Basically you need a lot more than a single 2D image. You need a stack, and from the stack you can measure the angle you suggest. Your very own link illustrates this. What this technique allows you to do is to measure depth of point-like objects to a very good resolution, better that can be usually done with confocal imaging, but this is not easily applicable to other modalities.
I think you might be confused because in confocal microscopy, a "single image" is a 3D stack of optical sections.
Also you need to insert a phase mask into the focal plane of the optical equipment. This is not easily done.
Basically the author's technique referenced here is applicable to a different field of modalities, so it is differently useful.
A. Orth and K. B. Crozier, "Light field moment imaging", non paywalled version:
From Crozier's web page: http://crozier.seas.harvard.edu/publications-1/2013/76_Orth_OL_2013.pdf
So commentators here can be a little more informed about what these guys are really doing. From quick reading, I'm not an expert but this is different from the usual depth from defocus. It allows for some 3D information but not a lot, obviously. Still could be useful.
No, because the light rays coming from the crab nebula are all parallel.This relies on light coming from the sample at various angles to the lens. Essentially the sample must be close to the optical system.
In radio astronomy, you can get some 3D information from radio sources, because radiotelescopes can measure the phase directly.
I'm not sure how you can achieve the DHPSF from standard optical equipment. How do you measure the phase mask from a single image? Do you have a reference ?
So these employees were forced to use the UK PO accounting software, which had bugs, and which showed in some instances imaginary shortfalls that they had to repay with no way of defending themselves. Sounds peachy! I hope some judge throws the book at the UK post office and finds some way to redress the situation.
Actually Intel is pretty much the king of the hill at the moment for HPC. They don't have a "GPU" solution, but they do have a massively parallel CPU + PCIe compute card available called the "Xeon Phi". Extremely confusing, yet this is what the current fastest supercomputer uses
1 exaflops (10^{18} flops) is 1000 petaflops (10^{15} flops). The Chinese are 3.4% of the way there. Exaflop-scale computers are realistically expected for 2019, i.e. in 6 years' time.
Meanwhile, India is supposedly building a 140 exaflop computer for 2017
One could look at other statistics to find out which of the alternative is more likely or if yet another series of reasons is the explanation for the current situation. For instance the rate of violent crime, which is rather high in the US compared with a lot of other Western countries (e.g. intentional homicide rate is basically 5x higher in the US than most of Western Europe), and the reasons why people are put (and made to stay) in prison. So violence rate is an explanation, but not the only one. In many Western countries, prison is the solution of last resort for violent and dangerous criminals. In the US, thanks to the ongoing "war on drugs" since President Nixon, it seems that even non-violent drug offenses can land you in jail for a long time. Also, sentences tend to be significantly longer in the USA compared to other countries. Sentencing laws (three strikes, etc) are way tougher, and last but not least, prison building and operating is a private business in the USA. Why don't you read about it?
Overall it perhaps emerges that the USA is vastly better at catching and managing criminal than Honduras, that has almost 20x as much crime than the US but only few prisoners. However it seems that the USA also likes to put and keep people in prison rather than look for alternatives for dealing with their problems (drug rehabilitation for instance).
Some statistics that I find interesting: about 90% of federal prisoners are there for non-violent offenses; there are currently more African-American in prison in the USA right now than there were slaves in 1850; and 67% of ex-prisoners reoffend within 3 years: when they get out of jail they cannot find a job, they are ineligible for welfare, and so they are caught in an endless spiral.
If you remember your app name, and if you type nimbly, then Win+R works. My parents for instance are not very good at either, but they can search visually in a list. They prefer the start button.
What do you mean "refuses to admit"? We are not talking about a level playing field here. Overall developer time is not quite the same between Windows & Linux, Photoshop and Gimp and MS-Office / Libre-Office, and commercial software can't survive if enough people don't pay up. People choose pirated software when they can but don't necessarily use it. Personally I have paid for all of these apps (or have my employer pay) yet I find that I'm enjoying Linux much more than Windows. An office suite is a nuisance (compared to LaTeX) I have to deal with, but I use Libre/Open as much as MS when I have to. I hardly use Photoshop or Gimp but I do use Premiere Pro.
Nonetheless I use to use Illustrator but Adobe left it behind somewhat so now I use Inkscape quite happily. I sometime have to use Matlab sometime to communicate with colleagues of for courses, but I much prefer Python, which is now almost as good even for the stuff Matlab is really good at (linear algebra).
In other words, choices are always good. Useful OSS is not going away. Eventually Gimp is going to be good enough for most things and people actually are using it. That is all that matters.
In fact the visible universe is significant bigger than 12 billion ly in radius, because the universe is expanding. The co-moving distance to the edge of the universe is more like 46 billion ly. See this wikipedia page for more details. If this is true, 3.6 billion ly is about 8% of the distance to the edge of the visible universe, which represents a partial volume of (3.6/46)^3 = 5 10^{-4} the size of the visible universe (or 0.05%).
It is hard not to be dismissive of engineers when all they care about is building stupidly expensive gadgets that entrap everyone into thinking they are doing something useful. Software engineers are the worst: they build buggy software and toy apps that are nothing but smoke and mirrors instead of helping solve the real world's problem. What value do they really create?
See how it goes? With this attitude no one does anything useful: farmers, politician, army, police, doctors, scientists, marketing, finance, accounting people are all good for the bin. The saying goes that the economy needs economist the same way the weather needs meteorologists. This is true to some extent, but one forgets that one might be becoming better at economics or meteorology as time goes by and methods improve. This has and this will pay off.
Professor Dyson, you have been an actor and a witness of the huge expansion of hard sciences into everyday life. Thanks to scientific progress, particularly in the understanding of the basic physical laws, we have been able to improve our way of life almost beyond belief: energy, transport, and even exploration, now going beyond our planet all have been hugely transformed in the last 100 years.
However in the last few decades, our understanding of physics has become good enough for "most" things, and physics seems stuck. It looks like string theory is mostly sterile, with no useful or even testable prediction in spite of huge efforts, with nothing to really replace it. Also, the standard model is too good. We found the Higgs almost exactly where it was expected. On the one hand do not have a working theory of quantum gravity, and on the other we seem unable to produce a useful fusion reactor in spite of foreseeable dire needs in safe energy consumption. We already know that other efforts including solar, sea and wind energy will not be sufficient to guarantee our children a safe future.
At the same time we witness a resurgence of religious beliefs at home and abroad that is causing societal and international concerns. According to some recent polls, many people do no longer "believe" in science, or even maths. We are seemingly stuck in a hard recession, pollution is a problem finally hitting planet-wide proportions. This is not looking too good.
What do we do? How do we restore the public's confidence in progress? How do we go forward? These are huge questions and it would be nice to have a perspective on this from your point of view.
I'm not really at liberty to describe the research culture at CSIRO in great detail, but it is, or at least was, as the articles say, very application-driven and short-term, external-earning motivated. This was only in one division, I cannot speak for the whole of the organization, however these stories seem to indicate that the problem is widespread.
I was at CSIRO between the mid-1990 to the mid 2000, and I have seen it progressively become a very tough place to do research. I was very very happy to leave. I'm not a top researcher by any stretch of the imagination, and I was never bullied, although I did experience unpleasant conflict. Ever since I've left (for academia) I've been more free to conduct my research the way I wanted it, I have found that it is indeed easier to find funding (so far). Looking for funding first and doing skunk research second is a sure way to kill imagination and generate stress, dissatisfaction and mistrust, not to mention poor results. Scientists are not necessarily good salespeople (too frank). Basically CSIRO was (and apparently still is in some places) in some ways a toxic place for scientists.
I hope it improves. CSIRO is nowhere near the top 10 rank it seeks to achieve, at least in the areas I'm familiar with, but there are still very good people working there.
Sure today's PC are insanely powerful for watching video, for running a word processor or a spreadsheet program. However they are just about good enough for real time speech processing and we are a long way from the processing power required for fast-enough computer vision, for instance. This would literallybe a game changer on your PC or your smartphone. Everybody was awed at first by the MS Kinect, but the reality is that the SDK is not fast enough and not precise enough to be all that useful beyond simple things.
Stronger AI still requires massive amount of computer power for machine learning. A lot of work is being done in this area. Look up the program of recent machine learning or computer vision conferences, and be awed.
The particles generated at the even horizon would be essentially photons of every wavelength, which are their own anti-particle. This is because photons can have any energy, which is not the case for massive particles. Only very light black holes (with very high energy output) would be able to create anything other than photons.
Although a common misconception, this is not correct. Please read the article. In the astronaut's frame of reference in freefall, nothing special happens at the event horizon (that is, if there is no firewall after all), and the even horizon is reached in finite time. For time to apparently accelerate like you describe, the astronaut would have to hover near the event horizon. They could not be in free fall.
That is a little schematic. Most codes running on HPC that I know are not trivial, typically expensive CFD calculations with 3D fluid / structure interaction, that may actually require quite a few thousand of hours of computation, e.g. blood flow simulations in the brain. Some scientists I know study physical geo-mechanical models that require them to analyze thousands of 2048^3 tomography data where they identify and simultaneously track tens of thousands of individual grains of sand as well as the relationship between thm, in order to find out what happens during terrain compression.
Access to the machines capable of handling such loads is competitive and in fact expensive. The technical staff managing these machines is competent. If you actually win time on these machines, they will help you make your software run well on them.
So yes, I'm perfectly willing to believe that some HPC machines are not very well managed but this does not mean that this is the case everywhere.
Slashdot Mirror
Sorry, with the clicky:
non-paywalled version of the article.
I've done some quick research into what you suggest:
http://www.stanford.edu/group/moerner/sms_3Dsmacm.html
Basically you need a lot more than a single 2D image. You need a stack, and from the stack you can measure the angle you suggest. Your very own link illustrates this. What this technique allows you to do is to measure depth of point-like objects to a very good resolution, better that can be usually done with confocal imaging, but this is not easily applicable to other modalities.
I think you might be confused because in confocal microscopy, a "single image" is a 3D stack of optical sections.
Also you need to insert a phase mask into the focal plane of the optical equipment. This is not easily done.
Basically the author's technique referenced here is applicable to a different field of modalities, so it is differently useful.
Cheers.
A. Orth and K. B. Crozier, "Light field moment imaging", non paywalled version:
From Crozier's web page: http://crozier.seas.harvard.edu/publications-1/2013/76_Orth_OL_2013.pdf
So commentators here can be a little more informed about what these guys are really doing. From quick reading, I'm not an expert but this is different from the usual depth from defocus. It allows for some 3D information but not a lot, obviously. Still could be useful.
All the best.
No, because the light rays coming from the crab nebula are all parallel.This relies on light coming from the sample at various angles to the lens. Essentially the sample must be close to the optical system.
In radio astronomy, you can get some 3D information from radio sources, because radiotelescopes can measure the phase directly.
I'm not sure how you can achieve the DHPSF from standard optical equipment. How do you measure the phase mask from a single image? Do you have a reference ?
Poorly constructed, but very insightful article. Thanks for the link!
So these employees were forced to use the UK PO accounting software, which had bugs, and which showed in some instances imaginary shortfalls that they had to repay with no way of defending themselves. Sounds peachy! I hope some judge throws the book at the UK post office and finds some way to redress the situation.
Yes everyone spies on everyone, but it is still embarrassing when one is caught red-handed.
Actually Intel is pretty much the king of the hill at the moment for HPC. They don't have a "GPU" solution, but they do have a massively parallel CPU + PCIe compute card available called the "Xeon Phi". Extremely confusing, yet this is what the current fastest supercomputer uses
http://www.datacenterdynamics.com/focus/archive/2013/06/xeon-phi-powered-supercomputer-tops-top500
Xeon phi is easier to deal with than Nvidia's solution for GPU, essentially because it is currently much easier to program.
http://goparallel.sourceforge.net/independent-test-xeon-phi-shocks-tesla-gpu/
1 exaflops (10^{18} flops) is 1000 petaflops (10^{15} flops). The Chinese are 3.4% of the way there. Exaflop-scale computers are realistically expected for 2019, i.e. in 6 years' time.
Meanwhile, India is supposedly building a 140 exaflop computer for 2017
Better get a move on I guess.
This is an interesting point, with some caveat.
One could look at other statistics to find out which of the alternative is more likely or if yet another series of reasons is the explanation for the current situation. For instance the rate of violent crime, which is rather high in the US compared with a lot of other Western countries (e.g. intentional homicide rate is basically 5x higher in the US than most of Western Europe), and the reasons why people are put (and made to stay) in prison. So violence rate is an explanation, but not the only one. In many Western countries, prison is the solution of last resort for violent and dangerous criminals. In the US, thanks to the ongoing "war on drugs" since President Nixon, it seems that even non-violent drug offenses can land you in jail for a long time. Also, sentences tend to be significantly longer in the USA compared to other countries. Sentencing laws (three strikes, etc) are way tougher, and last but not least, prison building and operating is a private business in the USA. Why don't you read about it?
Overall it perhaps emerges that the USA is vastly better at catching and managing criminal than Honduras, that has almost 20x as much crime than the US but only few prisoners. However it seems that the USA also likes to put and keep people in prison rather than look for alternatives for dealing with their problems (drug rehabilitation for instance).
Some statistics that I find interesting: about 90% of federal prisoners are there for non-violent offenses; there are currently more African-American in prison in the USA right now than there were slaves in 1850; and 67% of ex-prisoners reoffend within 3 years: when they get out of jail they cannot find a job, they are ineligible for welfare, and so they are caught in an endless spiral.
If you remember your app name, and if you type nimbly, then Win+R works. My parents for instance are not very good at either, but they can search visually in a list. They prefer the start button.
Compatibility mode? Virtual machine?
You'd be surprised. Nowadays some artists are not afraid to code C+OpenGL
What do you mean "refuses to admit"? We are not talking about a level playing field here. Overall developer time is not quite the same between Windows & Linux, Photoshop and Gimp and MS-Office / Libre-Office, and commercial software can't survive if enough people don't pay up. People choose pirated software when they can but don't necessarily use it. Personally I have paid for all of these apps (or have my employer pay) yet I find that I'm enjoying Linux much more than Windows. An office suite is a nuisance (compared to LaTeX) I have to deal with, but I use Libre/Open as much as MS when I have to. I hardly use Photoshop or Gimp but I do use Premiere Pro.
Nonetheless I use to use Illustrator but Adobe left it behind somewhat so now I use Inkscape quite happily. I sometime have to use Matlab sometime to communicate with colleagues of for courses, but I much prefer Python, which is now almost as good even for the stuff Matlab is really good at (linear algebra).
In other words, choices are always good. Useful OSS is not going away. Eventually Gimp is going to be good enough for most things and people actually are using it. That is all that matters.
In fact the visible universe is significant bigger than 12 billion ly in radius, because the universe is expanding. The co-moving distance to the edge of the universe is more like 46 billion ly. See this wikipedia page for more details. If this is true, 3.6 billion ly is about 8% of the distance to the edge of the visible universe, which represents a partial volume of (3.6/46)^3 = 5 10^{-4} the size of the visible universe (or 0.05%).
So this is indeed very close by and rare.
Cheers
Oh yeah ? How about:
It is hard not to be dismissive of engineers when all they care about is building stupidly expensive gadgets that entrap everyone into thinking they are doing something useful. Software engineers are the worst: they build buggy software and toy apps that are nothing but smoke and mirrors instead of helping solve the real world's problem. What value do they really create?
See how it goes? With this attitude no one does anything useful: farmers, politician, army, police, doctors, scientists, marketing, finance, accounting people are all good for the bin. The saying goes that the economy needs economist the same way the weather needs meteorologists. This is true to some extent, but one forgets that one might be becoming better at economics or meteorology as time goes by and methods improve. This has and this will pay off.
Professor Dyson, you have been an actor and a witness of the huge expansion of hard sciences into everyday life. Thanks to scientific progress, particularly in the understanding of the basic physical laws, we have been able to improve our way of life almost beyond belief: energy, transport, and even exploration, now going beyond our planet all have been hugely transformed in the last 100 years.
However in the last few decades, our understanding of physics has become good enough for "most" things, and physics seems stuck. It looks like string theory is mostly sterile, with no useful or even testable prediction in spite of huge efforts, with nothing to really replace it. Also, the standard model is too good. We found the Higgs almost exactly where it was expected. On the one hand do not have a working theory of quantum gravity, and on the other we seem unable to produce a useful fusion reactor in spite of foreseeable dire needs in safe energy consumption. We already know that other efforts including solar, sea and wind energy will not be sufficient to guarantee our children a safe future.
At the same time we witness a resurgence of religious beliefs at home and abroad that is causing societal and international concerns. According to some recent polls, many people do no longer "believe" in science, or even maths. We are seemingly stuck in a hard recession, pollution is a problem finally hitting planet-wide proportions. This is not looking too good.
What do we do? How do we restore the public's confidence in progress? How do we go forward? These are huge questions and it would be nice to have a perspective on this from your point of view.
You are perfectly correct of course, however these are pesky engineering problems :-) At least it is not theoretically impossible.
I'm not really at liberty to describe the research culture at CSIRO in great detail, but it is, or at least was, as the articles say, very application-driven and short-term, external-earning motivated. This was only in one division, I cannot speak for the whole of the organization, however these stories seem to indicate that the problem is widespread.
I was at CSIRO between the mid-1990 to the mid 2000, and I have seen it progressively become a very tough place to do research. I was very very happy to leave. I'm not a top researcher by any stretch of the imagination, and I was never bullied, although I did experience unpleasant conflict. Ever since I've left (for academia) I've been more free to conduct my research the way I wanted it, I have found that it is indeed easier to find funding (so far). Looking for funding first and doing skunk research second is a sure way to kill imagination and generate stress, dissatisfaction and mistrust, not to mention poor results. Scientists are not necessarily good salespeople (too frank). Basically CSIRO was (and apparently still is in some places) in some ways a toxic place for scientists.
I hope it improves. CSIRO is nowhere near the top 10 rank it seeks to achieve, at least in the areas I'm familiar with, but there are still very good people working there.
Sure today's PC are insanely powerful for watching video, for running a word processor or a spreadsheet program. However they are just about good enough for real time speech processing and we are a long way from the processing power required for fast-enough computer vision, for instance. This would literallybe a game changer on your PC or your smartphone. Everybody was awed at first by the MS Kinect, but the reality is that the SDK is not fast enough and not precise enough to be all that useful beyond simple things.
Stronger AI still requires massive amount of computer power for machine learning. A lot of work is being done in this area. Look up the program of recent machine learning or computer vision conferences, and be awed.
And crappy Intel video.
The particles generated at the even horizon would be essentially photons of every wavelength, which are their own anti-particle. This is because photons can have any energy, which is not the case for massive particles. Only very light black holes (with very high energy output) would be able to create anything other than photons.
Although a common misconception, this is not correct. Please read the article. In the astronaut's frame of reference in freefall, nothing special happens at the event horizon (that is, if there is no firewall after all), and the even horizon is reached in finite time. For time to apparently accelerate like you describe, the astronaut would have to hover near the event horizon. They could not be in free fall.
That is a little schematic. Most codes running on HPC that I know are not trivial, typically expensive CFD calculations with 3D fluid / structure interaction, that may actually require quite a few thousand of hours of computation, e.g. blood flow simulations in the brain. Some scientists I know study physical geo-mechanical models that require them to analyze thousands of 2048^3 tomography data where they identify and simultaneously track tens of thousands of individual grains of sand as well as the relationship between thm, in order to find out what happens during terrain compression.
Access to the machines capable of handling such loads is competitive and in fact expensive. The technical staff managing these machines is competent. If you actually win time on these machines, they will help you make your software run well on them.
So yes, I'm perfectly willing to believe that some HPC machines are not very well managed but this does not mean that this is the case everywhere.