We did eigenvalues in second year linear algebra, but we never actually had any real-world applications for it, hence the course had no relevance to the 'real world'. I came across eigenvalues in a journal on personal relationships would you believe it, but I didn't actually know what they were used for.
@frichikendz if we have a square matrix, which is basically an array of numbers with equal numbers of rows and columns, an eigenvalue is a number that is related to that matrix. for a newb like you, showing you how to get the eigenvalues would be difficult.
an algorithm is completely different. it is basically a set of steps for taking input and producing output. for example, do you remember long division? the procedure you learned is an algorithm called the division algorithm.
@gotama420 not in German, where the word 'eigen' in 'eigenvalue' is from. There it means 'own'. So it is it's 'own value' or a 'characteristic value' which is the English term.
"Eigen" translated from German: adjective own peculiar independent inherent proper individual particular separate strange typical fussy possessive Thumbs up to put curious information on the top
If your looking for computing power against cost right now you could do a lot worse than pick up a few second hand ps3's and cluster them up, like the us navy have famously done. IBM cell processor with gigabit LAN port running linux.
I feel as though this video would benefit from a follow-up which talks about the mathematical meaning of an eigenvalue/vector and some of its implications and uses. This video seems to give vague applications of eigenvalues/vectors without saying what they are. Obviously going into extreme detail about eigenvalues in 5-10 minutes is impossible, but I feel that the equation A*x=lambda*x is extremely important and should be covered in a video about eigenvalues and eigenvectors.
@BGood156 He basically did; he said it's a characteristic number of a matrix, which is easily understood as "This number is equivalent to a Matrix when used" similar to Ax=lamba x where lamba is equivalent to the matrix for scalar multiplication.
@loopdeloop17 That's quite a leap in logic there. That said, I can kind of see how you came to that conclusion as you almost certainly have some kind of background in linear algebra, however, your description is definitely not what came to my mind when I first watched this video. If someone never took linear algebra, the explanation that was given in the video would be woefully inadequate in giving any intuition about what eigenvalues and eigenvectors actually are.
in german it's Eigenfrequenz. which means litaraly "own frequence" so its better to translate it as distinctive frequence or something like build-in frequence;)
@uut0 I don't believe it is really necessary, we write on 2 dimensionally. But if you are talking about a matrix denoting dimensions, yes. You can have as many as you wish, he even mentioned a 10 x 10, which could because 10 dimensional.
I am an amateur physics enthusiast at best. That having been said, is there any relationship between eigenvalues and the term eigenstate which is commonly used in quantum mechanics?
wow this video is really good so glad to see the physics department getting a channel. we did eigenvalues a while ago in maths- good to see it in use!
ya, the half life symbol is actually used in chemistry for the half life of an atom of some kind of element. i.e. Mercury-196: Stable. But something like Mendelevium-258 has a half life of 51.5 days.
Buy some several Playstation 3s. There are instructions (on the internet) on how to turn several (I can't remember how many you need) of them into a supercomputer that I would imagine would increase the amount of information you can process.
still, the playstations have a low processing power to price ratio.. probably a core 2 quad would beat all those computers he's using.. but only if he could program it in a way that it used the 8 processors..
PS3 in a general sense may have relatively low processing power, but like all dedicated graphics chips they are hugely superior to general intel or AMD chips at certain types of mathematical calculations, often the sort of calculations needed in research, hence why many researchers around the world have switched renting out time on supercomputers for having their own "supercomputer" made up of PS3s (at a fraction of the cost and often with a better performance)
I Agree with you PS3 have low processing power typically there really good at math certain math equations and rendering (Folding @ Home) due to cell technology but Eigenvalues are really interesting if you study into it ;D
Price pr calculation wice (both aqusition and powerconsumtion) new GPUs are more efficient than a PS3. I myself own a GPU wich can do 2,4 TFLOPS (= Terra Floating Point Operations Per Second. 2400 billion operations per second with decimalnumbers). If this professors work could be made into parallell code, a graphicscard costing £100 could outperform all those computers easily. By clustering high-end cards in those machines he could do a years worth of computing in a day to a week.
@gulllars Sounds unlikely, since he's working with chaotic systems, and precision on GPGPUs is horrible. And if you use double-precision on the GPU, most of the speed advantage vanishes. (and even then the precision might still be too poor)
The power of DSPs. Solving complex equations by manipulating signals and analyzing the results. The speed and efficiency of modern DSPs is what gives GPU computing its edge.
@gulllars ima guess ur talking about a high end 5000 series from ati? i got something like dat too ;) i didnt quite get the gist of the calculations, but that would only work if the calculations arent linear. ie the next calculation doesnt depend on the previous result. if many calculations can be done in parallel, then yes, a couple GPUs could do the work an order of magnitudes more quickly than his set up there.
@dandil a year ago, i had a 4870x2. Now i have a 6850, since it's sufficient for my current gaming, and i got tired of the noise.
To harvest GPU FLOPS require highly parallell problems, or better, embarresingly parallel problems. You need high thread level parallelity (and instruction level parallelity too helps on ATI/AMD).
I can see chaotic systems being difficult to optimize for GPGPU.
Things like Ray-Tracing, rasterization, and biologic, geologic, or astronomic prosessing can use GPGPU well.
@gulllars Yes, but (1) parallel calculations are more in the marketing realm than they are of practical use, and (2) the programs used for generic math simulations are hard pressed to be even compiled for the right processor. One has to be more of a programmer than a physicist to leverage those GPUs.
@ZdenekJindra the latest generations of GPUs support OpenCL, which is not that different from C+, and support hardware agnostic massive parallell execution.
Anyways, for embarrassingly parallel workloads, GPUs offer unparalleled performance pr Watt and £.
@ZdenekJindra if you could get grad students to re-adapt your code to OpenCL as a project, and keep the front-end or GUI similar, it could be done fairly transparent so the researcher only has to adapt to minor changes in the procedures at a high level, and multiply output several times, maybe orders of magnitude depending on workload.
@gulllars I do not know about his work but I work with fluids and Computational Fluid Dynamics and in it, for research purposes, a computation that takes about 200h in the fastest supercomputers is considered acceptable.
Eigenvalues find their application also in quantum physics!
For instance to obtain the energy of a quantum system you take the wavefunction of the system, then you apply an operator, which can be very complicated, and the result is the same exact function, multiplicated by a factor, which is the energy and is an eigenvalue.
We did eigenvalues in second year linear algebra, but we never actually had any real-world applications for it, hence the course had no relevance to the 'real world'. I came across eigenvalues in a journal on personal relationships would you believe it, but I didn't actually know what they were used for.
pidgeoninthecoop 1 week ago
I did all this in school once. It's a shame I never used it afterwards and couldn't do it without a textbook anymore.
aluisious 3 weeks ago
Eigen is both dutch and german. It means "own" in both languages.
SemperVinco 1 month ago
@SemperVinco It's also (almost) Danish, though spelled egen and eigenvalue is egenværdi which is a literal translation
kas00078 5 days ago
Eigen is a Dutch word not a German word.
oranjeliniaal 2 months ago
@oranjeliniaal no, eigen is german (I am from germany) It's from an german scientist
highguyontour 1 month ago
Wait... So if I find out the Eigenvalue of a bridge... I can collapse it? Awesome...
Versudan 3 months ago
Comment removed
frichikendz 3 months ago
@frichikendz if we have a square matrix, which is basically an array of numbers with equal numbers of rows and columns, an eigenvalue is a number that is related to that matrix. for a newb like you, showing you how to get the eigenvalues would be difficult.
an algorithm is completely different. it is basically a set of steps for taking input and producing output. for example, do you remember long division? the procedure you learned is an algorithm called the division algorithm.
UltraProle21 3 months ago
1:33 Takes a scientist to excite a ruler with some initial condition.
ratje67 5 months ago 5
Please can you do a response video to explain what you are doing?
av733 5 months ago
@sixty symbols thanks a lot.I was wondering about the physical significance of eigen values .this was really really helpful.
rampriyadarshini 5 months ago
How long ago was this filmed? Surely not in 2009? There's a CRT monitor in it, and it's not in a landfill or museum!
Sachaztan 5 months ago
@gotama420 not in German, where the word 'eigen' in 'eigenvalue' is from. There it means 'own'. So it is it's 'own value' or a 'characteristic value' which is the English term.
guidovseeters 6 months ago 2
Comment removed
guidovseeters 6 months ago
those look like pentium 3 1 GHz rigs, does he know about moors law and how it is still holding?
danthemanzizzle 6 months ago
axelasdf 10 months ago
More examples. :>
Try to use an FPGA or DSP.
AndyKong51 10 months ago
civil engineering
FistsofVengeance 11 months ago
that dude just needs to use an abisk
isrealjason 1 year ago
someone needs to get CUDA GPU processing going in this place..
Pada007gangster 1 year ago 2
If your looking for computing power against cost right now you could do a lot worse than pick up a few second hand ps3's and cluster them up, like the us navy have famously done. IBM cell processor with gigabit LAN port running linux.
Tossphate 1 year ago
He might have excited the ruler he didn't excite me.
857frank 1 year ago
I need to make that man a computer, 256GB of RAM and 192 2-thread cores would keep him busy.
HWGuyEG 1 year ago
he is talking about resonance.....
theallseeingi237 1 year ago
omg math lesson flashbacks
Accursed2552 1 year ago
He's probably running a p3. lol
get this man a nvidia supercluster.
DarkKnightBob1o1 1 year ago
Comment removed
Alcatar 1 year ago
get an NVIDIA Tesla personal supercomputer.
muzammilali007 1 year ago
Eigen stands for own. You can say: "Das ist deine EIGENE Schuld" meaning this is your own fault.
EbayEuron 1 year ago
I feel as though this video would benefit from a follow-up which talks about the mathematical meaning of an eigenvalue/vector and some of its implications and uses. This video seems to give vague applications of eigenvalues/vectors without saying what they are. Obviously going into extreme detail about eigenvalues in 5-10 minutes is impossible, but I feel that the equation A*x=lambda*x is extremely important and should be covered in a video about eigenvalues and eigenvectors.
BGood156 1 year ago
@BGood156 He basically did; he said it's a characteristic number of a matrix, which is easily understood as "This number is equivalent to a Matrix when used" similar to Ax=lamba x where lamba is equivalent to the matrix for scalar multiplication.
loopdeloop17 1 year ago
@loopdeloop17 That's quite a leap in logic there. That said, I can kind of see how you came to that conclusion as you almost certainly have some kind of background in linear algebra, however, your description is definitely not what came to my mind when I first watched this video. If someone never took linear algebra, the explanation that was given in the video would be woefully inadequate in giving any intuition about what eigenvalues and eigenvectors actually are.
BGood156 1 year ago
in german it's Eigenfrequenz. which means litaraly "own frequence" so its better to translate it as distinctive frequence or something like build-in frequence;)
swunt10 1 year ago
oh my god....i just realised i am so stupid :))))) heheh, these guys are brain
MrEurgbp 1 year ago
Please god let me build you a PC
smeghead666 1 year ago
The best translation of "eigen" is probably "intrinsic" in this case.
conoba 1 year ago
The man wants Scotty on hand for a job like this. ^.^
temporaldisplacement 1 year ago
This comment has received too many negative votes show
probably the most boring sixty symbols video
kruset11 1 year ago
someone get this man some supercomputing time ASAP.
TigerSlashX 1 year ago 108
Has the professor featured here considered distributed computing?
A BOINC application (server, application, etc ...) and about 1000 users could return solutions on a dayly basis.
eyreland 1 year ago 3
are there 3 dimensional matrices?
uut0 1 year ago
@uut0 I don't believe it is really necessary, we write on 2 dimensionally. But if you are talking about a matrix denoting dimensions, yes. You can have as many as you wish, he even mentioned a 10 x 10, which could because 10 dimensional.
BoondockDeacon 1 year ago
ah k i see thanks :D. that would probably take up lotsa computing power to find the eigenvalues for those :O
uut0 1 year ago
I am an amateur physics enthusiast at best. That having been said, is there any relationship between eigenvalues and the term eigenstate which is commonly used in quantum mechanics?
wendighoul 1 year ago
I took linear algebra but i can hardly remember what eigenvalues are :(
VanillaShoelace 1 year ago
What do you reckon the sizes of those matrices are?
drokles 2 years ago
could someone explain me what exactly he searches with his 7 PCs ?
Moondye7 2 years ago
@Moondye7 he calculates
jelloshots5 1 year ago
I wish I had seen this video before learning about them, certainly would have studdied harder on these!
engjds 2 years ago 4
Surely you can get every computer on the universities network to join in during night time and idle periods, like 3D studio MAX's back burner.
Cusk0 2 years ago
Most university do this.
Reading University uses the CONDOR system to number crunch on unused pcs at night.
Skiddings 2 years ago
Setup a Boinc account and get thousands of private users to help crunch the data.
TEMPESTSPARK 2 years ago
I HATE MATRICES, ESPEcIALLY EIGEN VALUES!, HMP.
roendm 2 years ago
first born value
rabbitsib 2 years ago
I fear I study Eigenvalues and Eigenstates in later years:P...
Should be good lol:D
Great video:D
MichaelERutherford 2 years ago
The basic idea of eigenvalues is very simple. It's certainly not something to be intimidated by.
200609 2 years ago
I'm not actually scared or intimidated by them... Was just being daft:P
MichaelERutherford 2 years ago
Thankyou very much for these videos :)
nhojmabon 2 years ago
Eigenvalue
composed by 2 words: eigen and value
In Dutch EIGEN means own , in the meaning of: my own value = mijn eigen waarde
Aythamie 2 years ago
"I am speaking to you from El Paso. That's Spanish for 'the paso' " - Bush impersonator
PrivateSlacker 2 years ago
and it's the greek l
robotwarsfan03 2 years ago
wow this video is really good so glad to see the physics department getting a channel. we did eigenvalues a while ago in maths- good to see it in use!
neen222 2 years ago
just buy more memory, it's cheap!
danthemango 2 years ago
For something like that, memory would not really affect it too much. Processing power would what you would need more of.
bizzawizz 2 years ago
whatever, I don't really understand what he's doing so I'm not going to be much help anyways.
danthemango 2 years ago
ya, the half life symbol is actually used in chemistry for the half life of an atom of some kind of element. i.e. Mercury-196: Stable. But something like Mendelevium-258 has a half life of 51.5 days.
blicblak 2 years ago
I'm so glad that the Physics department in Nottingham has an output on YouTube as well!
cagedkiller360 2 years ago
Buy some several Playstation 3s. There are instructions (on the internet) on how to turn several (I can't remember how many you need) of them into a supercomputer that I would imagine would increase the amount of information you can process.
mhuycke0111 2 years ago
still, the playstations have a low processing power to price ratio.. probably a core 2 quad would beat all those computers he's using.. but only if he could program it in a way that it used the 8 processors..
ricardjorg 2 years ago
You can easily set up a cluster computer with Linux and several networked computers. A PS3 would be a nightmare to program for.
saintaureus 2 years ago
True
ricardjorg 2 years ago
PS3 in a general sense may have relatively low processing power, but like all dedicated graphics chips they are hugely superior to general intel or AMD chips at certain types of mathematical calculations, often the sort of calculations needed in research, hence why many researchers around the world have switched renting out time on supercomputers for having their own "supercomputer" made up of PS3s (at a fraction of the cost and often with a better performance)
ywrtywto09 2 years ago
I Agree with you PS3 have low processing power typically there really good at math certain math equations and rendering (Folding @ Home) due to cell technology but Eigenvalues are really interesting if you study into it ;D
tech33012 2 years ago
Price pr calculation wice (both aqusition and powerconsumtion) new GPUs are more efficient than a PS3. I myself own a GPU wich can do 2,4 TFLOPS (= Terra Floating Point Operations Per Second. 2400 billion operations per second with decimalnumbers). If this professors work could be made into parallell code, a graphicscard costing £100 could outperform all those computers easily. By clustering high-end cards in those machines he could do a years worth of computing in a day to a week.
gulllars 2 years ago 49
@gulllars In esssence a single pimpmachine could pay itself off in a matter of weeks compared to the shit these ones haul.
Vennificus 1 year ago
@gulllars Sounds unlikely, since he's working with chaotic systems, and precision on GPGPUs is horrible. And if you use double-precision on the GPU, most of the speed advantage vanishes. (and even then the precision might still be too poor)
whoppix 1 year ago 6
@gulllars
The power of DSPs. Solving complex equations by manipulating signals and analyzing the results. The speed and efficiency of modern DSPs is what gives GPU computing its edge.
douro20 1 year ago
@gulllars ima guess ur talking about a high end 5000 series from ati? i got something like dat too ;) i didnt quite get the gist of the calculations, but that would only work if the calculations arent linear. ie the next calculation doesnt depend on the previous result. if many calculations can be done in parallel, then yes, a couple GPUs could do the work an order of magnitudes more quickly than his set up there.
dandil 10 months ago
@dandil a year ago, i had a 4870x2. Now i have a 6850, since it's sufficient for my current gaming, and i got tired of the noise.
To harvest GPU FLOPS require highly parallell problems, or better, embarresingly parallel problems. You need high thread level parallelity (and instruction level parallelity too helps on ATI/AMD).
I can see chaotic systems being difficult to optimize for GPGPU.
Things like Ray-Tracing, rasterization, and biologic, geologic, or astronomic prosessing can use GPGPU well.
gulllars 10 months ago
@gulllars Yes, but (1) parallel calculations are more in the marketing realm than they are of practical use, and (2) the programs used for generic math simulations are hard pressed to be even compiled for the right processor. One has to be more of a programmer than a physicist to leverage those GPUs.
ZdenekJindra 10 months ago
@ZdenekJindra the latest generations of GPUs support OpenCL, which is not that different from C+, and support hardware agnostic massive parallell execution.
Anyways, for embarrassingly parallel workloads, GPUs offer unparalleled performance pr Watt and £.
gulllars 10 months ago 2
@gulllars I never denied that, but if one has invested years of experience into existing tools and isn't a computer geek, no GPU will happen.
ZdenekJindra 10 months ago
@ZdenekJindra if you could get grad students to re-adapt your code to OpenCL as a project, and keep the front-end or GUI similar, it could be done fairly transparent so the researcher only has to adapt to minor changes in the procedures at a high level, and multiply output several times, maybe orders of magnitude depending on workload.
Just a thought.
gulllars 10 months ago
@gulllars I do not know about his work but I work with fluids and Computational Fluid Dynamics and in it, for research purposes, a computation that takes about 200h in the fastest supercomputers is considered acceptable.
mignik01 5 months ago
Come on down to Lambda's -- for Eigenvalues in every aisle!
johnclavis 2 years ago
lol
Odjii 2 years ago
maybe eigenvalue means selfvalue
marylov 2 years ago
Synonyms of Eigen in German
Own, Separate, Individual, Proper, Inherent, Peculiar, Typical, Strange
Which pretty much agrees with the mathematical meanings of eigenvalue and eigenvector
jestertru 2 years ago
Eigenvalues find their application also in quantum physics!
For instance to obtain the energy of a quantum system you take the wavefunction of the system, then you apply an operator, which can be very complicated, and the result is the same exact function, multiplicated by a factor, which is the energy and is an eigenvalue.
xja85mac 2 years ago