Stories
Slash Boxes
Comments

News for nerds, stuff that matters

Slashdot Log In

Log In

Create Account  |  Retrieve Password

Water Computing

Posted by michael on Fri Oct 25, 2002 09:44 PM
from the wetware dept.
Andrew_Cronin writes "This is a nice project that some one did at MIT on building some logic computation systems without using electrons.. So why not use water..."
+ -
story
This discussion has been archived. No new comments can be posted.
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
 Full
 Abbreviated
 Hidden
More
Loading... please wait.
  • by packeteer (566398) <packeteer&subdimension,com> on Friday October 25 2002, @09:45PM (#4535095)
    Doesn't anyone know? Water and computers dont mix. Make up yuour mind. You can either make fun of water-cooling OR make computers out of water.
    • clean off your motherboard real good with rubbing alchol. (turn computer off, discharge all capacitators and DOUSE it with rubbing alchol) then what you do is get distilled water... and ofcourse you then poar that all over your computer whilest its running. i did this for a speach class. everyone was in shock... after the speach teacher yelled at me i was then given the only a.


      so what were you saying?

    • Water and computers dont mix

      You're right. Only Aqua and Macs mix.

      Oh wait....
    • Keynes (Score:3, Interesting)

      The idea predates the electrical computer. In the 1930s John Maynard Keynes had a water computer that was used to construct a model of the British economy. The computer stretched over a large country mansion.

      Konrad Zues' Z1 and Z2 machines were built somewhat later but used many of the same ideas, only in a much more compact space.

      Of course now we will have a bunch of idiot libertarians blasting Keynes. However Keynes and his computer are the reason why Britain pulled out of the depression before the war while in the US depression turned to slump. The problem came when Keynsianism became an idelology after his death, the solution to every problem was deficit spending, just like today some idiots think that the solution to every problem (including a deficit) is tax cuts.

  • by EvilCabbage (589836) on Friday October 25 2002, @09:46PM (#4535099) Homepage
    .. how long until we see a computer constructed using bong water?
  • by fmaxwell (249001) on Friday October 25 2002, @09:46PM (#4535103) Homepage Journal
    Here's the real aqua!
  • by srhuston (161786) on Friday October 25 2002, @09:48PM (#4535106) Homepage Journal
    Brings new meaning to the term "Bit bucket"
  • I think it is a cool concept. EMP wouldn't be such a problem. Lugging the thing to a LAN party would. Imagine having to carry Hinkley and Schmidt as well as the device.
  • So... (Score:5, Funny)

    by ackthpt (218170) on Friday October 25 2002, @09:48PM (#4535108) Homepage Journal
    If you overclock it, can it cool itself?
  • Will we use electricity to cool it? Well water is a very effective cooler on electronics, so why not do it the reverse when your water is your 'electronics'?

    ZZZZZZZZZZZZZZzzzzzzzzzzzzzzzAAAAAAAAAAAAAAAPppp pp pppppppppp

    YOUCH!!!!!

    DAMN ME!!!!! THAT !*!@# HURTS.

    Scratch that idea.
  • by wilburdg (178573) on Friday October 25 2002, @09:49PM (#4535112)
    He says without electronics, not without electrons... Last time I checked a molecule of water had, oh say about 10 electrons in it...
        • by wirelessbuzzers (552513) on Friday October 25 2002, @10:37PM (#4535294)
          How can people argue about this for so long!?!? If you really know nothing about chemistry, don't post about it.

          Oxygen is element 8. It has 8 electrons. 6 of them are valence elecrons (in the 2s and 2p orbitals) and 2 of them are "core" electrons in the 1s orbital. Only the valence electrons get drawn in those dot diagrams, that's why you can only see six on some sites.

          Hydrogen is element 1. It has 1 electron. This electron is in the 1s orbital.

          Water is H2O, where the 2 is subscript. It has 2 hydrogens and one oxygen, with polar-covalent bonds between them, so there are 2*1+1*8=10 electrons. Two of them are in oxygen's 1s orbital, four of them are in two of oxygen's four sp3 hybrid orbitals, and four of them are shared between hydrogen's 1s orbital and the other two of oxygen's sp3 hybrids (one orbital and two electrons for each hydrogen).

          Don't even get me started on sp3* anti-bonding pairs.

          Sorry for being inconsistent as to whether numbers should be spelled out.
        • Except, of course, that assertion involves simplifying assumptions, too.

          We could go as far as high school chemistry and decide that there are 2 core 1s electrons definitely associated with the oxygen, plus four more electrons that are part of lone pairs on the oxygen (probably also mostly belonging to the oxygen atom). Then there's four electrons involved in the two sigma bonds joining the oxygen to the hydrogen. Simplest story is that it shares two with each hydrogen.

          If we break out the molecular orbital theory, then it starts to get kind of messy. At the lowest level, we have a really ugly n-body problem. We can't solve the Schrodinger equation analytically for this case, so we're limited to approximate numeric solutions. (Technically, we should really account for relativistic effects and use the Dirac equation, but that's overkill for lightweight atoms like these.)

          Even then, solving for the wavefunctions by whatever method only gives us a probability that electrons will be located nearest a given atom. In principle, occasionally all 10 electrons could actually be closest to one of the protons, but you would have to wait a loooooong time for it to happen.

          Oh, yes--if I wanted to be picky, I could also mention that pure water will still undergo spontanous autoionization to form H+ (H3O+, actually) and OH- ions, containing the same number of electrons, but now the wrong number of protons...

          One more, then I'm done. Liquid water actually tends to get kind of clumpy. In the so-called 'flickering cluster' model, water molecules in the liquid phase form short-lived hydrogen bonded clumps containing several (or even several tens of) water molecules. These clumps have an electron count that depends (of course) on their size.

          Water is actually an incredibly interesting beast, chemically speaking. We take it for granted because it is ubiquitous, but there is a tremendous amount of very interesting stuff that it can do.

  • by Lobsang (255003) on Friday October 25 2002, @09:49PM (#4535114) Homepage
    Either that or the Slashdot crew already crashed the "Water Computer". :))
  • by EvilAlien (133134) on Friday October 25 2002, @09:51PM (#4535128) Journal
    ... its half full of beowulf clusters of water computing power.

    I couldn't resist, sorry.

    As we get closer and closer to microscoptic or at least very small computers, how long until the inevitable complex systems of neural nets combined with tiny computing devices self-organizes into something with capabilities outstripping an expert system... and into something like SkyNet in the Terminator movies?

    The self-organization of a complex system into a self-aware artificial intelligence is a chaos theory wet dream.

  • by dillon_rinker (17944) on Friday October 25 2002, @09:55PM (#4535149) Homepage
    Scientific American had an article about water-based logic gates and circuits some decades ago. IIRC, they even created circuits that had no electronic analogy. I can find no reference to this on the web - perhaps some other science geek with access to a complete collection could find it. I believe it was in the Amateur Scientist, but it's been about fifteen years since I read through the stacks of magazines in the cabinets of my chemistry classroom, so I could be mistaken.

    For a similar concept (ie, non-silicon machine logic) that I first read about in the pages of Scientific American, check out the Apraphulians here [rustrans.co.uk]. For more info on this ancient race, Google is your friend. [rustrans.co.uk]
    • by friscolr (124774) on Friday October 25 2002, @10:10PM (#4535202) Homepage
      In "The Way Things Work" (1988) David Macaulay also illustrates logic gates as plumbing in a building. It's on page 333 of the original book; the newer version even has an interesting rant about Bill Gates.

      i always liked old school water clocks.

      i seem to recall something about logic gates or some sort of logic being built out of matchboxes and beans. it played tictactoe, deciding the best move by plopping out a bean of a certain colour? i can remember neither the details nor the source.

        • i seem to recall something about logic gates or some sort of logic being built out of matchboxes and beans. it played tictactoe, deciding the best move by plopping out a bean of a certain colour? i can remember neither the details nor the source.
          I've seen it in a book by Martin Gardner, the game was called Hexapawn. A quick search on google should turn up more details.
          Interesting! In short: it's an exhaustive search of all moves from black's perspective whereby all moves are initially possible and a loss causes the last move to be removed as a choice. I found the info here [mactech.com]. There's more at the site including how to implement it programatically.

          Here's their explanation of the matchbox version of the game:

          Gardner's machine is implemented as a set of 24 matchboxes, one for each possible board position when it is Black's move. Each matchbox has pasted on it a drawing showing this board position, as well as all possible moves from that position, drawn in different colors. Inside each matchbox are several colored beads, one for each move on the top. When it is the machine's turn to move, the human operator finds the matchbox showing the current position, draws a bead at random from the matchbox, replaces it, and makes the move thus chosen. The machine learns from its losses: when it loses, the operator removes and discards the last bead drawn. This ensures that the machine will never lose in this way again.

          To keep this on topic: this game could be implemented with a water computer, too. Replace the matchboxes with different colored buckets of water. Instead of removing a bead for a loss, dump out the corresponding bucket.

          For a more enjoyable game, replace the buckets of water with shots of beer. =)

    • Speaking of logic, here's a October 1989 Scientific American article detailing the tinkertoy tic-tac-toe playing machine [rutgers.edu]. Anybody want to make this out of these water logic components? Didn't think so... ;)
    • The "apraphulian computer" article in 1988 was an April Fool's joke. A.K. Dewdney has a history of presenting interesting scientific concepts as fiction. You can find examples of this style of presentation in his book Planiverse [amazon.com] as well as some of the earlier corewars articles in Scientific American IIRC.
  • Very good (Score:5, Interesting)

    by Istealmymusic (573079) on Friday October 25 2002, @09:58PM (#4535163) Homepage Journal
    I remember of Neal Stephenson's Cryptonomicon, where when L.P.Waterhouse was walking amongst the beach at sundown, in a starlit avenue, and he drew lines in the sand. This was inspiration for barcodes, but that's beside the point. The point is Stephenson had the narrator narrate L.P.Waterhouse notice how the ocean is a Turing machine, it interacts with the sand, due to certain mathematical fluidity properties, to leave an indentation of predictable properties.

    I've considered water-based computation long ago, but hats off to this student for logic design and implementation. My idea was to have water push open another wate gate, much like a flow-controlled valve, allowing for a water-based transistor. Combine this with other transistors, and you can build virtually any gate--I take that back, any gate you want. XOR and AND are good choices, as with a XOR a you can get NOT, to make a NAND, and as we all know NAND is the Univesal Binary Gate.

    • NAND is the Univesal Binary Gate.

      In principle, couldn't XOR be used to construct all the other gates as well? Who needs a NAND explicitly? You can make an AND from XORs. An as stated, you can also get a NOT from XORs.

      Yes, of course you can build up all the NANDs you need from XORs, and then use those composite gates to build the rest of the gates--but it's more efficient to skip the middle step of constructing the NANDs.

  • Who else tried this? (Score:5, Interesting)

    by thekernel32 (240428) on Friday October 25 2002, @09:59PM (#4535167) Homepage
    I really remember seeing a documentary on how the russians tried using water logic for computations and got pretty good with it back in the 40's. Then again they did alot of cool stuff over in that part of the world. They actually managed to stick with 50's technology for 40 years. yay communism! I knew someone who had no more after the berlin wall came down just because his job was dependent upon the crappy stuff they made breaking down. Reliable products from the west caused alot of people to loose their jobs.

    I know it's an off topic rant, wanna give me some points for being interesting anyway?
  • by Cheetah86 (136854) on Friday October 25 2002, @10:01PM (#4535175) Homepage
    <sarcasm>Looks like we "flooded" the server...</sarcasm>
  • by 7-Vodka (195504) on Friday October 25 2002, @10:03PM (#4535182) Journal
    But the thought just came to me of using beer instead of water. That way you could make a beer computer! Who says beer makes you shite at math? you CAN get drunk and still do 4 bit additions :)

    Hell, take that thing on a pub crawl and have your beer do it's own calculations of how much tip you should leave as it's on it's way down to your stomach!

  • by matman (71405) on Friday October 25 2002, @10:37PM (#4535295)
    I wonder if it would be possible to create such gates that function on water pressure changes. It's quite quick to propigate a wave through water. I'm talking about pressurizing water inside of the system, and then inducing shock waves from your inputs. Could you make gates that trigger on those pressure changes? You could probably get quite a few bps (by creating shock waves in the water) I wonder if you could just use plain old speakers to generate sound waves... I wonder what the attenuation characteristics of water are.
  • Why not extend the metaphor...

    Inductance is the same as momentum. You could build a gadget that has a turbine in the water flow with a fly wheel attached. The gadget would resist water flow starting up, and would resist the water flow slowing down once it's moving. (same as an inductor fighting a change in current)

    Capacitance is the same as a flexible membrane across the pipe, which will transmit AC changes in pressure, but not DC.

    You could build a capacitor/inductor tuned circuit that either filters or passes certain frequency water waves

    Also, water transistors should be fun. A small flow or pressure of water controls a larger flow or pressure (in either an analog or digital fashion)

    It would be a way fun tool for teaching electronics.

    • Actually in Russia (Score:5, Informative)

      by WetCat (558132) on Friday October 25 2002, @11:47PM (#4535523)
      ... water analog computers have been used since 1949... till mid 80-x for modelling differential equations.
      They were used for large-scale projects, such as modelling of water dams.
      • In the London science museum they have an analog water computer that represents the British economy. I believe it predates 1949. I am not sure how accurate a model it was, but it cant be much worse that the digital model they have now.

        Reseach shows the more higly qualified an economist is, the poorer his predictions!

  • by Alsee (515537) on Friday October 25 2002, @10:51PM (#4535349) Homepage
    Every time I try to overclock it, all the circuits freeze up.

    We're thinking of giving up on liquid nitrogen and trying liquid helium.

    -
  • by Alien Being (18488) on Friday October 25 2002, @11:05PM (#4535401)
    better pipelining.
  • a water adder (Score:5, Insightful)

    by drDugan (219551) on Friday October 25 2002, @11:37PM (#4535498) Homepage
    a friend of mine in college (Aron) made a water adder capable of adding two 8-bit values -- all with water streams.

    LINK [bowdoin.edu]

  • by Chagrin (128939) on Friday October 25 2002, @11:51PM (#4535532) Homepage
    Didn't we just have a different "font" article?
  • by Mark Garrett (607692) on Saturday October 26 2002, @01:29AM (#4535756)
    ... but I bet that in a few months, it'll all just turn out to be vaporware.
  • by Animats (122034) on Saturday October 26 2002, @01:39AM (#4535773) Homepage
    Fluidic logic [si.edu] has been around since the 1960s. It's often used in industrial process control. It works for both air and liquids. Most industrial systems use air logic, but automatic transmissions often have fluidic logic running on hydraulic fluid.

    The MIT students didn't quite get it right. Their gadget doesn't seem to have gain. The key insight needed for fluidics is that a jet of fluid can be diverted with a smaller jet coming in from the side. This allows building a fluidic amplifier.

    Once you have an amplifier, you can do switches, gates, flip-flops, and other logic elements. Analog control systems, with fluidic sensors and amplifiers driving pneumatic or hydraulic cylinders, are also possible. When the inputs and outputs are pneumatic or hydraulic, it's often convenient if the control system is, too. Fluidic elements are very reliable, too - there are no moving parts except the working fluid.

    One wierd fluidic application is this kosher public address system. [216.239.53.100]

  • by haggar (72771) on Saturday October 26 2002, @03:04AM (#4535908) Homepage Journal
    (I say "calculators" because I think that computer is underrated).

    I have studied about these machines at Uni (I studied in eastern Europe): they use fluids and analog pseudo-circuits to create things like integrators, adders etc. and are capable of solving systems of differential equations in real time. This kind of equations is still a non-trivial problem for digital computers.

    However, with the advent of gigabit-clocked CPUs, these machines are definitely out. Their models are, sometimes, replicated in software, though.

  • by rwa2 (4391) on Saturday October 26 2002, @10:01AM (#4536647) Homepage Journal
    I saw a guy do a presentation on this during an ASME conference. The Russians have been using the same technology with air rather than water to construct logic gates and op amps for missile controllers and such. I suppose the benefits were not needing electricity - just run the thing off of compressed air, which is readily available in a missle.

    Actually, the main benefit was that it was no longer susceptible to electromagnetic interference, so you could build very noise-free amplifiers.

    The samples that the guy had were much smaller... about 2-3cm squre wafer-thin metal plates. Each plate had a pattern on them which were machined out to form components. Circuits were built by "sandwiching" a series of plates together, so the outputs from one component could feed into the input of the next component. The resulting devices would be several cm long consisting of dozens of components bolted together with long stringers.

    The theory looked about the same as the MIT guy's device, though. I think the Russians developed most of the counterparts to electronic components, though - various transducers and transistors. I'm not sure how they would have done something like a diode,though...
    • by Anonymous Coward on Friday October 25 2002, @10:12PM (#4535211)
      how to implement other operation: OR, NOT

      Simple. The author gives XOR and AND gates, formed from joining two streams together, without and without a control. (See the article for details, I haven't taken the time to look into it very deeply).

      Anyways, XOR's function number is 0110. Split it in two, and you get "A(01) when B=0, NOT A(10) when B=1"--two unary gates formed a binary gate. Split AND's function, 0001, and you get "0 when B=0, A when B=1". Trust me, this is easier than it seems. The unary gates are: 00=0, 01=B, 10=NOT B, and 11=1. Now that we got that out of the way:

      • 0 XOR A = A
      • 1 XOR A = NOT A
      • 0 AND A = 0
      • 1 AND A = A

      None of those are useful except 1^A=!A. We need NOT to complete our library of functions too. Now we can combine it with other gates like so using Boolean Algebra:

      NOT(A XOR B) = A XNOR B
      NOT(A AND B) = NOT(A) OR NOT(B)DeMorgan's Law
      NOT(A AND NOT(B)) = NOT(A) OR NOT(NOT(B)) = NOT(A) OR B
      And now, ladies and gentlement, I present to you, The OR Gate:

      not(not(a) and not(b)) = not(not(a)) or not(not(b)) = a or b

      This is constructed from: 1 xor ((1 xor a) and (1 xor b)), and of course the 1 is simply a constant flowing stream of high-power water. And obviously, since NAND is a universal gate, this can be done like so:

      1 xor (a and b) = a nand b
      NAND can make any gate, including NOT, which is then combined as we saw above to form OR, NOR, XOR, XNOR, and even inhibitation and implication if you please.

      Did that answer your question?

      -jc

    • > 1) how to implement other operation: OR, NOT,

      well, following on the system he started you can probably get the effect of OR or NOT by altering how the "computer" reacts to the outputs, using the existing gate.

      You could make an OR gate by wiring the two outputs together. Get water in either jet, and you get a 1. Put water in both jets and you get a 1. Put no water in either jet, you get a 0.

      A NOT is just as simple, except you need a constant jet feeding through the gate. No water in the other jet means a 0 converts to a 1, water in both jets dumps into waste and creates a 0.

      So there you have it.. by tweaking he inputs/outputs of the single existing gate you can create pretty much any conditional you desire.
    • a not gate: an xor gate with true -- ie a constant stream -- applied to one input

      an or gate: an and gate with not gates at each input and output -- or just join two streams with a simple junction

      A recycling system is easily added. A more fundamental problem is that the gates are passive: there is no amplification. You can use gravity, but with feedback some lines will have to go upwards and need a complex pump for each line. Also you may need a lot of height per stage.

      Fluid actuated valves would solve this problem, and would be more efficient. Only one pump would be needed -- as in electronics -- to provide the supply pressure. But in the simple case of a four-bit adder, it might be harder to make.

    • Probably the first "modern"computer, the Z1 (Germany, early 30's) used aluminum moving parts (!)but because of difficulties with these sort of parts, the Z2, Z3, Z4, and Z5 used electronic relays. (Z1 was destroyed by allied bombing).