Slashdot Mirror
Chaotic Computing In Practice
codyhess writes "The Economist published a great article detailing efforts to use Chaos in computing - "Speaking at the American Physical Society's annual March conference, William Ditto of the University of Florida told of his efforts to create a 'chaotic computer'."
Dr. Ditto can create standard logic gates (AND, OR, etc) that output a value according the their chaotic threshhold. Different logic operations can be performed by simply changing the threshhold, making an incredibly flexible computer that can perfom different functions instantaneously."
Nothing new, I've been doing that since Windows 3.11! :)
...is already well versed in chaotic computing.
I was just going to hire really crazy programmers, and change their meds based on what I needed.
... I type in random characters in Google and hit "I feel Lucky".
Skynet is being born!! .. or not!!
But I like the idea of logically flexible computers, and fear it at the same time. Sometimes, especially after work, it's nice to come home to something that can think in a straight line.
Google was no help...in translating this article into English.
sarchasm: The gulf between the author of sarcastic wit and the person who doesn't get it.
I first read that as Catholic Computing.
Pearly Gate logic will have to wait a few years yet, I guess.
Check out the date of the article - April 1st...
I am not a physicist, engineer, or scientist (or anything else qualified to answer this) but it seems to me, the simple minded one, that once you start controlling something, it isn't chaotic. I mean- if they are basing decisions on this, then it can't be completely chaotic, can it? How can you derive an AND, OR, etc, from chaos without controlling it (thus negating the chaos). Can someone dumb this down a little for those of us who aren't in the know?
Celebrate Steak and a Blowjob Day!
analog computers of old. IIRC they were used for ballistics calculations and similar by the military.
Here is an example.
Look into what kind of mathematical operations can be realized with multiplying DACs.
Dr. Ditto
Wait...Rush Limbaugh has a Ph.D?
I have discovered a truly marvelous
i think this paragraph really sums things up. the editor is such a moron as to explicitly state the obvious grammatical correlation between mathematically chaotic logic circuits and the general "chaos" users experience with their computers. and that preceded by a description that sounds like some kind of vampirian (or is it vampirical?), frankensteinian, technological monster. (rob zombie brings you "attack of the chaotic leech borgs"!).
p.s. the chaotic leech borgs would be a good name for a band
The way I see it (although I am not a mathematician), the major hurdle to realizing this is the fact that generating random numbers usually results in patterns.
"Teleporting Rodents with D-Cell Battery Displacement" theory -- IgnoramusMaximus (692000)
Evil Scientist: My clone army will soon be complete!
Secret Agent: Not so fast, Doctor Ditto!
Reading this article reminds me of the Improbability Drive in the Hitch Hikers Guide to the Galaxy.
The last problem to be solved was to find a perfect source of randomness, which the galaxies best professors had been trying to solve for decades as whole departments had been built up on trying to solve this problem. Then one day, a brilliant student solves the problem by realizing a a cup of hot coffee provides this data. He is immediately awarded the highest Physics prize in the universe, and immediately lynched by his peers for being a smart-ass.
Couple of thoughts:
This isn't quite the same thing as having randomly perturbed input thresholds, which is how neurons work. And, as anyone who's tried it knows, neurons are only about 95% efficient in determining the correct result. It takes a lot of logical processing on top of the neural bitwise decisionmaking to distill the 95% to the 99% or so correct answer rate that constitutes "intelligent thought".
And, they'd better look into real-world noise margin requirements for thresholding electrical switching decisions, or "chaotic" is all their output will ever be.
Basically, I invented a simple but mind-blowingly fast algorithm for solving complex equations:
#include <stdlib.h>
double solve(void) {
return rand()
}
Sometimes, it will give you a root of x^2 - 7; other times, value of pi or phi. Once it even gave me the answer to the meaning of life, the universe and everything!
Chaotic systems are actually quite controlloable in a very interesting way. The key property that makes a chaotic system so unpredictable is divergence -- if two copies of the system differ by delta, that delta will grow exponentially in time (doubling according to a coefficient call the Lyapunov coefficient). Yet, the divergence is never arbitrary. Instead, the divergence in chaotic systems happen within a space called the strange attractor - the diverging trajectories stay within in the attractor zone even as the split from each other.
If you map the strange attractor and nudge the system are the right point of the cycle, you can push the system into what ever mode of behaviro you want. Although you cannot predict the longterm behavior of the chaotic system, you can perturb it periodicaly to stabiize it or rapidlly shift its behavior. Scientists are looking at how to use this chaotic control theory to control unstable systems such as ultrahigh power lasers, manuerable jet aircraft, and heart tissue.
The key controlling a chaotic system is to understand how the chaotic system diverges (the shape of the strange attractor) and use that knowledge to deftly inject perturbations at just the right moment.
Two wrongs don't make a right, but three lefts do.
or is he just the 'friend' of this guy?
The only thing I want to know is; are these computers Chaotic Lawful, Chaotic Neutral, or Chaotic Evil?
I hear there's rumors on the Slashdots
this is "self modifying hardware", similar to "self modifying code". but is it fair to call it chaotic? In a chaotic system, the process remains the same but the output varies. In this situation, both the process and the data change over time. Or can a chaotic system also be one where the rules change as well?? Experts??
personally, SMC is a bitch to debug, I can't imaging how one would begin to debug THIS beast...
https://www.accountkiller.com/removal-requested
But this would still be limited to the confines of the Church-Turing hypothesis. Fundamentally, it doesn't seem to be a stronger model of computation. Even quantum computing is a different model than the Turing machine, however it is not yet known if it is strictly stronger. At a first glance, this just seems to be a novel way of making reconfigurable circuits. But can it beat the Turing machine? I doubt it...
You wish.
To protect the perpetrator I won't mention his name, but here's a warning about people developing off in a corner, by themseleves rather than collaborating with their peers.
I worked for two years at one job before learning there was another programmer (besides the other two I worked with.) The group I worked with remained within the same office or no more than a room away and we frequently bounced ideas off each other, creating some damn fine products (if I do say so myself.) The other guy, actually a personal friend of the director, always worked on his own. When he retired and I inherited his work I was truly pissed off. The code was horrible and reflected the skills of a novice (a poor one at that) and was littered with GOTO statements and demonstrated a severly retarded understanding of documentation, coding style (i.e. 3000 line for-next loop with GOTOs out and back in again) and zero knowledge of library functions, which would have cut hundreds of lines from the code. (Since the code would be replaced by a full system a year later, all I had to do was just keep it running and fix corrupt data, which was frequent.)
You might get the impression that the lone coder was chaotic, but you would have it backward. His procedure was orderly, straight forward, rarely diverging from his approach or skill set. The three (of which I was part) was Chaos -- we thought outside the box, tried things, introduced new approaches to old problems. Where we once would say, "no, that can't be done", we went to, "Yes, that can be done, and has, furhter, it's more useful and versitile than you ever imagined."
Seize chaos, it's your real friend.
A feeling of having made the same mistake before: Deja Foobar
...adding a good/evil axis to computer alignment. Because otherwise, if you get a chaotic computer, how do you know whether it's chaotic good, chaotic neutral, or chaotic evil?
Better to have a computer with a good heart and a general distrust of authority than one which wants to enslave everyone and reduce the world to a desolate wasteland.
It is the hardware side that uses chaos, not the software. Details are sketchy in the article, but I believe they are looking at chaotic systems and tweaking the hardware to use different regions of behavior depending on the desired use.
And, yes, there are reasons we're not all programming in LISP.
Things to do today: See list of things to do yesterday
We're already seeing quantum computing, as this story is in two places simulataneously. Remember, you saw it here first, and second, on Slashdot.
Dude
Did you check the date on that Economist article.
Think of it this way:
Imagine you could watch two one-hour long TV shows simultaneously superimposed onto each other on the same TV (and understand both shows seperately.) Now imagine you have have two TV with the same capability. Now you can watch four shows in one hour. This is the essence of this computing theory: you can do more calculations in less time but not in the normal computing sense. I prefer to think of standard binary computing as a direct derivative of quantum computing, much like velocity is a derivative of acceleration.
The chaos theory simply describes the elements that are involved in forming these calculations. This would directly affect a computer's ability to multitask - instead of a data flow going in a straight line and different parts of the processor performing different operations on it, a function could be self-contained and processed recursively, with the data passing through the same matrix, while that matrix changes itself to perform different functions. In theory, this could take far less ticks, increasing speeds exponentially. Add more matrices to this and you can see the benefits.
The idea behind his work is to be able to control the input into these "chaotic elements" thus producing a predictable and reproducable output. A true 'quantum leap', if you will. This would be a significant jump in computing technology, skipping over "trinary" computing altogether.
Humans don't think in straight lines, why should computers? Then again, I could just be reading it wrong.
From the Lockheed Martin Analog computing link "A picture of a GEDA center showing (from the left) an R-2 unit, two L-2 units, (maybe) an N-2 unit behind the woman, (maybe) two L-1 units and another recording unit between the women."
Perhaps because they were hoping to do some gradual revisions and eventually reach Windows 3.14?
Yes, digital computers may use analog signals, but the basic operations of a digital computer (AND, OR, XOR, etc.) are fundamentally digital operations. They quantize the analog signals into 1s and 0s, and output quantized signals based on those digital values (of course, with some amount of analog error).
An analog computer does no such thing. If it wants to add two signals, it adds them. In analog. You can do integrals and derivations in analog as well, amongst other things.
A digital computer may have to use analog signals to operate on some level, but that does not make it an analog computer.
manuerable jet aircraft
Eww, why would someone want to fly in that?
// file: mice.h
#include "frickin_lasers.h"
They have also made a logic element out of a pair of leech neurons (nerve cells from blood-sucking worms) placed on a microchip.
I remember for my System's Analyst and Design class my teacher mentioning how they were already wiring organic matter to computer chips. One unfortunate student who made the great mistake to vocalized his complete shock over this, from which this cynical and suggestive instructor bluntly responded to him, "Read a book!" Mind you this particular student had the appearance of a squeegee-kid roadie.
Although I was not the one told off by the teacher that I should enlighten myself,I am very amazed over the use of organic matter is even possible. Are they running some sort of voltage through these fibers? Do they obviously react differently than say a copper wire, and why? Will it be just a matter of time before some medical students at a frat party get bored and hook up a cadaver's brain up to laptop to string search and download what the deceased use to listen to?
Some aim to please, I aim to tease.
That's too narrow definition of chaotic system, because Lyaponov coeefficients and strange attractors realted only to dynamical systems wich have a toplogy - that is some underlying continuity. However there is another type of object which exhibit chaotic behavior, though only in infinite areas - discrete objects like cellular automata, which have no notion of divergence, and discussed more in term of complexity This chatic computing idea is in fact related to cellular automata. Cellular automata is a perfect example simple, completly deterministic discrete system, which behavior very difficalt and sometimes impossible to predict
Here is a paper that describes using chaotic gates as "universal gates".
with multiple images. Years ago a company I was working at had an FPGA that could store 4 images and switch between them every clock cycle, no chaos required.
For those who don't know, an FPGA is a flexible computer chip. Imagine a motherboard full 100,000s or millions of solid state "glue logic" gates that could be re-aranged by little elves repeatedly, and that's an FGPA, but larger, and less expensive. You could build an 8088, then a DSP, then a fast FFT, a converter, then a crypto processor, whatever. Creative uses them on some soundblasters so the hardware (yes, the hardware) can be upgraded ith more features in the future. On mine they added a few digital effects and the ability to handle another few hundred MIDI voices.
There are some problems with the article: it makes claims that aren't backed up. So what's new on slashdot? Anyway, here are the gory details from my point of view. The original source reference appears to suffer from the same problem.
The gist of the new idea is a clever way to create a special type of gate whose dynamical threshold value can be modified to implement one of several possible logic gates. An interesting idea, but not computationally revolutionary. These gates would still implement the same chips we use today.
Now, the article goes on to claim that there is a wonderful new horizon of modifiable computation. I see a lot of words and no details. How are those modifiable threshold levels in these gates stored, anyway? Don't tell me it's with something like a flip-flop. It would be asinine to need 6-8 gates to store each bit of the modifiable threshold value for one "chaotic" gate.
Also, there's the small problem that we can MODEL any type of strange new computational paradigm and have been able to for years. We're no closer to a replacement for Turing-style computation than we were decades ago. I've seen one paper about Analog computers being able to compute some esoteric set of functions that discrete computers can't touch, but I haven't seen anyone explain how this helps in any useful way.