good for a signal that doesn't change over time.
In fact, even the best pitch shifters assume that the pitch can be modeled as function of time and are unable to deal with randomly changing pitches very well. Lots of artifacts are still introduced when dealing with an "instrument" as complex as the human voice (on the other hand, they work great for flutes). Of course, if you don't change the pitch very much, you can get away with less artifacts.
"Support for x86, AMD64, PowerPC, UltraSparc, Alpha and MIPS processors"
Actually, there are 13 ports officially supported by Gentoo (and there are also a few side project ports which are still experimental).
Gentoo doesn't consider embedded processors part of the support base. I think the only one that isn't supported actually supported in any way in that list is VAX. It considers embedded processors a separate family of packages. Further, it covers processor families. The sparc port actually covers all Sparcs.
There are various levels of support for the 8000+ packages in Gentoo. Have a look here for the list which specifies the level of support for any given package.
VMware uses a processor emulation layer. In other words, its sandbox contains a virtual processor that interprets the output and converts it into native code. This means that at best you can get 1/3 of native speed with VMWare. Linux can run at close to native speed under Windows (using coLinux or others), under BSD (using userspace linux), or under MacOS (using userspace linux). The way this works is that there are hooks built into linux that basically make it act less like an operating system and more like an app.
VMWare can make anything run on anything else, but it isn't pretty. And unlike those things I mentioned, it's not free either.
Portage has been ported to BSD. Incidentally, there's also a cygwin and a MacOS port. I just call it a Gentoo port to NetBSD because that's what the people who made it call it.
Wow...it runs on almost as many things as Gentoo does. :)
Maybe that's why there's a Gentoo port to BSD.
Seriously, though, at this point, Linux can run under other operating systems without a CPU emulation layer, and it can do just as many strange processors. Not that I'm discounting NetBSD; there are plenty of reasons why NetBSD has an edge.
This just isn't one of them.
Yeah...evolution is quite stable, and it will work with exchange. So I guess "until" is "now."
The thing you're missing (and the one that I'm sure is mentioned elsewhere on this thread, and that the author is hinting at) is that one of the main arguments for a lower Linux TCO than Windows TCO is that it avoids the Microsoft tax, thus saving you a lot up front to offset the possibly higher costs of learning the system.
I guess this isn't usually true in the corporate world.
Perhaps you should take a psychology course or two. Why do you suppose young men kill young women?
It's for pleasure. It's a sensual experience like rape. Remember the serial killer Ted Bundy? Do you know what he claimed set him off? I'll give you a huge hint: porn.
Seeing violence on TV has not been substantially linked to actual violence - especially not serial violence - but sexual awakening before one is ready can create all kinds of sexual deviance - including a homocidal fetish. This is much worse, and only one of the reasons why people should already have a good understanding of who they are before they do something like look at porn.
Actually, the typing class I took in middle school helped me learn to type better.
As I said, this was the first in the school's collection of computer classes. Ultimately, that typing class was just a waste of my time.
Lucky you. I graduated in 97. I took two of the three computer classes my school offered before giving up in disgust.
Class 1: Typing. Not on computers - on typewriters.
Class 2: Microsoft Works for dos FOR A YEAR (for all those who don't know, it's less complicated than office. It took all of a week to learn everything we covered in that course). The teacher optimized it for the XTs that we were using so that it would fit on our floppies. That's right - no harddrives on these machines.
Of course, there were computers with win 3.1 in the libraries, but the libraries weren't open after school, we weren't allowed to use disks on them, and teachers didn't schedule regular trips to the library for that purpose. They just sat there doing nothing.
I don't blame the football program, though - that pays for itself. I was in the band, though, and I know for a fact we had more than a few luxuries that could easily have been spent on twenty or so pentiums (for instance, the $2000 stereo system we got my junior year).
Nearest I can figure, they used this logic for giving the band so much money:
90% of AP and gifted students were in our band => a lot of extra federal money given to support these students (not needed in most cases) => spend it on what the AP and gifted students do in case it helps them
Teachers who become more afluent and wise don't become administrators. Politicians become adminstrators, and not because they care, usually - usually it's to get a better political job later. This is the problem, I think.
That would be true in most cases, but Purdue doesn't have any contractual relationship with Netscape.
All they have is that warm fuzzy feeling.
If you're rich because of what's being shared
What? What happens if you're rich becauseo f what's being shared?
The economy will collapse?
You'll be less rich?
The forces of Skeletor will invade your home and send you to the dungeons?
I'm sure that this could have been an insightful post but you didn't finish your
RTF and doc are just too complicated for them to be used by programs outside of Open Office and Microsoft products.
I would like a simpler format.
I would prefer to allow any program that is capable of printing a layout to export to some document format, and right now the only possibility is pdf and ps, both of which have no WYSIWG editors.
Anyone else feel that way?
My almamater uses it (Purdue university). My guess is that it's a compromise to suit the business drones. They like closed products for some reason. I think it gives them a warm, fuzzy feeling.
Netscape will always be able to feed on Mozilla as long as such drones exist.
You don't know what "Just in Time" means, do you?
Here's an overview of it.
You can only reorder so much. If the order of execution depends upon data - for example, programs that accept data from the users, from a database, from chaotic systems, or from other programs, then you can't predict it at compile time. Of course, you may not believe this, but that doesn't make it any less true. Maybe one day you'll learn this.
As far as "a lot of CISC compilers," I can only assume that you're referring to gcc and Microsoft, because the other two players in optimization, IBM and Intel, have already moved way beyond that.
I'm just looking forward to seeing how totally completely different it's going to be from the actual game. I'd really like to see how they pull this one off.
You know, I hear they've got "Tetris the Playing Card Game" and "PacMan the pen&paper RPG" in the works as well.
It's a RISC core. But the entire CPU is not the core. The CISC instructions are broken down by the CPU. Reread my post. I mentioned this.
All things being equal, sending CISC instructions to the CPU which are then converted into RISC is faster than RISC because the CPU can figure out good ways to rearrange the ops. What you discribed is generally considered CISC because the instruction set actually sent to the CPU is a complex set of instructions, and there are multiple clock ops.
Compile-time optimizations are only really good for pure numerical stuff (of which the most common is graphics, as I said) because pipelining optimizations can be done at compile-time. For everything else, most advantage goes to just-in-time optimization, which modern day CISC does.
You should take a compiler class. They usually teach these things there.
If there were a box moving instruction, CISC would win.
CISC always wins when executing the special complex instructions thanks to pipelining - which, by the way, is much more complicated than RISC pipelining. CISC boxes usually optimize the code to help this along. This is why CISC is winning: it has a lot of the features of a JIT compiler built-in.
Each trip would take 130 seconds, but it'd be able to do 1.5 of them at a time.
The only place that RISC wins is when the instructions must all be sequential, can't be improved upon more during runtime than during compile time, and require roughly the same amount of hardware to do for the entire operation.
Where would that be? Primarily in graphics. However, for your average multiuser system, CISC is a better idea.
Lyx is not capable of reading any latex file - only Lyx produced Latex files are guaranteed to work.
.txt and .tex. Rather, I am thinking about virtually the only way to produce still-editible fancy documents (for free) given the choice between pdf and plaintext is with latex, since .tex is a plaintext format. Reread my post.
And no, I am not confusing