Well, that is true. I had forgotten about that detail. (Big detail though!)
But, what about hypothesizing certain criteria for the state of a folded protein? Such as... minimization of free energies, auto/cross correlation functions (pair distribution functions) going to zero (likely shell values), ergodicity in ensemble of states (time average = ensemble average). These are all indications of a system gone to equilibrium. Now, that's not necessarily the folded state..but it is a stable state of the protein (of which the folded state might be the dominant one).
The Pande group shoot 100,000s of trajectories through space state space to calculate those rates. Most of those trajectories do not go in the right direction, but some do. It's certainly possible to start testing some of the criteria listed above and see which ones tell you which directions are the "right ones" (and then see if those criteria are universal, of course).
The folding process can take place over microseconds of time.
Many processes, such as conformational changes, ion flow, and ligand binding can occur much faster (1 to 100 nanoseconds).
So if you start to simulate an already folded protein and watch its dynamics you can still discover what sort of behaviors it may do. Without having folded the protein.
They've folded proteins whose kinetics are 1st order (ie. small enough proteins).
The folding problem is the one of the hardest ones. So don't get all blustery about it not being a solved problem. Cancer hasn't been cured yet either.
What looks like a laptop when folded and a Death Star access panel when open is one of the most ridiculous third-party controllers ever conceived. Infra-red beams shot out of this unit's surface and tried to interpret hand motions as controller movements, but did it work? Know anyone who owned this thing? There you go.
and it sorta worked. If you used the included aircraft controller thingy and didn't move it too quickly then it would respond decently fast. Of course, if you tried to "punch" the sensor areas when playing Mike Tyson's Punch Out then it usually just went berzerk and did a bunch of crazy moves.
Incidentally, are there any Nintendo fans that want to buy one slightly used U-Force (emphasis on the slightly)? Ebay beckons...
For scientific computing, I think the Cell's advantages will heavily depend on how many vectorizable loops are in the existing code. For example, in Molecular Dynamics (MD) code, one frequently calculates the forces between each pair of atoms (basically solving a = F/m over and over). Systems with N atoms have N*(N-1)/2 pairs so when N is large that loop requires significant calculations (this excludes cut off distances and other tricks). MD code is used to (try to) fold proteins and to solve a host of other important problems.
So if you can store all of the force calculation instructions in the SPEs, insert atom positions, and retrieve the force numbers, then you should be able to minimize wasted cycles decently well. Like you said, the amount of instructions has to be small and all of the atom positions will not fit into memory at the same time, requiring shuffling. But I think, for this application, that the Cell will make a big difference.
For other scientific codes, like Fast Fourier Transforms, the situation is very similar. There are a large number of vectorizable loops which do not require branch prediction and typically use a small number of instructions per loop. While the shuffling of data in/out of memory has to be minimized IBM should come up with some automatic ways to do this. Also, I think IBM is trying to put most of the burden of determining out of order execution on the compiler. I don't know how well this will work, considering that Intel tried to do the same thing with Itanium (or something similar), but maybe the status of compiler research has improved since then.
The other staple of scientific computing is the solution of linear equations (LINPACK, LAPACK, etc). If they can perform well in a LAPACK benchmark for large matrices then I think they will be ok. So many scientific applications, including ODE integrators, PDE integrators, etc, use a Ax=b solver that once they provide a fast port of one, it'll encourage the porting of code that heavily uses it.
And, of course, since graduate students do a lot of the grunt work of scientific computing, the fun of working on some exotic hardware like the Cell might be enough to overcome the barrier-to-port. As long as they give away the SDKs for free and are generous with the hardware.
Either way, it's exciting that a completely new type of hardware will soon be available.
Of the many FPSs I've played (some futuristic, some realistic), BF2 is of the best ever. You really do feel like you're on a battlefield (except the dying part). I wonder if the Army uses it for training (a la America's Army).
I agree that a SCM/Wiki would the ultimate killer app for all sorts of legislative areas. Maybe include contracts and other business agreements in there as well. The Wiki would provide the debate/discussion utility while the SCM handles revisioning/etc and linking.
It's a nice idea until someone sneaks in an amendment that would never pass on the floor.
Unfortunately, I think this frequently happens already. It's the legislation by obscurity trick...stick a small bit of language into a 700+ page, $1b+ appropriation law and it's rare that someone will notice it or try to line item veto it. And it's especially rare that a huge bill is not passed because of a small bit of language (usually containing pork).
Now, if every individual subsection had its own link and "discussion" section, then maybe people can draw attention to the porks parts that no one previously noticed. And if all of the pork parts are removed from the bill, then we can be talking about some serious reductions in spending. But, and let's remember this, it is in Congress's interest to keep the pork under wraps so any mechanism that provides more debate or more transparency will be fought against. However, if pork-fighters successfully paint the pork-entrenced Congressmen as self-serving, gluttonous, and deceitful...well they might favor transparency over not being reelected.
And, another thing...most legislators are much older than the 'Internet generation'. How many of them know what source revisioning is? Maybe some of them know what a Wiki is. Any killer app needs to be extremely user-friendly. Our legislative branch didn't get elected because they were the smartest of the bunch....
First, let's pretend that U.S Congressman carefully debate the pros and cons of proposed legislation.
Then, I think that Wikis would make a wonderful forum for developing legislation. Especially those 400+ page variety bills that seem to contain all sorts of pork. Of course, modification of a Wiki page would be restricted to Congressman and their aids (hopefully not lobbyists..). State and local legislative wikis might also be useful.
Hyperlinking between chapters/portions of bills might make reading the bill easier and focus attention on poorly written parts.
Of course, for a Wiki to be useful, the debate has to be in the open and not in the proverbial "smoking room". O well!
Somebody should patent... "An open patent review program that would set up a system on the patent office Web site where visitors could submit search criteria and subscribe to electronic alerts about patent applications in specific areas." and "the creation of a patent quality index that would serve as a tool for patent applicants to use in writing their applications". (From the Article)
That way, the USPTO will have to pay royalties to reform the patent process. Oh, the irony.
It does. But the game is turn-based (you take your time to move, press enter, let your opponent move, repeat). So it's not exactly as thrilling as jumping off a high pedestal, turning 180 degrees, and (midair) firing a rocket at your soon-to-be-dead opponent. It's certainly less twitch.
I like CivIV though and it's a great game. By its never nature, though, it's not a fantastic action packed multiplayer game...unless you like to play chess on the internet. Yes, it feels like playing chess on the internet.
I agree that the major challenge is to figure out what DNA to synthesize. And we need some procedural method to explicitly determine the DNA sequence that performs some specific function. Although, I don't know if thinking about these systems as modular electronics is the best way. These systems certainly are modular, but the fact that there are no wires and no clear separation between modular devices means that a full modularization of the entire system into separate parts is a real challenge. It can probably be achieved, but it requires a lot of engineering of the individual parts.
But, even after that, I think the techniques in building organisms will be very different from building electronics. Even simple bacteria use regulatory mechanisms acting on multiple different layers: transcriptional, translational, post-translational. The code itself (the DNA) can get cut, reversed, and reinserted to generate "dynamic coding". And then there's diffusion, signal tranduction, active transport, etc that are all completely foreign to the design of electronic chips or "programming".
So maybe thinking as an electrical engineer or computer scientist will help in the beginning, but by the end a new type of engineering has to be created. I'm a chemical engineer myself so it's natural to think of these systems as small chemical reactors, engaging in numerous biochemical reactions with extremely dilute concentrations. So maybe I'm biased in that regard. I describe these systems using chemical kinetics and the answers often match experimental data. It's probably possible to use a detailed kinetic model of a "part" to fully characterize its modular behavior. And then use that information to better stitch together multiple parts. But, who knows, the field is young and we will eventually figure out the best ways of doing things.
Re:Unlikely, but exciting if they pull it off
on
Writing Genetic Code
·
· Score: 1
Well, they know the sequence of many different bacterium. What happens if they were to synthesize the entire DNA chromosome for one of those bacteria and inject it into an plasma membrane whose DNA was removed? Would you call that synthetically creating a bacteria or just synthetically creating all of the programming of the bacteria?
Now, what happens if you decide to change parts of the DNA chromosome by inserting your own genes or entire pathways that perform interesting and useful functions? If you do the same procedure you will have created something which does not exist in nature (that we of know of). So once the procedure is practiced it can be used to generate entirely novel bacterium.
I think that's where the excitement comes from. Our ability to create arbitrary DNA sequences will make it a lot easier to engineer bacteria.
And this isn't BS....there are numerous DNA synthesis companies who will sell by the base pair (around $1/bp). Typical bacterial chromosomes are ~1 million bp and I believe a few of the synthesis companies have actually produced contiguous strands of DNA of that length or greater. So it's happening and it's real! (But still too expensive for my budget!)
On the main page they do write "the free encyclopedia that anyone can edit."
1) You always get what you pay for.
2) If it says anyone can edit it, that does include people who have no idea what they're writing about.
3) It's amazing that even though (1) and (2) are true, that Wikipedia is, on average, very factual. This is especially true in math and science, where there really is a right answer to be written. But even "experts" can debate the meaning of the Civil War...so when controversy erupts on the Civil War page among both experts and non-experts, it's not because the non-experts have no clue, it's because the experts disagree as well. All Wikipedia articles must also have sources. So when a student cites Wikipedia, they should really be citing the original source.
Slashdot Mirror
FTA,
"Each virus, and thus each wire, is only 6 manometers -- 6 billionths of a metre -- in diameter, and 880 manometers long, the researchers said."
Wow. Only 6 manometers! I wish I knew what a manometer was.
Go Reuters.
Ok, true true. This just means I (and every other person studying stat mech) will have a job for the forseeable future. :)
Well, that is true. I had forgotten about that detail. (Big detail though!)
;)
But, what about hypothesizing certain criteria for the state of a folded protein? Such as... minimization of free energies, auto/cross correlation functions (pair distribution functions) going to zero (likely shell values), ergodicity in ensemble of states (time average = ensemble average). These are all indications of a system gone to equilibrium. Now, that's not necessarily the folded state..but it is a stable state of the protein (of which the folded state might be the dominant one).
The Pande group shoot 100,000s of trajectories through space state space to calculate those rates. Most of those trajectories do not go in the right direction, but some do. It's certainly possible to start testing some of the criteria listed above and see which ones tell you which directions are the "right ones" (and then see if those criteria are universal, of course).
There, that's a PhD project right there.
It was 50 nanoseconds.
The folding process can take place over microseconds of time.
Many processes, such as conformational changes, ion flow, and ligand binding can occur much faster (1 to 100 nanoseconds).
So if you start to simulate an already folded protein and watch its dynamics you can still discover what sort of behaviors it may do. Without having folded the protein.
http://www.stanford.edu/group/pandegroup/
They've folded proteins whose kinetics are 1st order (ie. small enough proteins).
The folding problem is the one of the hardest ones. So don't get all blustery about it not being a solved problem. Cancer hasn't been cured yet either.
I don't know if having a low ID is a badge of honor or shame.
This is Slashdot, of course.
I think you should STFU Up.
;)
Happy?
Bah. Posers.
"While a baby girl copies the scene where Lisa appears to be driving Marg's car."
Not Lisa. Maggie.
Duh?
I owned one of these ....
...
What looks like a laptop when folded and a Death Star access panel when open is one of the most ridiculous third-party controllers ever conceived. Infra-red beams shot out of this unit's surface and tried to interpret hand motions as controller movements, but did it work? Know anyone who owned this thing? There you go.
and it sorta worked. If you used the included aircraft controller thingy and didn't move it too quickly then it would respond decently fast. Of course, if you tried to "punch" the sensor areas when playing Mike Tyson's Punch Out then it usually just went berzerk and did a bunch of crazy moves.
Incidentally, are there any Nintendo fans that want to buy one slightly used U-Force (emphasis on the slightly)? Ebay beckons
That's a very good summary of the Cell's SPEs.
For scientific computing, I think the Cell's advantages will heavily depend on how many vectorizable loops are in the existing code. For example, in Molecular Dynamics (MD) code, one frequently calculates the forces between each pair of atoms (basically solving a = F/m over and over). Systems with N atoms have N*(N-1)/2 pairs so when N is large that loop requires significant calculations (this excludes cut off distances and other tricks). MD code is used to (try to) fold proteins and to solve a host of other important problems.
So if you can store all of the force calculation instructions in the SPEs, insert atom positions, and retrieve the force numbers, then you should be able to minimize wasted cycles decently well. Like you said, the amount of instructions has to be small and all of the atom positions will not fit into memory at the same time, requiring shuffling. But I think, for this application, that the Cell will make a big difference.
For other scientific codes, like Fast Fourier Transforms, the situation is very similar. There are a large number of vectorizable loops which do not require branch prediction and typically use a small number of instructions per loop. While the shuffling of data in/out of memory has to be minimized IBM should come up with some automatic ways to do this. Also, I think IBM is trying to put most of the burden of determining out of order execution on the compiler. I don't know how well this will work, considering that Intel tried to do the same thing with Itanium (or something similar), but maybe the status of compiler research has improved since then.
The other staple of scientific computing is the solution of linear equations (LINPACK, LAPACK, etc). If they can perform well in a LAPACK benchmark for large matrices then I think they will be ok. So many scientific applications, including ODE integrators, PDE integrators, etc, use a Ax=b solver that once they provide a fast port of one, it'll encourage the porting of code that heavily uses it.
And, of course, since graduate students do a lot of the grunt work of scientific computing, the fun of working on some exotic hardware like the Cell might be enough to overcome the barrier-to-port. As long as they give away the SDKs for free and are generous with the hardware.
Either way, it's exciting that a completely new type of hardware will soon be available.
Of the many FPSs I've played (some futuristic, some realistic), BF2 is of the best ever. You really do feel like you're on a battlefield (except the dying part). I wonder if the Army uses it for training (a la America's Army).
Hmm. Why is that? I do scientific computing, but my knowledge of chip design is iffy.
The SPEs can do double precision, but at half the flops.
I'm just curious. How much do programmers usually charge to write good GUIs?
How about Web based (Ajax) GUIs?
Instead of creating sequel after sequel of movies, they should just create a cartoon series based on it.
Think Saturday morning / after school + DVD sales.
And then they don't have to worry about dilution of the franchise...a series can go on and on and on and no one cares!
I agree that a SCM/Wiki would the ultimate killer app for all sorts of legislative areas. Maybe include contracts and other business agreements in there as well. The Wiki would provide the debate/discussion utility while the SCM handles revisioning/etc and linking.
....
It's a nice idea until someone sneaks in an amendment that would never pass on the floor.
Unfortunately, I think this frequently happens already. It's the legislation by obscurity trick...stick a small bit of language into a 700+ page, $1b+ appropriation law and it's rare that someone will notice it or try to line item veto it. And it's especially rare that a huge bill is not passed because of a small bit of language (usually containing pork).
Now, if every individual subsection had its own link and "discussion" section, then maybe people can draw attention to the porks parts that no one previously noticed. And if all of the pork parts are removed from the bill, then we can be talking about some serious reductions in spending. But, and let's remember this, it is in Congress's interest to keep the pork under wraps so any mechanism that provides more debate or more transparency will be fought against. However, if pork-fighters successfully paint the pork-entrenced Congressmen as self-serving, gluttonous, and deceitful...well they might favor transparency over not being reelected.
And, another thing...most legislators are much older than the 'Internet generation'. How many of them know what source revisioning is? Maybe some of them know what a Wiki is. Any killer app needs to be extremely user-friendly. Our legislative branch didn't get elected because they were the smartest of the bunch
First, let's pretend that U.S Congressman carefully debate the pros and cons of proposed legislation.
Then, I think that Wikis would make a wonderful forum for developing legislation. Especially those 400+ page variety bills that seem to contain all sorts of pork. Of course, modification of a Wiki page would be restricted to Congressman and their aids (hopefully not lobbyists..). State and local legislative wikis might also be useful.
Hyperlinking between chapters/portions of bills might make reading the bill easier and focus attention on poorly written parts.
Of course, for a Wiki to be useful, the debate has to be in the open and not in the proverbial "smoking room". O well!
Somebody should patent ... "An open patent review program that would set up a system on the patent office Web site where visitors could submit search criteria and subscribe to electronic alerts about patent applications in specific areas." and "the creation of a patent quality index that would serve as a tool for patent applicants to use in writing their applications". (From the Article)
That way, the USPTO will have to pay royalties to reform the patent process. Oh, the irony.
(This is a joke.)
It does. But the game is turn-based (you take your time to move, press enter, let your opponent move, repeat). So it's not exactly as thrilling as jumping off a high pedestal, turning 180 degrees, and (midair) firing a rocket at your soon-to-be-dead opponent. It's certainly less twitch.
I like CivIV though and it's a great game. By its never nature, though, it's not a fantastic action packed multiplayer game...unless you like to play chess on the internet. Yes, it feels like playing chess on the internet.
I agree that the major challenge is to figure out what DNA to synthesize. And we need some procedural method to explicitly determine the DNA sequence that performs some specific function. Although, I don't know if thinking about these systems as modular electronics is the best way. These systems certainly are modular, but the fact that there are no wires and no clear separation between modular devices means that a full modularization of the entire system into separate parts is a real challenge. It can probably be achieved, but it requires a lot of engineering of the individual parts.
But, even after that, I think the techniques in building organisms will be very different from building electronics. Even simple bacteria use regulatory mechanisms acting on multiple different layers: transcriptional, translational, post-translational. The code itself (the DNA) can get cut, reversed, and reinserted to generate "dynamic coding". And then there's diffusion, signal tranduction, active transport, etc that are all completely foreign to the design of electronic chips or "programming".
So maybe thinking as an electrical engineer or computer scientist will help in the beginning, but by the end a new type of engineering has to be created. I'm a chemical engineer myself so it's natural to think of these systems as small chemical reactors, engaging in numerous biochemical reactions with extremely dilute concentrations. So maybe I'm biased in that regard. I describe these systems using chemical kinetics and the answers often match experimental data. It's probably possible to use a detailed kinetic model of a "part" to fully characterize its modular behavior. And then use that information to better stitch together multiple parts. But, who knows, the field is young and we will eventually figure out the best ways of doing things.
Well, they know the sequence of many different bacterium. What happens if they were to synthesize the entire DNA chromosome for one of those bacteria and inject it into an plasma membrane whose DNA was removed? Would you call that synthetically creating a bacteria or just synthetically creating all of the programming of the bacteria?
Now, what happens if you decide to change parts of the DNA chromosome by inserting your own genes or entire pathways that perform interesting and useful functions? If you do the same procedure you will have created something which does not exist in nature (that we of know of). So once the procedure is practiced it can be used to generate entirely novel bacterium.
I think that's where the excitement comes from. Our ability to create arbitrary DNA sequences will make it a lot easier to engineer bacteria.
And this isn't BS....there are numerous DNA synthesis companies who will sell by the base pair (around $1/bp). Typical bacterial chromosomes are ~1 million bp and I believe a few of the synthesis companies have actually produced contiguous strands of DNA of that length or greater. So it's happening and it's real! (But still too expensive for my budget!)
On the main page they do write "the free encyclopedia that anyone can edit."
1) You always get what you pay for.
2) If it says anyone can edit it, that does include people who have no idea what they're writing about.
3) It's amazing that even though (1) and (2) are true, that Wikipedia is, on average, very factual. This is especially true in math and science, where there really is a right answer to be written. But even "experts" can debate the meaning of the Civil War...so when controversy erupts on the Civil War page among both experts and non-experts, it's not because the non-experts have no clue, it's because the experts disagree as well. All Wikipedia articles must also have sources. So when a student cites Wikipedia, they should really be citing the original source.
Four: Stop whining and go purchase the Encyclopedia Britannica for $125 or whatever.
Wikipedia is free, useful, and better than anything you could produce yourself. So stop whining!
That's very similar to another game I've heard of.
;)
There's 191 teams that all run around the field for months on end with no one scoring or even attempting to score.
It's called the UN.