I'm a bit of a skeptic but In your list I would like to see (as necessary features of a modern file system): - online defrag (you don't take the volume offline to do a defrag) - online integrity checking - online self-healing. - a pluggable infrastructure for various filters: antivirus, encryption, compression, folder redirection, etc.
I would view them as unnecessary: - snapshots: yes, they can be implemented in the file system (or below) but note that a software implementation will never match a hardware implementation at storage array level. Most storage arrays today support high-performant snapshots at the LUN/disk level. You can have software snapshots also, of course but perf will not be very good. Even Sun storage arrays have snapshots. - Auto-striping: stripping, spanning, mirroring, RAID5, etc are just some features that a storage array would provide. Anyway, these are implemented usually below the file system level, usually at volume level (for software implementations) or LUN level (for hardware implementations)
In my case it took me about one hour of C# programming and head scratching to found out the fifth number. The puzzle is not very hard but it requires a little bit of unorthodox thinking. Don't start with classical extrapolation techniques...
If you want hard math puzzles, check the "Ponder this" site from IBM - there are several beautiful puzzles there: http://www.research.ibm.com/ponder/
Re:We've been seeing a lot of this "safe" nukes st
on
Port-A-Nuke
·
· Score: 1
Seriously now, there is something interesting about nuclear reactors - their generated power is not really related with their size. So there is no such thing as "nuclear mini-reactor". You have almost the same quantity of nuclear fuel in each of them.
The reason is simple: the mass of fissionable material should be just a little above the critical mass (which is minimum 13 Kg for Pu-239 and slightly larger for U-235). The critical mass of course depends on the core design but you can't really extend that beyond a certain safety limit, and you also cannot keep that under the criticality limit (otherwise you won't have any self-sustainable reaction; only a tiny percentage of the fuel ends up consumed; the vast majority of the fuel is present there just to ensure criticality).
You need to be just above the critical mass in a small interval mainly for control reasons. The neutrons are produced in two class of reactions: (1) neutrons that are produced by the main fission reaction (which generates around 80% of the total quantity of neutrons) and (2) neutrons produced by secondary fission reactions from the extremely radioactive fission by-products from the first reaction. The production of these neutrons can be therefore considered as "delayed". These delayed neutrons are essential for reactor control - here is why:
At the end, fission is a chain reaction where you get almost 3 neutrons per fission (assuming U-235), from which you loose 2 and the remaining neutron is used to initiate the next fission. The average time for this generated neutron to participate in the next fission is in the order of milliseconds. In other words, every millisecond or so a new chain reaction produces a new generation of neutrons of class (1).
Let's forget a second about the second generation: if you are above the critical mass, over a very short period of time you get more and more neutrons which will build up in tens or hundreds of milliseconds way above the safety limit. For example let's say that you are 1.001 above the limit. This means that in the first generation you have a 0.001 excess, in the second generation 3*0.001, etc. After N generations you have 3^N * 0.001 which can be way beyond 1 in less than one second. The conclusion is that it is impossible to control a reactor by inserting the control neutron-absorber rods in the core - since again it takes less than one second to double up the neutrons.
Now, since you have these neutrons from the second generation, it is possible to control the reactor without quick build-up only if you manage that those 80% neutrons produced in the first class to be below criticality (otherwise you will have the quick build-up phenomenon above), but the total quantity of neutrons to be above. This way, you will ensure a delayed reaction time of a couple of minutes in which you can safely control the reactor.
The conclusion is that it doesn't really matter how you build the reactor - you will have the almost the same quantity of nuclear fuel there. There are no "micro/mini" reactors.
>>> Cryptography will die when the last human draws its breath.
You meant, cryptography will die when the third to last human draws its breath, correct? When you have two humans left there is no need for cryptography.
1) Nobody considers now breeder reactors as a good alternative for nuclear power generation.
There are several reasosn, assuming that you want a U-238 based breeder reactor: a) One reason is that is very hard to intensively reprocess the spent fuel. This requires huge installations that has to dissolve the spent fuel after a couple of weeks/months in nitric acid or equivalent. This is pretty challenging, even for US who has the experience of dozens of years in U-238/Pu-239 separation. The main problem is that you have all these dangerous byproducts like Cs-137, Sr-90, which appear in small quantities but still are pretty reactive properties (a medium half-life of a couple of years, etc). They are still pretty dangerous and hard to isolate chemically.
b) Another issue: the Pu-239 becomes a very dangerous material that fall off in the hands of terrorists. You need around 15-30 Kg of Pu-239 to build something that would detonate. This is a very small quantity. Today, there is a lot of Pu-239 that has to be either consumed or buried such that it won't fall in the wrong hands.
c) And finally, nuclear reactors are not that efficient. For one thing, the amount of generated power doesn't really depend on nuclear reactor size. Mainly because you have to be in the region between the critical and supercritical mass for a certain configuration. This also means that the main challenge in designing a reactor is optimizing the heat transfer process. Smaller reactors are better.
Second, you get about 1% energy from the fission fuel in one pass; there is no way to get 10%. And, reusing the other 99% for anything is pretty hard.
Well, probably there is still some room for some message, like "Alien XXX please fix your software" or something like this:-)
Some advanced viruses such as bacteriophages of the family Leviviridae, can reach almost 300 Kbp (kilobase pairs) which should be enough to store some additional data.
Somewhere I read that some flu viruses might be of alien form. Indeed, they seem to be the ideal organisms built for space travel. So why don't we search for alien messages in their DNA too?
First, you have come up with a general plan. - What exactly do you want to test - and to what degree of certainity - What are your priorities - here I would say that the accuracy must be less than 1% - What are your resources (i.e. how many testers, their qualifications) - What is the timeline - What is the estimated cost (and ultimately, what is the real profit from this stuff?)
Then you start to develop a small test plan. First, you need a precise behavioral model of your system. This means that if you have 1000 terminals, you have to define a model which says how the votes "add up" together in the main system.
Next step you design a real, physical interface to enter data in the system. This needs to be completely automated. Yes, you will need to use rubber fingers (or something equivalent) hooked up to some electromagnetic devices that triggers a finger to be "pushed" at a certain time. Then you define a module that would accept, say, RS232 commands and in response will push a certain finger. This is not very complicated and not expensive either. Then you just need one PC to send these commands over a set of serial cables to each "voting monkey" device.
If you have the ability to run the touch-screen software in a simulator test the software separately, using some scripts that simulate the "touch" commands directly in the simulator. This would be a cheaper solution although you will still need to do some manual/semi-automated end-to-end testing.
Finally, you have to one or several machines that will: - Run the main test programs - Run the "monkey scripts" that ultimately will cause these rubber fingers to go in action - Gather various activity logs from the affected systems - Correlate the data gathered from the voting systems with the actual votes.
The main difficulty would be writing the whole test package, depending on how complicated your test needs to be. You can decide how much time you need to spend on functionality testing vs. longhaul tests etc.
Anyway, what I wrote is a very high-level overview of a plan. The main point is that developing an end-to-end, 100% automated test for a system containing things like touch screens is not rocket science as long as you know what you are doing and you hired the right people.
It is actually better for a different reason. Each of these states evolves in parallel according to the Schrodinger equation for that particular quantum system. In other words you have some sort of parallelism.
>>> I don't see how this is 100% simultaneous. Let's say one pair breaks down.
It is simultaneous simply because as long as you don't perturb it, a pair of entangled particles constitute a single quantum system, a single "particle" if you want. As soon as you touch one of them, the pair is broken and now suddenly you can imagine having two separate particles popping up in space at certain distance between them.
>>> We call that event A in the observer X's IRF. Then another pair breaks down - event B. Let's use A as our reference point. An observer Y in a different IRF will say that event B did not happen at the same time oberver X said it did. There will be a difference in the time elapsed.
No. The state vector collapse is not an observable event in itself. Only the measurement is an observable event. All you can do (as observer X) is to measure both particles and hope/presume that the collapse already hapened - but you have no 100% guarantee that of course, unless you setup the experiment in certain ways.
>>> Particle A and B are entangled. Let them be 1 lightyear apart. When the state changes, observers of each particle send a signal to each other. The observer at A gets the signal about B one year later. The observer at B gets the signal about A one year later. I don't see how anything really changes by adding in a bunch of IRFs.
Of course, the outcome of the measurement depends on the state of the entangled pair and the type of the measurement. But there will be always a certain non-linear correlation between the two measurements done by observers at A and B. This means that the measurements were already correlated before the observers could possibly communicate. How would you explain this correlation? The actual measurement happened one year in each IRF before getting the data from the other observer.
There are several versions of the Schrodinger equation. For example the time dependent one looks like this:
H * v(t) = i * h * v'(t)
where H = the hamiltonian of the system (if you want, this represents the "hardware") i = the unit imaginary number h = the Planck constant divided by 2 * pi v(t) = the state vector
So basically, the product between the hamiltonian and the state vector is the variation of the state vector in time, multiplied by the constant (i * h)
Correct - there is no way to transmit pure information through photon entanglement for example. But it is possible to use this technique to verify some information transmitted in conjunction with a separate (classic) channel.
This has two consequences: 1) First, it is practically possible to use entanglement to build networks that are 100% guranteed to transmit either correct information or error. 2) Second, since measuring any particle will necessarily change it state gives an interesting conclusion: it is impossible to tamper the communication channel that transmits entangled photons. As soon as you attempted to measure what's on the channel, the verification mentioned above (i.e. the correlation between the final measurement of the two entangled particles at the two ends) will fail!
Therefore you have a bullet proof method that will prevent active/passive attacks on the entangled channel. The technique was actually employed in practice - see this link for example.
NB - this technique still doesn't prevent attacks that fully substitute one of the ends with a completely identical device so the other end still thinks it is talking to the right person. But in combination with standard cryptography techniques for the insecure channel, this techniue is almost impossible to break. A nice overview is presented here
Note that entanglement is just one approach in building quantum computers, and it is not really the ONLY approach.
Generally, a quantum computer consists in several quantum systems (for example captured particles, etc). The (quantum) state of these systems varies according to a well-known equation, called the Schrodringer equation. This is a very simple equation that describes the evolution of the system (the derivative of the current vector state) in respect to the current current state & time.
The nice thing about quantum computers is that they operate with multiple simultaneous states, therefore achieving some sort of parallelism. Basically a quantum system can be considered to have a superposition of states - it has two states at once if you want. Some of these states might converge to the same state depending on the hamiltonian or on the external interactions.
The hard part is that you never know when such a computer stops its calculation since the transformation state is fully reversible and goes on ad infinitum. If you want simply to test if the computer reached the end of the calculation, you will affect the current state. Anywyay, this challenge plus many others (for example the precision of the measurement, etc) makes quantum computing very challenging.
Still, there is a theoretical possibility that you can get a high degree of parallelism in certain configuration. A classical result from Shor (you can search on Google) shows that one of the classic problems in arithmetic - integer factorization - can be done in a polynomial time on a quantum computer. This simply means that RSA encryption can be potentially broken, irrespective to the length of the key. But we are still safe - so far nobody built a working quantum computer that would carry on simple calculations like factorizing the number 15.
On the other side, entanglement is an interesting quantum fenomenon which works like this: 1) First, you have to have a way to build pairs of entangled particles. There are several ways to do this, for example by having any quantum process that generates a pair of photons. 2) Second, if you modify the vector state of one particle, the vector state of the other one will be equally affected, regardless of the distance between these two particles!
What's interesting is that entanglement guarantees instantaneous quantum state change therefore contradicting somehow the theory of special relativity. This theory says that events cannot be 100% simultaneous if they occur in different points in space - there is a timing separation based on the particular reference chosen. Practically, no standard matter interaction can be faster than the speed of light.
But there is an exception here - "collapsing the vector state". If you measure the state of a particle, its state will collapse along one of the measured dimensions (according to certain probabilities). The corresponding entangled particle will suffer a similar change, so if you measure now the state of the this second particle you will see that its vector state has already changed - and you can even perform a partial correlation between the results of the two measurements.
In conclusion, enanglement guarantees instantaneous "interaction" regardless of the distance between these paired particles (this is why Einstein called it "spooky action at a distance" - because technically it is propagated with infinite speed). Anyway, it has be proven a while back that this does NOT contradict the special theory of relativity since this is not a standard matter interaction, like gravity, etc.
Going back to computers, entanglement is an interesting approach which might enable new algorithms or new ways to build such computers. But keep in mind that we are in the stone age of quantum computing right now...
But "writing" bacteria would be hard. We barely understand the organic life at this point. In contrast, we understand technology pretty well.
On a pessimistic note, let's look to computer viruses today. Computers are a very fertile ground for self-replication programs, especially malicious ones. The main reason is ease writing of replication code.
In the case of nano-technology this problem is very hard but not impossible. Unfortunately, I think that, as nanotechnology becomes more and more ubiquitous, the possibility of a malicious mechanical nano-virus will also become real. Compared with its software counterpart, such a nano-virus will probably have an extremely limited lifetime environment. In other words, its replication capabilities will be severely limited.
But still, one could still devise such a virus such that replication will be possible. I will leave aside motivation/ intelligence/ psychology/etc. and concentrate only on the design and technical aspects.
First, let's move to 2100 or sometime in the future where computers are everywhere, and the average size of today's PC would be less than one millimeter in size (or even smaller?)
The actual replication will be done in a hidden nest containing a colony of probably of hundreds of thousands of such viruses, each one performing a different function. This colony will be build near to a place where electric/electronic materials will be around and go unnoticed for years. The colony will also parasitize some existing electrical infrastructure. In this colony, these viruses will perform various functions: - Some of them will gather current from some existing electrical circuitry. - Some will gather existing electronical/electrical components/raw materials from around. - Some will explore the territory near the colony - Some will initially self-assemble and then build more and more powerful computer networks that will run the whole colony.
By all means, replication will be expensive and it will take a lot of time. But keep in mind that these nano-viruses can have all the time at their discretion - probably building the first self-replicating colony will take months if not years. This will also severely limit the speed of replication (so, no, the nano-equivalent of Slammer will not exist).
Such a virus should have a pretty complex structure: - It should be able to move around by himself (which means several mechanical parts, possibly reused from "good nanotechnology components") - contain complex electrical circuits (the equivalent of a today's computer, really), including NVRam, Flash, CPU, etc. - and some source of power, enough to keep the whole thing running. - It can use light as an energy source just to move around. It will move around looking for other sources of energy on which it can parasitize, possibly by teaming with other similar viruses to form more complex electrical circuits.
You can see that the whole stuff is at the upper limit at the nanotech scale, probably beyond (probably around 0.1 - 1 mm if not bigger). I don't see how this stuff can be scaled down to smaller dimensions but who knows.
Replication would be the hardest part. It will need some sort of raw materials like silicon for the solar panels, metals, various type of insulator, spare computer pieces (assuming that computers are extremely small and everywhere as all these prophets predict in 2100:-). Next, these viruses will team up in large colonies to build some sort of "virus" assembly factories which will probably contain a population of 10,000-100,000 viruses. These factories will grow unnoticed until it is too late.
Of course, all this is purely fictional stuff although I think it is technically doable. The hardest part is designing such a virus and here I can only hope that hacker creativity has an upper limit...:-) Or, bad governments have an upper money limit, whatever.
The proof (or, better said, the sketch of the proof) actually starts at the end of page 21, very close to the last page. The original work is actually pretty hard to find since it is buried in so many unrelated side notes.
Here is the general outline: 1) At the end of page 19 he mentions that "The positivity condition which is introduced implies the Riemann hypothesis if it applies to Dirichlet zeta functions." 2) After some introduction of the quantum gamma functions that lasts two pages, the actual proof starts at the end of page 21 with the phrase "A quantum gamma function is obtained when is nonnegative. A proof of positivity is given from properties of the Laplace transformation." 3) The proof ends in the middle of page 23 with the a verification that W(z) is a quantum gamma function with quantum q = exp(-2*pi), obtained from a spectral theory of the shift operator.
Overall this is just a very brief sketch of the whole proof.
BTW, to add gas on fire, here is an exceprt from mathworld.com, which surprisingly was missed by/. until now:-)
http://mathworld.wolfram.com
Riemann Hypothesis "Proof" Much Ado About Noithing (sic) A June 8 Purdue University news release reports a proof of the Riemann Hypothesis by L. de Branges. However, both the 23-page preprint cited in the release (which is actually from 2003) and a longer preprint from 2004 on de Branges's home page seem to lack an actual proof. Furthermore, a counterexample to de Branges's approach due to Conrey and Li has been known since 1998. The media coverage therefore appears to be much ado about nothing.
The counterexample to Brangles approach can be reached here: http://arxiv.org/abs/math.NT/9812166
No. It is a common misconception that irradiated food becomes radioactive. This is simply not true.
For one thing, beta and gamma rays are not causing nuclear reactions. Food is being usually irradiated with gamma rays from Cobalt-60 sources. This kills most living organisms (microbes etc) but the fod itself does not become radioactive.
In contrast, alpha rays (which consist in high-speed helium nucleus) are known to cause further nuclear reactions in some atoms (for example berillium irradiated with alpha rays is a good source of neutrons). Also, neutron radiation also causes nuclear reactions in many elements as we all know.
It would be surprising that Senda has a moon. After all, Sedna itself is comparable in size with our own moon (Sedna has less than 1700 Km in diameter, and our moon has around 3500 Km in size).
Now I am wondering if our Moon has another moon orbiting around:-) I am sure that somebody searched for it.
>>> If you think even a few percent of Google's customers will come close to 1GB of email within the first few years then you overestimate the average email user. Even if they did, using an aggressive compression algorithm they can cut store a full account (1GB of uncompressed email) down by at least 50%, if not to 20-25%. Since it has to be email then it has to follow normal email encoding standards (7bit, base64 encoding for binary, etc)
Actually the real numbers are eight times larger. In my original calculation I was multiplying Gigbabits (Gb) instead of Gigabytes (GB). I am still learning the Google calculator:-)
I would say that, even with a lot of compression techniques and autoamtic spam collection, 1% of that would be 2 Petabytes which is still a lot.
Additionally, I wouldn't overestimate the ability of current users to delete spam mail (or large email). With a 1 Gb of storage your usage habits are very different than with a 30 Mb mailbox. In the former case, you can easily end up with tens of thousands of emails over time and you do not necessarily know what to retain. In the later case, you have a much smaller quantity of emails and you will clean up your inbox much more often.
The key point here is psychology - users will be so confortable dealing with such a large inbox that they won't know what to clean and how to clean. On the contrary, having a small inbox forces you to clean up mess often.
In some sort of sense, this is getting similar with managing space on your own computer. I have probably hundreds of thousands of files on my PC and god knows what's there. Even the smartest and fastest search algorithms won't help me clean up this mess.
>>> In the end, hard drive space is cheap They can set up a fully backed up terabyte array for under $1000. That terabyte array will support thousands of 'average' users, and hundreds of 'power' users
Harddisk might be cheap but it consumes power. You have to spin around all these plates - and this is wasted energy compared with tape, for example. The problem here is that we are dealing with thousands of storage arrays, not with only one.
One thing that makes me skeptical on Gmail is the huge amount of storage required to keep the system running.
1) Let's make a simple calculation: let's pick up the number of Hotmail accounts (200,000,000 as I heard last time). Multiply this with 1 Gb and you get 24 Petabytes of data!
(See Google for more details http://www.google.com/search?hl=en&ie=UTF-8&oe=UTF -8&q=200000000+*+1+Gb )
It would be interesting to know how much data does Google store today.
2) Now, let's compute how much power will this system consume? Assuming at least a RAID 1 configuration, you would need at least 48 Petabytes of storage since we all know that harddisks fail.
Let's assume that one harddisk stores around 250 Gb of data. Let's assum uncompressed data (since those 1 Gb can contain anything after all... This means that we need around 200,000,000 * 2 / 250 = 1,600,000 harddrives running all the time!
Now, let's pick up the power consumption to be around 10 W. We then get around 1,600,000 * 10 = 16 Gigawatts of power to be dissipated. Now THAT is a lot of power... Think of all the maintenance costs for running this for only one year.
Anyway, the engineering challenges are pretty strong here. I imagine that Google is taking a risky bet here and hopes to develop storage rack/ventiation technology "on the go".
In conclusion, I really think that either Gmail won't be free, or the 1 Gb limit is a marketing number.
Actually this is just a game to set the right expectations. They designed whole thing with a much larger project life right from the beginning.
But, given the fact that the rover technology is low-cost and still unproven, they expected a certain risk for various glitches. So, a 250 days "published" interval followed by a deadly clitch would mean a very bad image for NASA.
NASA played the same "stay on the safe side" tune on many otehr missions - see for example the Voyager missions, etc.
Slashdot Mirror
I'm a bit of a skeptic but In your list I would like to see (as necessary features of a modern file system):
- online defrag (you don't take the volume offline to do a defrag)
- online integrity checking
- online self-healing.
- a pluggable infrastructure for various filters: antivirus, encryption, compression, folder redirection, etc.
I would view them as unnecessary:
- snapshots: yes, they can be implemented in the file system (or below) but note that a software implementation will never match a hardware implementation at storage array level. Most storage arrays today support high-performant snapshots at the LUN/disk level. You can have software snapshots also, of course but perf will not be very good. Even Sun storage arrays have snapshots.
- Auto-striping: stripping, spanning, mirroring, RAID5, etc are just some features that a storage array would provide. Anyway, these are implemented usually below the file system level, usually at volume level (for software implementations) or LUN level (for hardware implementations)
Hmmm.. does this 8% include reboots caused by patch installation?
In my case it took me about one hour of C# programming and head scratching to found out the fifth number. The puzzle is not very hard but it requires a little bit of unorthodox thinking. Don't start with classical extrapolation techniques...
If you want hard math puzzles, check the "Ponder this" site from IBM - there are several beautiful puzzles there: http://www.research.ibm.com/ponder/
Seriously now, there is something interesting about nuclear reactors - their generated power is not really related with their size. So there is no such thing as "nuclear mini-reactor". You have almost the same quantity of nuclear fuel in each of them.
The reason is simple: the mass of fissionable material should be just a little above the critical mass (which is minimum 13 Kg for Pu-239 and slightly larger for U-235). The critical mass of course depends on the core design but you can't really extend that beyond a certain safety limit, and you also cannot keep that under the criticality limit (otherwise you won't have any self-sustainable reaction; only a tiny percentage of the fuel ends up consumed; the vast majority of the fuel is present there just to ensure criticality).
You need to be just above the critical mass in a small interval mainly for control reasons. The neutrons are produced in two class of reactions:
(1) neutrons that are produced by the main fission reaction (which generates around 80% of the total quantity of neutrons) and
(2) neutrons produced by secondary fission reactions from the extremely radioactive fission by-products from the first reaction. The production of these neutrons can be therefore considered as "delayed". These delayed neutrons are essential for reactor control - here is why:
At the end, fission is a chain reaction where you get almost 3 neutrons per fission (assuming U-235), from which you loose 2 and the remaining neutron is used to initiate the next fission. The average time for this generated neutron to participate in the next fission is in the order of milliseconds. In other words, every millisecond or so a new chain reaction produces a new generation of neutrons of class (1).
Let's forget a second about the second generation: if you are above the critical mass, over a very short period of time you get more and more neutrons which will build up in tens or hundreds of milliseconds way above the safety limit. For example let's say that you are 1.001 above the limit. This means that in the first generation you have a 0.001 excess, in the second generation 3*0.001, etc. After N generations you have 3^N * 0.001 which can be way beyond 1 in less than one second. The conclusion is that it is impossible to control a reactor by inserting the control neutron-absorber rods in the core - since again it takes less than one second to double up the neutrons.
Now, since you have these neutrons from the second generation, it is possible to control the reactor without quick build-up only if you manage that those 80% neutrons produced in the first class to be below criticality (otherwise you will have the quick build-up phenomenon above), but the total quantity of neutrons to be above. This way, you will ensure a delayed reaction time of a couple of minutes in which you can safely control the reactor.
The conclusion is that it doesn't really matter how you build the reactor - you will have the almost the same quantity of nuclear fuel there. There are no "micro/mini" reactors.
>>> Cryptography will die when the last human draws its breath.
You meant, cryptography will die when the third to last human draws its breath, correct? When you have two humans left there is no need for cryptography.
Remember: Alice, Bob, Eve...
This post is way overrated. Photoremediation is simply a hoax.
You cannot induce nuclear reactions by irradiating matter with light.
There are certain considerations here:
1) Nobody considers now breeder reactors as a good alternative for nuclear power generation.
There are several reasosn, assuming that you want a U-238 based breeder reactor:
a) One reason is that is very hard to intensively reprocess the spent fuel. This requires huge installations that has to dissolve the spent fuel after a couple of weeks/months in nitric acid or equivalent. This is pretty challenging, even for US who has the experience of dozens of years in U-238/Pu-239 separation. The main problem is that you have all these dangerous byproducts like Cs-137, Sr-90, which appear in small quantities but still are pretty reactive properties (a medium half-life of a couple of years, etc). They are still pretty dangerous and hard to isolate chemically.
b) Another issue: the Pu-239 becomes a very dangerous material that fall off in the hands of terrorists. You need around 15-30 Kg of Pu-239 to build something that would detonate. This is a very small quantity. Today, there is a lot of Pu-239 that has to be either consumed or buried such that it won't fall in the wrong hands.
c) And finally, nuclear reactors are not that efficient. For one thing, the amount of generated power doesn't really depend on nuclear reactor size. Mainly because you have to be in the region between the critical and supercritical mass for a certain configuration. This also means that the main challenge in designing a reactor is optimizing the heat transfer process. Smaller reactors are better.
Second, you get about 1% energy from the fission fuel in one pass; there is no way to get 10%. And, reusing the other 99% for anything is pretty hard.
Well, probably there is still some room for some message, like "Alien XXX please fix your software" or something like this :-)
Some advanced viruses such as bacteriophages of the family Leviviridae, can reach almost 300 Kbp (kilobase pairs) which should be enough to store some additional data.
Somewhere I read that some flu viruses might be of alien form. Indeed, they seem to be the ideal organisms built for space travel. So why don't we search for alien messages in their DNA too?
This doesn't sound too complicated.
First, you have come up with a general plan.
- What exactly do you want to test - and to what degree of certainity
- What are your priorities - here I would say that the accuracy must be less than 1%
- What are your resources (i.e. how many testers, their qualifications)
- What is the timeline
- What is the estimated cost (and ultimately, what is the real profit from this stuff?)
Then you start to develop a small test plan. First, you need a precise behavioral model of your system. This means that if you have 1000 terminals, you have to define a model which says how the votes "add up" together in the main system.
Next step you design a real, physical interface to enter data in the system. This needs to be completely automated. Yes, you will need to use rubber fingers (or something equivalent) hooked up to some electromagnetic devices that triggers a finger to be "pushed" at a certain time. Then you define a module that would accept, say, RS232 commands and in response will push a certain finger. This is not very complicated and not expensive either. Then you just need one PC to send these commands over a set of serial cables to each "voting monkey" device.
If you have the ability to run the touch-screen software in a simulator test the software separately, using some scripts that simulate the "touch" commands directly in the simulator. This would be a cheaper solution although you will still need to do some manual/semi-automated end-to-end testing.
Finally, you have to one or several machines that will:
- Run the main test programs
- Run the "monkey scripts" that ultimately will cause these rubber fingers to go in action
- Gather various activity logs from the affected systems
- Correlate the data gathered from the voting systems with the actual votes.
The main difficulty would be writing the whole test package, depending on how complicated your test needs to be. You can decide how much time you need to spend on functionality testing vs. longhaul tests etc.
Anyway, what I wrote is a very high-level overview of a plan. The main point is that developing an end-to-end, 100% automated test for a system containing things like touch screens is not rocket science as long as you know what you are doing and you hired the right people.
It is actually better for a different reason. Each of these states evolves in parallel according to the Schrodinger equation for that particular quantum system. In other words you have some sort of parallelism.
You are right - I meant to say beyond 15... :-)
>>> I don't see how this is 100% simultaneous. Let's say one pair breaks down.
It is simultaneous simply because as long as you don't perturb it, a pair of entangled particles constitute a single quantum system, a single "particle" if you want. As soon as you touch one of them, the pair is broken and now suddenly you can imagine having two separate particles popping up in space at certain distance between them.
>>> We call that event A in the observer X's IRF. Then another pair breaks down - event B. Let's use A as our reference point. An observer Y in a different IRF will say that event B did not happen at the same time oberver X said it did. There will be a difference in the time elapsed.
No. The state vector collapse is not an observable event in itself. Only the measurement is an observable event. All you can do (as observer X) is to measure both particles and hope/presume that the collapse already hapened - but you have no 100% guarantee that of course, unless you setup the experiment in certain ways.
>>> Particle A and B are entangled. Let them be 1 lightyear apart. When the state changes, observers of each particle send a signal to each other. The observer at A gets the signal about B one year later. The observer at B gets the signal about A one year later. I don't see how anything really changes by adding in a bunch of IRFs.
Of course, the outcome of the measurement depends on the state of the entangled pair and the type of the measurement. But there will be always a certain non-linear correlation between the two measurements done by observers at A and B. This means that the measurements were already correlated before the observers could possibly communicate. How would you explain this correlation? The actual measurement happened one year in each IRF before getting the data from the other observer.
Here is a more detalied explanation.
There are several versions of the Schrodinger equation. For example the time dependent one looks like this:
H * v(t) = i * h * v'(t)
where
H = the hamiltonian of the system (if you want, this represents the "hardware")
i = the unit imaginary number
h = the Planck constant divided by 2 * pi
v(t) = the state vector
So basically, the product between the hamiltonian and the state vector is the variation of the state vector in time, multiplied by the constant (i * h)
For more details see this link.
There are many examples, such as the square-root of NOT gate.
Correct - there is no way to transmit pure information through photon entanglement for example. But it is possible to use this technique to verify some information transmitted in conjunction with a separate (classic) channel.
This has two consequences:
1) First, it is practically possible to use entanglement to build networks that are 100% guranteed to transmit either correct information or error.
2) Second, since measuring any particle will necessarily change it state gives an interesting conclusion: it is impossible to tamper the communication channel that transmits entangled photons. As soon as you attempted to measure what's on the channel, the verification mentioned above (i.e. the correlation between the final measurement of the two entangled particles at the two ends) will fail!
Therefore you have a bullet proof method that will prevent active/passive attacks on the entangled channel. The technique was actually employed in practice - see this link for example.
NB - this technique still doesn't prevent attacks that fully substitute one of the ends with a completely identical device so the other end still thinks it is talking to the right person. But in combination with standard cryptography techniques for the insecure channel, this techniue is almost impossible to break. A nice overview is presented here
Note that entanglement is just one approach in building quantum computers, and it is not really the ONLY approach.
Generally, a quantum computer consists in several quantum systems (for example captured particles, etc). The (quantum) state of these systems varies according to a well-known equation, called the Schrodringer equation. This is a very simple equation that describes the evolution of the system (the derivative of the current vector state) in respect to the current current state & time.
The nice thing about quantum computers is that they operate with multiple simultaneous states, therefore achieving some sort of parallelism. Basically a quantum system can be considered to have a superposition of states - it has two states at once if you want. Some of these states might converge to the same state depending on the hamiltonian or on the external interactions.
The hard part is that you never know when such a computer stops its calculation since the transformation state is fully reversible and goes on ad infinitum. If you want simply to test if the computer reached the end of the calculation, you will affect the current state. Anywyay, this challenge plus many others (for example the precision of the measurement, etc) makes quantum computing very challenging.
Still, there is a theoretical possibility that you can get a high degree of parallelism in certain configuration. A classical result from Shor (you can search on Google) shows that one of the classic problems in arithmetic - integer factorization - can be done in a polynomial time on a quantum computer. This simply means that RSA encryption can be potentially broken, irrespective to the length of the key. But we are still safe - so far nobody built a working quantum computer that would carry on simple calculations like factorizing the number 15.
On the other side, entanglement is an interesting quantum fenomenon which works like this:
1) First, you have to have a way to build pairs of entangled particles. There are several ways to do this, for example by having any quantum process that generates a pair of photons.
2) Second, if you modify the vector state of one particle, the vector state of the other one will be equally affected, regardless of the distance between these two particles!
What's interesting is that entanglement guarantees instantaneous quantum state change therefore contradicting somehow the theory of special relativity. This theory says that events cannot be 100% simultaneous if they occur in different points in space - there is a timing separation based on the particular reference chosen. Practically, no standard matter interaction can be faster than the speed of light.
But there is an exception here - "collapsing the vector state". If you measure the state of a particle, its state will collapse along one of the measured dimensions (according to certain probabilities). The corresponding entangled particle will suffer a similar change, so if you measure now the state of the this second particle you will see that its vector state has already changed - and you can even perform a partial correlation between the results of the two measurements.
In conclusion, enanglement guarantees instantaneous "interaction" regardless of the distance between these paired particles (this is why Einstein called it "spooky action at a distance" - because technically it is propagated with infinite speed). Anyway, it has be proven a while back that this does NOT contradict the special theory of relativity since this is not a standard matter interaction, like gravity, etc.
Going back to computers, entanglement is an interesting approach which might enable new algorithms or new ways to build such computers. But keep in mind that we are in the stone age of quantum computing right now...
But "writing" bacteria would be hard. We barely understand the organic life at this point. In contrast, we understand technology pretty well.
:-). Next, these viruses will team up in large colonies to build some sort of "virus" assembly factories which will probably contain a population of 10,000-100,000 viruses. These factories will grow unnoticed until it is too late.
:-) Or, bad governments have an upper money limit, whatever.
On a pessimistic note, let's look to computer viruses today. Computers are a very fertile ground for self-replication programs, especially malicious ones. The main reason is ease writing of replication code.
In the case of nano-technology this problem is very hard but not impossible. Unfortunately, I think that, as nanotechnology becomes more and more ubiquitous, the possibility of a malicious mechanical nano-virus will also become real. Compared with its software counterpart, such a nano-virus will probably have an extremely limited lifetime environment. In other words, its replication capabilities will be severely limited.
But still, one could still devise such a virus such that replication will be possible. I will leave aside motivation/ intelligence/ psychology/etc. and concentrate only on the design and technical aspects.
First, let's move to 2100 or sometime in the future where computers are everywhere, and the average size of today's PC would be less than one millimeter in size (or even smaller?)
The actual replication will be done in a hidden nest containing a colony of probably of hundreds of thousands of such viruses, each one performing a different function. This colony will be build near to a place where electric/electronic materials will be around and go unnoticed for years. The colony will also parasitize some existing electrical infrastructure. In this colony, these viruses will perform various functions:
- Some of them will gather current from some existing electrical circuitry.
- Some will gather existing electronical/electrical components/raw materials from around.
- Some will explore the territory near the colony
- Some will initially self-assemble and then build more and more powerful computer networks that will run the whole colony.
By all means, replication will be expensive and it will take a lot of time. But keep in mind that these nano-viruses can have all the time at their discretion - probably building the first self-replicating colony will take months if not years. This will also severely limit the speed of replication (so, no, the nano-equivalent of Slammer will not exist).
Such a virus should have a pretty complex structure:
- It should be able to move around by himself (which means several mechanical parts, possibly reused from "good nanotechnology components")
- contain complex electrical circuits (the equivalent of a today's computer, really), including NVRam, Flash, CPU, etc.
- and some source of power, enough to keep the whole thing running.
- It can use light as an energy source just to move around. It will move around looking for other sources of energy on which it can parasitize, possibly by teaming with other similar viruses to form more complex electrical circuits.
You can see that the whole stuff is at the upper limit at the nanotech scale, probably beyond (probably around 0.1 - 1 mm if not bigger). I don't see how this stuff can be scaled down to smaller dimensions but who knows.
Replication would be the hardest part. It will need some sort of raw materials like silicon for the solar panels, metals, various type of insulator, spare computer pieces (assuming that computers are extremely small and everywhere as all these prophets predict in 2100
Of course, all this is purely fictional stuff although I think it is technically doable. The hardest part is designing such a virus and here I can only hope that hacker creativity has an upper limit...
The proof (or, better said, the sketch of the proof) actually starts at the end of page 21, very close to the last page. The original work is actually pretty hard to find since it is buried in so many unrelated side notes.
/. until now :-)
Here is the general outline:
1) At the end of page 19 he mentions that "The positivity condition which is introduced implies the Riemann hypothesis if it applies to Dirichlet zeta functions."
2) After some introduction of the quantum gamma functions that lasts two pages, the actual proof starts at the end of page 21 with the phrase "A quantum gamma function is obtained when is nonnegative. A proof of positivity is given from properties of the Laplace transformation."
3) The proof ends in the middle of page 23 with the a verification that W(z) is a quantum gamma function with quantum q = exp(-2*pi), obtained from a spectral theory of the shift operator.
Overall this is just a very brief sketch of the whole proof.
BTW, to add gas on fire, here is an exceprt from mathworld.com, which surprisingly was missed by
http://mathworld.wolfram.com
Riemann Hypothesis "Proof" Much Ado About Noithing (sic)
A June 8 Purdue University news release reports a proof of the Riemann Hypothesis by L. de Branges. However, both the 23-page preprint cited in the release (which is actually from 2003) and a longer preprint from 2004 on de Branges's home page seem to lack an actual proof. Furthermore, a counterexample to de Branges's approach due to Conrey and Li has been known since 1998. The media coverage therefore appears to be much ado about nothing.
The counterexample to Brangles approach can be reached here: http://arxiv.org/abs/math.NT/9812166
No. It is a common misconception that irradiated food becomes radioactive. This is simply not true.
For one thing, beta and gamma rays are not causing nuclear reactions. Food is being usually irradiated with gamma rays from Cobalt-60 sources. This kills most living organisms (microbes etc) but the fod itself does not become radioactive.
In contrast, alpha rays (which consist in high-speed helium nucleus) are known to cause further nuclear reactions in some atoms (for example berillium irradiated with alpha rays is a good source of neutrons). Also, neutron radiation also causes nuclear reactions in many elements as we all know.
It would be surprising that Senda has a moon. After all, Sedna itself is comparable in size with our own moon (Sedna has less than 1700 Km in diameter, and our moon has around 3500 Km in size).
:-) I am sure that somebody searched for it.
Now I am wondering if our Moon has another moon orbiting around
>>> If you think even a few percent of Google's customers will come close to 1GB of email within the first few years then you overestimate the average email user. Even if they did, using an aggressive compression algorithm they can cut store a full account (1GB of uncompressed email) down by at least 50%, if not to 20-25%. Since it has to be email then it has to follow normal email encoding standards (7bit, base64 encoding for binary, etc)
:-)
& oe =UTF-8&q=200000000+*+1+GB
Actually the real numbers are eight times larger. In my original calculation I was multiplying Gigbabits (Gb) instead of Gigabytes (GB). I am still learning the Google calculator
So the real number is 190 Petabytes!
http://www.google.com/search?hl=en&lr=&ie=UTF-8
I would say that, even with a lot of compression techniques and autoamtic spam collection, 1% of that would be 2 Petabytes which is still a lot.
Additionally, I wouldn't overestimate the ability of current users to delete spam mail (or large email). With a 1 Gb of storage your usage habits are very different than with a 30 Mb mailbox. In the former case, you can easily end up with tens of thousands of emails over time and you do not necessarily know what to retain. In the later case, you have a much smaller quantity of emails and you will clean up your inbox much more often.
The key point here is psychology - users will be so confortable dealing with such a large inbox that they won't know what to clean and how to clean. On the contrary, having a small inbox forces you to clean up mess often.
In some sort of sense, this is getting similar with managing space on your own computer. I have probably hundreds of thousands of files on my PC and god knows what's there. Even the smartest and fastest search algorithms won't help me clean up this mess.
>>> In the end, hard drive space is cheap They can set up a fully backed up terabyte array for under $1000. That terabyte array will support thousands of 'average' users, and hundreds of 'power' users
Harddisk might be cheap but it consumes power. You have to spin around all these plates - and this is wasted energy compared with tape, for example. The problem here is that we are dealing with thousands of storage arrays, not with only one.
One thing that makes me skeptical on Gmail is the huge amount of storage required to keep the system running.
F -8&q=200000000+*+1+Gb )
1) Let's make a simple calculation: let's pick up the number of Hotmail accounts (200,000,000 as I heard last time). Multiply this with 1 Gb and you get 24 Petabytes of data!
(See Google for more details http://www.google.com/search?hl=en&ie=UTF-8&oe=UT
It would be interesting to know how much data does Google store today.
2) Now, let's compute how much power will this system consume? Assuming at least a RAID 1 configuration, you would need at least 48 Petabytes of storage since we all know that harddisks fail.
Let's assume that one harddisk stores around 250 Gb of data. Let's assum uncompressed data (since those 1 Gb can contain anything after all... This means that we need around 200,000,000 * 2 / 250 = 1,600,000 harddrives running all the time!
Now, let's pick up the power consumption to be around 10 W. We then get around 1,600,000 * 10 = 16 Gigawatts of power to be dissipated. Now THAT is a lot of power... Think of all the maintenance costs for running this for only one year.
Anyway, the engineering challenges are pretty strong here. I imagine that Google is taking a risky bet here and hopes to develop storage rack/ventiation technology "on the go".
In conclusion, I really think that either Gmail won't be free, or the 1 Gb limit is a marketing number.
Actually this is just a game to set the right expectations. They designed whole thing with a much larger project life right from the beginning.
But, given the fact that the rover technology is low-cost and still unproven, they expected a certain risk for various glitches. So, a 250 days "published" interval followed by a deadly clitch would mean a very bad image for NASA.
NASA played the same "stay on the safe side" tune on many otehr missions - see for example the Voyager missions, etc.
EMachines never had any stores and did pretty well... I guess this is why they are merging with Gateway.