Talk:Real computation
From Wikipedia, the free encyclopedia
[edit] could a real computer ever be built?
Right now, obviously, (and perhaps this is exactly the intent), the section "could a real computer ever be built" is far from a NPOV. It says that it can't be built out of "wood." Really, there should be some counterarguments in there. Such as "Noone ever expected it to be made out of wood, why would you even mention such a thing?". I'm not saying that a Real computer can be built as is defined here, I'm just saying that currently, the article is more bias than any other wiki article I've seen. Don't you guys have a sense of shame? -- Kevin Baas 21:29, 18 Jan 2004 (UTC)
- I disputed the NPOV of this section with template:POV-section. I very much agree. Scottydude 23:53, 7 January 2007 (UTC)
[edit] Quantized space/time
Some physical theories such as loop quantum gravity hypothesize that space and time themselves are quantized: this would also present major problems for building a "real computer".
I'm not a physicist, but if space and time itself was quantized, surely a perfect analogue computer wouldn't just present major problems, but it would be completely impossible? There'd be no way to input, process, or output infinitely precise values.
Even assuming continuous space/time, while an amazingly accurate analogue system could be created, I don't get how infinite precision could ever be achieved - if there is an argument against this seemingly intuitive conclusion, could it be put in the article?
--82.46.194.14 14:15, 9 September 2005 (UTC)
Well then there's the distinction between analog computation and real computation. analog computation doesn't require or assume infinite precision. And i was just thinking about this the other day: essentially, since a system based on a finite number of logical axioms becomes inconsistent when too many axioms are introduced, a system based on a infinite number of axioms, which an analog system ultimately is, is neccssarily inconsistent, i.e. "imprecise", in that a given input never produces the same output, it's alwasys offset by a litttle. But the question then becomes one of usefullness, or "what is computation?" apart from the conventional definition of computation. Is there a usefull way to "process" information that does not require the same result every time? For instance, in simulating systems that exhibit chaotic dynamics?
Then we come to the question of quantized space time - under the scope of "consistency": if one assumes, from the get-go, that any accurate definition of "reality" must be "consistent"; reproducable, then only those processes that are "computable", by the standard godel/turing definition, and therefore only discrete space/time, is "real" - but this amounts to assuming the conclusion. Kevin Baastalk: new 03:42, September 10, 2005 (UTC)
[edit] "may" enable?
The entry says "Depending on the model chosen, this may enable real computers to solve problems that are inextricable on digital computers". This doesn't clearly state whether there exist models of real computation that are strictly more powerful than Turing machines. But the paper http://citeseer.ist.psu.edu/siegelmann94analog.html purports to give an example of such a model (on page 2, it says they prove that their analog neural net can solve the halting problem in exponential time). So, unless I am misunderstanding that paper, I think that this entry should have the following sentence added, "There exist models of real computation which are provably more powerful than Turing machines (siegelmann and sontag, '94)".
I am not an expert in this area so I won't add that myself, in case I am misunderstanding. Bayle Shanks 20:53, 21 August 2006 (UTC)
