Slashdot Mirror

I'm sure it'll be perfect for this application:

http://en.wikipedia.org/wiki/Plenoptic_camera (a type of camera that can let you re-focus (and to a certain extent re-position) images after taking the shot. The problem is that it requires a LOT of resolution to produce acceptable images).

http://graphics.stanford.edu/papers/lfcamera/

http://www.youtube.com/watch?v=9H7yx31yslM&NR=1 (demo video from paper above)

http://www.youtube.com/watch?v=o3cyntPC2NU

Here's one built with a 250 MP Flatbet scanner:

http://www.youtube.com/watch?v=4O5fPoacF3Q&feature=related

My CS lecturer prepared a number of these for an intro CS course. I took the class back in '06-'07, but IIRC the videos were taken from Mac OS X. That said, it is pretty neat.

If there's some simple GPIO/parallel/etc. interface (could always rig up something USB based...), it'd be great for controlling lights and other appliances via SMS, IRC, etc.

Lest we forget everything Folding@Home has done..

The important difference between deduction on the one hand and induction and abduction on the other hand is that deductive reasoning holds necessarily, whereas inductive and abductive reasoning do not. This is because with deductive reasoning, so long as the premises are true, the conclusion must be true. With inductive and abductive reasoning, even though the premises are true, the conclusion may be still be false. A classic example of induction is that (p1) All swans that I've observed are white; (c1) All swans are white. Let's imagine that there are 1000 swans in the world, and I've observed 999 of them. Let's say that (p1) is true. (c1) may still be false. That one remaining swan could be some other colour than white. (Aside: Even if you said (c1') At least xx% of swans are white, it still does not hold necessarily since the total amount of swans is a contingent fact that you have no access to.) This is what is meant when we say that induction does not hold necessarily. On the other hand, if we say (p1) All swans that I've observed are white; (c2) This particular swan that I've observed is white. Now if (p1) is true, (c2) is necessarily true. The conclusion cannot be false. If this is unsatisfying, there's a much better explanation up on the Stanford Encyclopedia of Philosophy.

The sibling post pointed out that many fields within science and math try to solve this issue, but I don't think any of them actually try to justify induction. Most of the solutions that I know of creates techniques that if used will increase the reliability of induction. But you can never justify induction deductively, and any attempt to use deduction to justify induction must include induction in its premises (resulting in circular reasoning). Furthermore, deductive reasoning that includes results from inductive reasoning is logically invalid. This may be a bit abstract, so let me try to illustrate with an example.

There's a difference between the decimal number 2, the string "2", and the binary number |10|. Now, there's also the real abstract idea of [2] that we all have. Necessarily, the decimal number 2 and the binary number |10| both refer to this abstract [2]. However, when we write down the string "2", that does not necessarily refer to the abstract [2]. It just so happens to refer to the abstract [2] because we have thus defined it.

So let's say we have 3 formulas. (1) 2 + 2; (2) |10| + |10|; and (3) "2" + "2". Even though all 3 formulas represent [2] + [2], only formulas (1) and (2) represent [2] + [2] necessarily. What happens if you mix and match? So let's say you have (1a) 2 + |10|; (2a) 2 + "2"; and (3a) |10| + "2". In this case, only (1a) is necessarily equal to [2] + [2].

In this example with numbers, the decimal number 2 and the binary number |10| represent conclusions derived from deductive reasoning. They hold necessarily. The string "2" represents results from inductive and abductive reasoning, where the conclusion may so happen to hold, but it does not hold necessarily. Each time you introduce a result from inductive reasoning to a deductive formula, you are adding another element that does not hold necessarily. The problem is that necessity does not have degrees. Something is either necessary or not. Once you use inductive or abductive reasoning, your conclusion does not hold necessarily and is thus logically invalid (all logically valid moves must preserve truth necessarily).

Now as I said in my previous post, this is not a major concern for scientists or most other people. While induction is not a logically valid move (since it does not necessarily preserve truth), it is a move that is reliable enough for practical (and even theoretical) purposes. For science, as previously mentioned (and as you've mentioned), there are many techniques to increase the reliability of induction. Everyone uses induction a lot in their every day lives. We use it all the time. It's impossible to function without induction. For

Folding@Home

Our goal: to understand protein folding, misfolding, and related diseases

Protein folding is linked to disease, such as Alzheimer's, ALS, Huntington's, Parkinson's disease, and many Cancers

that's okay! :)

forget about Foldit! Just download Folding@Home and let your CPU/GPU do it for you!

http://folding.stanford.edu/

Same goes for optical interconnect to memory: the flood may be Biblical when it arrives

But it won't be - the system is fundamentally limited by all of the rest of the components. A top end front-side bus can already push 80Gb; scaling that upto the 400Gbit that this optical link promises will probably be practical within a few years, but the latency of encoding and decoding a laser signal and pushing it over several meters is going to be a killer for computational applications. It will be great for USBX, and for high end networking it will challenge Infiniband (which currently tops out at around 300Gb). Infiniband is already used for networking high-performance computational clusters, but nobody is using it for the CPU to memory bus because of the high latency. Even with high bandwidth, computation still has to be carried out on the data, and so it still makes sense to put the data and processor as close together as possible.

In the last decade there were many research papers proposing that co-processors would be placed on DRAM cards, or Embedded DRAM would allow CPU and processors to be fabricated on a single die (e.g. 1, 2). But if you have a processor and DRAM connected to similar units via an optical interconnnect, guess what - the architecture begins to look awfully similar to a regular network with optical ethernet. So, it looks likely that this will be just another incremental improvement in architecture rather than the radical shift that TFA envisions.

It's nice to see someone involved in the industry "getting it".

For me, the deal-breaker with current 3D movies is the simple fact that I have to look at whatever is in focus to avoid being distracted or getting a headache. I like the freedom to explore the scene, and 3D takes that away from me.

That being said, there are some developments that could be a bit of a game changer for 3D. Where I'd like 3D right now is FPS games. Combined with a webcam and eye-tracking the effect could be configured to put whatever I look at in focus, much like the eye would. I'm guessing this could do a lot for immersion. (I'm not aware of anyone having done this yet.)

With movies it's much trickier. Depth of field is something you decide when shooting. For now:
http://graphics.stanford.edu/papers/lfcamera/

Using a flexible depth of field technique it could perhaps be possible to remove the restriction to have to decide the depth of field when shooting. The viewer could decide that part when watching, using the same kind of eye tracking I described above. Granted, this would work only if there is a single viewer, or with a technology to feed individual video feeds to all viewers. This also would strip the director of a creative technique, and in that sense would lessen the cinematic experience.

I agree completely with you about your thoughts on the human visual system. The brain in effect makes a 2D scene with 3D ques into 3D. For now, as far as I'm concerned, 3D is a gimmick in current movies. A cool gimmick, (and technically very well executed in Avatar,) but a gimmick none the less.

Does that 0.06 USD/kWh include the costs of mining & associated health costs, disposal of waste & associated health costs, cost of security to protect plants from terrorists? Point being: unless one can scrutinize the details, there's no way to trust such numbers.

The French people have come to terms with nuclear energy, which apparently has worked well enough for them there. US results have been less spectacular; many unplanned excursions resulted in lies and coverups that make it difficult for millions of Americans to accept claims for nuclear. The cost of new plants has skyrocketed into the $10-20 billion range. They remain uninsurable. Many make claims for new nuclear technologies, but there are few if any working demonstrations.

The costs of solar may be based on online technology, rather than one of the dozens of technologies in the pipeline. See for example this one: http://news.stanford.edu/news/2010/august/new-solar-method-080210.html

Storage is not a problem, there are many ways of doing storage. Britain for example has a huge facility that pumps water uphill into a lake 2000 feet above during the day; it is capable of powering a city for hours and can come online within a few minutes at any time. Nuclear is currently pumping the water; it could just as easily be solar thermal or wind. As someone else has pointed out: fossil-fuel energy saved during the day can be used at night. And, in the US, there is still a lot of waste that can be ameliorated by conservation measures, further reducing energy needs.

With wind energy quickly becoming a clear win around the world, and growing cheaper all the time (already competitive with the French rates your quoting) it's quite clear that the time for nuclear plants is drawing to an end. The dangers of nuclear proliferation are never far away so long as nuclear power continues to be acceptable. For cost, safety, and obsolescence reasons, it would be a poor idea to invest in such plants.

Cue Basil Fawlty-esque voice from that classic episode:
"Ohhhhhhhh, it's Obama's fault is it. Here's me thinking it might be Bush's fault for starting an illegal war, or Cheney's fault for encouraging lying about the presence of WMD, or Powell's fault for playing the willing patsy at the UN, but nooooo, it's Obama's because he's Muslim and black and won the election when I wanted the Republicans to be in power for longer. Well, he'll have to be punished then: who's a naughty boy, Barack"

http://www-cs-students.stanford.edu/~clee/fawltytowers.html

The progress report from March 2010, available at http://gcep.stanford.edu/research/factsheets/petesolor_results.html, provides a more detailed and understandable summary of what they are doing

In films that don't pretend to be 3D there's no conflict between the parallax and my focal distance, so I never have any reason to attempt to focus at some other depth. In fact, the limited field of focus is often used to show differing distances, to help make up for the lack of real depth information.

Plus it's *not* impossible to capture a re-focusable image -- you just need to capture the entire light field as opposed to the 2D projection of the light field captured by traditional photography. While not in common use, such devices actually exist: http://graphics.stanford.edu/papers/lfcamera/

Yes! I spent hours playing that. It looks like this is a download of the game: http://www.stanford.edu/~cammat/HOVER/index.html

I'd also like to thank the windows 95 disc for introducing me to Weezer.

It's worth noting that page nine of the Frankencamera team's paper mentions the work of Joshi et al when it discusses deblurring pictures. Neel Joshi was the lead researcher from the article we are discussing.

Step back! This is a job for Frankencamera. Run it on your Nokia N900 today.

OTOH having that Arduino board and a mess of wires attached to your camera does score you a lot more geek cred than photographing using an plain old mobile phone.

/greger

No one is doubting Global Warming.

That's simply not true. There's a large contingency of folks who are outright denying even the temp rises. They're typically the mindless followers of Beck & Limbaugh.

By "solar weather theory" are you referring to the false arguments that AGW is caused by cosmic rays and/or temps are increasing on other planets? If so, no problem. Here's 34 different scientific papers that refute each aspect of them. :)

So, you ready to change your business model now?

If you read these articles, you'll see that modern IFR reactors can be started on the existing nuclear WASTE from our current reactors, and need only a milkcrate-sized chunk of essentially unrefined uranium metal per month to continue operating ad infinitum.

And of course, just because they were writing code doesn't mean they were programmers by trade. I've rewritten a decent amount of old F77 Fortran code into C (C99), where the original Fortran was written by mathematicians with enough knowledge to directly translate their equations into code, but not to do it well. This code was running on vector processing machines, and Fortran was designed to take advantage of that in a way the C never was, so the Fortran code should have had the advantage, but they made such elementary algorithmic mistakes that those benefits were wasted.

For example, in a trebly nested loop (the most touched code in the program), they were counting the bits set in an integer (that was most of the work in the innermost loop actually). They used the naive approach, as opposed to one of the many, many, many portable, efficient means of doing so, and as opposed to using a provided header that defined these sorts of operations as a macro, such that it defaulted to a performant portable technique, but used the most efficient method available for a number of known processors (many of them have an assembly instruction that does what these portable algorithms do, but requires a single instruction, instead of cluttering the line cache with the step by step instructions). The mathematics were beyond my skills at the time, but it was trivial to find and fix all the "mathematically correct, but practically inefficient" sections of the code without needing to understand it all in detail, and the end result in C ran on the same machine in less than 10% of the time, and over half of that savings came simply from swapping in the efficient bit counting macro.

The end result was not only significantly faster, but was half the length (since macros did a lot of work that had been hand-coded poorly), and easier to read (even before I commented it). Though frankly, anything is easier to read than F77, with its horrible label based loops that make no distinction between the end of the loop and a C-style "continue" statement; it was my first experience with Fortran, and most of the time converting to C was spent figuring out that the fifty "continues" could be converted to C as three loops with a few break and continue statements scattered about (and far less than the original program used; in a nested if statement where each branch triggered a return to the top of the loop, it would be coded individually for each case, rather than putting it in the outer if once, even though it was functionally equivalent).

In summary, don't let your professional mathematicians implement their own equations.

And of course, just because they were writing code doesn't mean they were programmers by trade. I've rewritten a decent amount of old F77 Fortran code into C (C99), where the original Fortran was written by mathematicians with enough knowledge to directly translate their equations into code, but not to do it well. This code was running on vector processing machines, and Fortran was designed to take advantage of that in a way the C never was, so the Fortran code should have had the advantage, but they made such elementary algorithmic mistakes that those benefits were wasted.

For example, in a trebly nested loop (the most touched code in the program), they were counting the bits set in an integer (that was most of the work in the innermost loop actually). They used the naive approach, as opposed to one of the many, many, many portable, efficient means of doing so, and as opposed to using a provided header that defined these sorts of operations as a macro, such that it defaulted to a performant portable technique, but used the most efficient method available for a number of known processors (many of them have an assembly instruction that does what these portable algorithms do, but requires a single instruction, instead of cluttering the line cache with the step by step instructions). The mathematics were beyond my skills at the time, but it was trivial to find and fix all the "mathematically correct, but practically inefficient" sections of the code without needing to understand it all in detail, and the end result in C ran on the same machine in less than 10% of the time, and over half of that savings came simply from swapping in the efficient bit counting macro.

The end result was not only significantly faster, but was half the length (since macros did a lot of work that had been hand-coded poorly), and easier to read (even before I commented it). Though frankly, anything is easier to read than F77, with its horrible label based loops that make no distinction between the end of the loop and a C-style "continue" statement; it was my first experience with Fortran, and most of the time converting to C was spent figuring out that the fifty "continues" could be converted to C as three loops with a few break and continue statements scattered about (and far less than the original program used; in a nested if statement where each branch triggered a return to the top of the loop, it would be coded individually for each case, rather than putting it in the outer if once, even though it was functionally equivalent).

In summary, don't let your professional mathematicians implement their own equations.

And of course, just because they were writing code doesn't mean they were programmers by trade. I've rewritten a decent amount of old F77 Fortran code into C (C99), where the original Fortran was written by mathematicians with enough knowledge to directly translate their equations into code, but not to do it well. This code was running on vector processing machines, and Fortran was designed to take advantage of that in a way the C never was, so the Fortran code should have had the advantage, but they made such elementary algorithmic mistakes that those benefits were wasted.

For example, in a trebly nested loop (the most touched code in the program), they were counting the bits set in an integer (that was most of the work in the innermost loop actually). They used the naive approach, as opposed to one of the many, many, many portable, efficient means of doing so, and as opposed to using a provided header that defined these sorts of operations as a macro, such that it defaulted to a performant portable technique, but used the most efficient method available for a number of known processors (many of them have an assembly instruction that does what these portable algorithms do, but requires a single instruction, instead of cluttering the line cache with the step by step instructions). The mathematics were beyond my skills at the time, but it was trivial to find and fix all the "mathematically correct, but practically inefficient" sections of the code without needing to understand it all in detail, and the end result in C ran on the same machine in less than 10% of the time, and over half of that savings came simply from swapping in the efficient bit counting macro.

The end result was not only significantly faster, but was half the length (since macros did a lot of work that had been hand-coded poorly), and easier to read (even before I commented it). Though frankly, anything is easier to read than F77, with its horrible label based loops that make no distinction between the end of the loop and a C-style "continue" statement; it was my first experience with Fortran, and most of the time converting to C was spent figuring out that the fifty "continues" could be converted to C as three loops with a few break and continue statements scattered about (and far less than the original program used; in a nested if statement where each branch triggered a return to the top of the loop, it would be coded individually for each case, rather than putting it in the outer if once, even though it was functionally equivalent).

In summary, don't let your professional mathematicians implement their own equations.

And of course, just because they were writing code doesn't mean they were programmers by trade. I've rewritten a decent amount of old F77 Fortran code into C (C99), where the original Fortran was written by mathematicians with enough knowledge to directly translate their equations into code, but not to do it well. This code was running on vector processing machines, and Fortran was designed to take advantage of that in a way the C never was, so the Fortran code should have had the advantage, but they made such elementary algorithmic mistakes that those benefits were wasted.

For example, in a trebly nested loop (the most touched code in the program), they were counting the bits set in an integer (that was most of the work in the innermost loop actually). They used the naive approach, as opposed to one of the many, many, many portable, efficient means of doing so, and as opposed to using a provided header that defined these sorts of operations as a macro, such that it defaulted to a performant portable technique, but used the most efficient method available for a number of known processors (many of them have an assembly instruction that does what these portable algorithms do, but requires a single instruction, instead of cluttering the line cache with the step by step instructions). The mathematics were beyond my skills at the time, but it was trivial to find and fix all the "mathematically correct, but practically inefficient" sections of the code without needing to understand it all in detail, and the end result in C ran on the same machine in less than 10% of the time, and over half of that savings came simply from swapping in the efficient bit counting macro.

The end result was not only significantly faster, but was half the length (since macros did a lot of work that had been hand-coded poorly), and easier to read (even before I commented it). Though frankly, anything is easier to read than F77, with its horrible label based loops that make no distinction between the end of the loop and a C-style "continue" statement; it was my first experience with Fortran, and most of the time converting to C was spent figuring out that the fifty "continues" could be converted to C as three loops with a few break and continue statements scattered about (and far less than the original program used; in a nested if statement where each branch triggered a return to the top of the loop, it would be coded individually for each case, rather than putting it in the outer if once, even though it was functionally equivalent).

In summary, don't let your professional mathematicians implement their own equations.

Everyone would want the highest power possible. The law would have the opposite affect of its promoters.

You're probably right. Don't have the source available at the moment, but I remember reading that when fast food joints in some location was forced to put calories on their menu, the average number of calories per meal purchased went UP.

Ah - Stanford study of Starbucks: 6% Fewer calories

However - Yale and NYU study of McD's, Wendys, BK, and KFC:
"they found that people had, in fact, ordered slightly more calories than the typical customer had before the labeling law went into effect, in July 2008."

The haters are out this morning:

"... Shrink, I want to kill. I mean, I wanna, I
wanna kill. Kill. I wanna, I wanna see, I wanna see blood and gore and
guts and veins in my teeth. Eat dead burnt bodies. I mean kill, Kill,
KILL, KILL." And I started jumpin up and down yelling, "KILL, KILL,""

http://www.arlo.net/resources/lyrics/alices.shtml :-)

RISC blows CISC away: [skip or walk]

http://en.wikipedia.org/wiki/Reduced_instruction_set_computing#RISC_design_philosophy

- so much so, that they still bolt it on CISC [with some success]

[Don't bother with the subheading "Diminishing benefits", it's BS, look at IBM's POWER]

RISC vs. CISC:

http://www-cs-faculty.stanford.edu/~eroberts/courses/soco/projects/2000-01/risc/risccisc/

Our G5 x2 2.5 is soon to be a companion to our Xserve x2 1.33 [redundant DNS].
Just add:

Swift Data 200:

http://www.transintl.com/store/category.cfm?Category=2490

Inside your Power Mac G5:

http://support.apple.com/kb/HT1305

"catastrophic coolant leak":

http://www.xlr8yourmac.com/systems/G5_CoolantLeak_Repair/G5_CoolantLeak_Repair_p1.html

This has been said elsewhere in this thread, the real breakthrough here is due to compressed sensing, but here are some extra information:

1- Compressed sensing basically used the idea that it is not necessary to sample an image (or a projection in this case) everywhere because natural data is fairly redundant. This is why you can capture a 10 Mpixel image in a digital camera and have it compressed to a 2 Mbyte JPEG file without losing much visible information. Compressed sensing basically does the compression *before* the sampling and not after. Researchers at Rice University for instance built a working, one-pixel camera using this brilliant principle.

2- Compressed (or compressive) sensing was proposed by Emmanuel Candes and Terence Tao respectively at Stanford and UCLA. Tao is a recent Fields medalist. I recommend reading his blog if you like mathematics.

3- This field is really less than 10 years old, it has completely turned on its head classical ideas about sampling-limited signal processing (Nyquist, Shannon, etc). It is a brilliant combination of signal, image processing and recent advances in combinatorial and convex optimization.

4- However this is only the beginning. Because compression happens before sampling, you need to make so-called sparsity assumptions about the signal ; in other words you need to know a great deal about what you are going to try to image. In interventional therapy, precise imaging of the patient is made beforehand in a classical way (CT or MRI), and this kind of technique is only used to make fine adjustments as therapy is ongoing. This is extremely useful and safe because of lower radiation output and because the physicians know what to expect.

5- Here the GPU is useful because it makes the processing fast enough to actually be used. It is an essential brick in the application, but of course not in the theory.

Best.

The best way to make a product people want it is to have them make a prototype themselves and submit it. However most things cannot be prototyped but a regular Joe. But tshirts can.

Perhaps Threadless has stolen designs in the past. Can you cite a non-anecdotal instance of this theft by them? Searches for "Threadless stole my design" return zero results. THe phrase ""Threadless stole my design" has nothing about this (but is does return results about people who steal from Threadless). And searches for "Threadless stole my idea" only returned one deliciously ironic result. If it's happening often (or at all) nobody is hurt enough to talk about it online.

And are aware of how easily dismissible your claim of having your punchline 'stolen' is? Even if you had a million-dollar-an-hour law firm behind you, the claim would be laughed out of court. A short turn of phrase or other idiom, cannot be copywrited (source). It can be trademarked, but that only applies if the phrase is used to sell a product or promote an organization. Your punchline 'I hate your children', and the gag of hating 'trendy' children's names, was not at all new...even in 2004. What's more, the Gawker shirt didn't even take the exact phrase...and they added artwork. How long after you posted you strip did it appear on Gawker. From what I can tell it was years later (source). Do you really think someone stumbled across your strip and submitted it to Gawker? That reasoning seems really thin. Unless, of course, you have some kind of source...

at the end of the day we still need email clients in our pockets

I definitely don't want or need (or have) an email client in my pocket. I hope to one day be as successful as Dr. Knuth, so I won't need any email clients at all.

USENET is very resistant to censorship. The comment "The network interprets censorship as damage and routes around it" was actually made about USENET, not the Internet, in response to an incident at Stanford where the administration tried to censor "rec.humor.funny". As long as someone, somewhere on campus had a dialup USENET feed, the missing messages would be recovered and put back into on-campus servers.

There was something to be said for that, instead of a huge number of business-controlled forum systems.

My university course on databases used the text book A First Course in Database Systems by Jeff Ullman and Jennifer Widom. I rather enjoyed the book, and plan to have it above my desk in case any sort of database design or maintenance project comes up for me. The book's page is here; links to purchase are at the bottom.

Aside from the fact that it uses Google Voice, you can make a very open system for controlling lights/power/etc. from a cell phone/email/IRC/whatever.

C++ can be faster than C... This is an old one, which proves the point...

Remember, C++ is not just OO - that's one of the paradigms it supports, but not the only one.

Otherwise, back it up and cancel your bank account and start paying for everything by cash. (*)

After reading this FAQ item on Donald Knuth's webpage, I'm beginning to wonder if it's not whether my bank account will be owned, but when...

"Where? Everywhere. I don't know what enterprises you work in, but I've been in enterprise development for 35 years and I've never seen such a sea change..."

Yeah, yeah. I know. And I'm absolutely not interested. The world of enterprise is a world of stale old tenchnologies, it's a world of hopeless legacy. Java gave it a brief time of excitement, but now it'll fall back on being soul-crushing boredom. Only now with Java instead of COBOL.

While small and nimble companies will use good tools (.NET, Python, etc.) to leapfrog 'enterprisey' companies. I'm seeing this already.

"Right. We should rely on your word rather than actual statistics."

Which statistics? Number of Google searches? Don't make me laugh.

"What? Java 1.6.0 was released in 2006. We're now up to release 20. By my count that's better than one major release per quarter."

I'm speaking about major updates. Java 1.1 was released in 1997 with anonymous and inner class support. Java 5 was released in 2004 with generics (7 years). Java 7 is on track to be released in 2011 (7 years).

"Sorry, not even close and most of what is available is closed source."

Can you provide examples of Java projects absent in .NET?

"Having to convince management to pony up money for third party .Net libraries all the time is tiring, which is why, if given the choice, I develop in java where I can get better quality and better tested libraries for no charge." .NET ecosphere is just different. A lot of projects are closed source, but are usually fairly cheap. Also, MS is quite friendly for small companies - it's possible to get everything required for development almost for free (using their BizSpark program).

"Guess it depends on your cerebral capacity. BTW, you might be interested to know that apparently Scala 2.8 will support LINQ."

Nope, it won't. I observe Scala development quite closely, and there's no infrastructure for LINQ. There's an interesting project http://ppl.stanford.edu/wiki/images/8/8f/Porting_LINQ_to_Scala.pdf but it is not open (I asked its author). There's also ScalaQL and other projects, but they are inadequate.

Right now, I'm using QueryDSL as the most practical approach.

What is a concern is that those guys are now dead (Alan Turing), retired (Don Knuth) or in very old age (lots of 60s people who worked on C-like languages), or not presentable. For those alive, their contributions were made at young age, so we are missing something because our twenty-somethings are not earning the same respect, or somehow achievements magically stopped mattering.

In comparably important to human development as science, we have the political and business world. There, there are influential and WELL-known figures on TV and media articles providing quotes, as well as market-moving long-term predictions (like certain American financial advisers.)

Besides Bill Gates, there is no other known figure that mainstream people recognize (Steve Jobs and Linus Torvalds are names that carry no recognition in the third world countries, for instance.) Our equivalently bright, twenty-something uber-gods are busy running their successful Facebooks, Googles or whatever. These people aren't interested to know about what Linus thinks. Matter of fact, I cannot name a single world-famous doctor alive today. Maybe medicine and IT jobs where we're regarded as "plumbers" and fixer-uppers, rather than geniuses, have a downside when it comes to mainstream fame.

I suspect that individual recognition affordable in other human endeavors where respect and breakthroughs are encouraged, became forbidden in favor of abstracting your achievement to uncopiable patents attached to a big name like IBM.

You think Knuth uses Emacs? Heresy!

From http://www-cs-faculty.stanford.edu/~knuth/programs.html:

"The emacs-oriented desktop layouts I use on my home computer to write books..."

Looks like you just copied a definition from a dictionary without understanding the question. No Points.

Morals are basic principles of right and wrong-- "it is wrong to kill", "it is right to show compassion" and so on.

Ethics are methods of reasoning about those morals. "If it is wrong to kill, what actions should I take in order to live my life according to this principle?" "Is it possible to apply this moral principle in a consistent manner, and if not, how should my moral precepts change to allow for living a consistently good life?"

See, for example Kantian ethics

As for sampling front ends: they don't really cut it due to parasitics -- they can be fast, but they aren't really accurate -- 12 bits of performance would be pushing it I think. Surely if you would be very clever and do, say, FEM of an integrated sampling bridge, and would characterize well all the parasitics and how they can be balanced out, it could probably be done. But using off-the-shelf parts with nothing exotic: I will believe it when I see it done.

I am not sure they even got 8 bits worth of performance. Once you get into that realm, sample and holds are more like sample and slow downs. :)

The only practical way I know of is with a fast variable gain amp to blank overdrive -- this can readily be done, and this way you can test settling of 18 bit DAC's, and it's really no big deal once you try it and get it working. It doesn't cost much, either.

I'm playing with a blanking front-end that uses JW's variable transconductance amp approach, gated with fast comparators, and so far I've got it to recover within 100ns to 12 bits -- that's on the first try, on a real crappy breadboard. It's still far away from 10ns one expects the recovery to be in a 100MHz scope, but I should be able to cut it down to 25ns or so without doing anything extraordinary. And I don't really have all that much analog tinkering experience. Surely someone who knows what they are doing could get it to work way better and cheaper.

Hehe. You read that application note also?

Non-saturated switching is always going to have an advantage over saturated switching but unless you can control the parasitics you are doomed. The big improvement now has been monolithic variable gain amplifiers with good DC and settling performance. I suspect they got that by using the same techniques which were applied to operation amplifiers to prevent thermal gradients generated by the other stages from unbalancing the input stage differential pair. You could probably do something similar using dual and quad matched transistor arrays to build a Gilbert cell multiplier but it would be complicated and likely not as fast and I suspect parasitics would be a problem. Does anybody make or sell cross coupled quad transistor arrays or something equivalent? Linear Integrated Systems sells matched duals and they might be enough. If you use discrete duals then you do not have to worry about thermal feedback from other stages anyway.

As far as construction goes, dead bug construction over a ground plane is probably going to be mandatory for good performance. If you can go the printed circuit board route, you would still need a good ground plane. I tend to use double sided copper board for a ground plane and shielding and will build an entire small enclosure if necessary. I have even built a couple of rugged helical resonators that way. Kynar insulated wire wrap wire (the whole color table is available) or just enameled copper works well for hook up wire if you include strain reliefs. If you have a precision shear (or a good tabletop sander and time), you can even make little microstrip transmission lines to be soldered or glued down onto the ground plane.

http://qwiki.stanford.edu/wiki/High_Frequency_Op_Amp_Circuits

For an oscilloscope to be a truely universal instrument, it should have a minimum number of caveats. Poor signal fidelity (measured in single % - gimme a break), poor overload recovery, no antialiasing protection on many DSOs, ridiculous trigger holdoff times (orders of magnitude worse than on a $100 tek 7K mainframe from the 70s) -- those are the gripes I have with current technology.

Luckily the stuff that used to be out of reach financially is now either affordable or free: you can easily get a dev board with fast FPGA with multipliers on it, 64MB of DDR2 and a USB 2.0 connection for $200 IIRC. The software to do logic de

That's not just a rumor. Knuth explains it here.

He has announced it, and he has paid. Many times. For some reason people rarely cached his checks but stuck them in frames instead. Since pictures of these ended up on the web, Knuth had to stop sending out checks. These days you can get a check from the Bank of San Seriffe instead.

With effective computer languages and compiler tools this should not be an issue.

Memory per CPU is a bad way to think of it thats quite ill defined, e.g. a modern CPU has a few MB of L2 cache, working out at around 256-512k per core, on some architectures there's now L3 cache before you get to the relatively slow DRAM. Look at architectures like Picochip or Imagine, which are embedded architectures designed for Signal processing, one is a 2D grid array or processors and the other is a SIMD cluster of VLIW processors.

Given that Knuth doesn't have an email account, I'm betting it's one of these:

- Knuth now has his secretary sending tweets for him.
- Knuth got a Facebook account. It's literally a book of faces.
- Knuth has convinced his secretary to view the most popular YouTube videos on a daily basis, and then act them out. (Her kitten impressions are awesome.)

But seriously, I'm hoping that he's releasing his works under creative commons. Bibles are free in hotels, but if you want the bible of programming and algorithms, you have to pay $70 per volume!

just fucking google it
http://www-cs-faculty.stanford.edu/~knuth/mmix.html

I get the sense that this is a tongue-in-cheek announcement? It's 2010, so maybe it'll be the MMX machine?

Let's see. Wednesday: July 7, 2010 = 7-7-7DA. 20th anniversary of TeX. Hmm. I can't figure it out, but I'd put my money on an elegant technical curiosity which doubles as elaborate pun and extended joke, kind of like MMIX.

Found it:

From SAIL "autobiography" (here):

I got proper air conditioning a short time later, but unfortunately developed a bad case of hiccups that struck regularly at 12 second intervals. My assistants spent a number of days trying to find the cause of this mysterious malady without success. As luck would have it, somebody brought a portable radio into my room one day and noticed that it was emitting a "Bzz" at regular intervals -- in fact, at the same moment that I hicced. Further investigation revealed that the high-powered air defense radar atop Mt. Umunhum, about 20 miles away, was causing some of my transistors to act as radio receivers. We solved this problem by improving my grounding.

I've been using Google Voice as an SMS-email relay to remotely control appliances. Glad more people will have the option of doing the same!

It is a bundled disable-able plugin, you retarded fuck. Ditto with Chrome's included Flash support. Do you idiots read anything beyond the 8-word article title?

Let's see if I can make you feel even more stupid. From TFA:

• To further protect users, PDF functionality will be contained within the security “sandbox” Chrome uses for web page rendering.

Don't be too quick to disregard this as a joke, since breeder reactors have been outlawed for some time here, it not an unreasonable suggestion. Of course, as far as breeders are concerned, past political boondoggles are certainly a major issue.

http://www.google.fr/search?&q=bp+bribes+royalties

3rd result: http://iis-db.stanford.edu/pubs/20814/Corruption_transparency_Angola1_No36.pdf

Start reading on page 7

Unfortunately, he no longer gives out reward checks for finding bugs in his texts. This seems to be mostly because proud bug-finders inevitably post images of the checks online, which of course, contain Knuth's bank account numbers. More discussion here.

Some context that presaged this trend independent of the current depression in the larger economy:

http://www-cs-faculty.stanford.edu/~knuth/joalet.pdf

http://stlq.info/2004/05/commentary_the_crisis_in_schol.html