没有合适的资源?快使用搜索试试~ 我知道了~
《Simulating Physics with Computers》
需积分: 9 19 下载量 89 浏览量
2009-12-20
17:34:14
上传
评论
收藏 1.13MB PDF 举报
温馨提示
试读
22页
费曼的《Simulating Physics with Computers》,量子计算机这个概念应该就是从这篇文章提出的,UCBerkerly的一门课程的阅读材料,课程网址:http://www.eecs.berkeley.edu/~christos/classics/
资源推荐
资源详情
资源评论
International Journal of Theoretical Physics, VoL 21, Nos. 6/7, 1982
Simulating Physics with Computers
Richard P. Feynman
Department of Physics, California Institute of Technology, Pasadena, California 91107
Received May 7, 1981
1. INTRODUCTION
On the program it says this is a keynote speech--and I don't know
what a keynote speech is. I do not intend in any way to suggest what should
be in this meeting as a keynote of the subjects or anything like that. I have
my own things to say and to talk about and there's no implication that
anybody needs to talk about the same thing or anything like it. So what I
want to talk about is what Mike Dertouzos suggested that nobody would
talk about. I want to talk about the problem of simulating physics with
computers and I mean that in a specific way which I am going to explain.
The reason for doing this is something that I learned about from Ed
Fredkin, and my entire interest in the subject has been inspired by him. It
has to do with learning something about the possibilities of computers, and
also something about possibilities in physics. If we suppose that we know all
the physical laws perfectly, of course we don't have to pay any attention to
computers. It's interesting anyway to entertain oneself with the idea that
we've got something to learn about physical laws; and if I take a relaxed
view here (after all I'm here and not at home) I'll admit that we don't
understand everything.
The first question is, What kind of computer are we going to use to
simulate physics? Computer theory has been developed to a point where it
realizes that it doesn't make any difference; when you get to a
universal
computer,
it doesn't matter how it's manufactured, how it's actually made.
Therefore my question is, Can physics be simulated by a universal com-
puter? I would like to have the elements of this computer
locally intercon-
nected,
and therefore sort of think about cellular automata as an example
(but I don't want to force it). But I do want something involved with the
467
0020-7748/82/0600-0467503.0£1/0 © 1982 Plenum Publishing Corporation
468
Feylaman
locality of interaction. I would not like to think of a very enormous
computer with arbitrary interconnections throughout the entire thing.
Now, what kind of physics are we going to imitate? First, I am going to
describe the possibility of simulating physics in the classical approximation,
a thing which is usuaUy described by local differential equations. But the
physical world is quantum mechanical, and therefore the proper problem is
the simulation of quantum physics--which is what I really want to talk
about, but I'U come to that later. So what kind of simulation do I mean?
There is, of course, a kind of approximate simulation in which you design
numerical algorithms for differential equations, and then use the computer
to compute these algorithms and get an approximate view of what ph2csics
ought to do. That's an interesting subject, but is not what I want to talk
about. I want to talk about the possibility that there is to be an
exact
simulation, that the computer will do
exactly
the same as nature. If this is to
be proved and the type of computer is as I've already explained, then it's
going to be necessary that
everything
that happens in a finite volume of
space and time would have to be exactly analyzable with a finite number of
logical operations. The present theory of physics is not that way, apparently.
It allows space to go down into infinitesimal distances, wavelengths to get
infinitely great, terms to be summed in infinite order, and so forth; and
therefore, if this proposition is right, physical law is wrong.
So good, we already have a suggestion of how we might modify
physical law, and that is the kind of reason why I like to study this sort of
problem. To take an example, we might change the idea that space is
continuous to the idea that space perhaps is a simple lattice and everything
is discrete (so that we can put it into a finite number of digits) and that time
jumps discontinuously. Now let's see what kind of a physical world it would
be or what kind of problem of computation we would have. For example,
the first difficulty that would come out is that the speed of light would
depend slightly on the direction, and there might be other anisotropies in
the physics that we could detect experimentally. They might be very small
anisotropies. Physical knowledge is of course always incomplete, and you
can always say we'll try to design something which beats experiment at the
present time, but which predicts anistropies on some scale to be found later.
That's fine. That would be good physics if you could predict something
consistent with all the known facts and suggest some new fact that we didn't
explain, but I have no specific examples. So I'm not objecting to the fact
that it's anistropic in principle, it's a question of how anistropic. If you tell
me it's so-and-so anistropic, I'll tell you about the experiment with the
lithium atom which shows that the anistropy is less than that much, and
that this here theory of yours is impossible.
Simulating Physics with Computers
469
Another thing that had been suggested early was that natural laws are
reversible, but that computer rules are not. But this turned out to be false;
the computer rules can be reversible, and it has been a very, very useful
thing to notice and to discover that. (Editors' note: see papers by Bennett,
Fredkin, and Toffoli, these Proceedings). This is a place where the relation-
ship of physics and computation has turned itself the other way and told us
something about the possibilities of computation. So this is an interesting
subject because it tells us something about computer rules, and
might
tell us
something about physics.
The rule of simulation that I would like to have is that the number of
computer elements required to simulate a large physical system is only to be
proportional to the space-time volume of the physical system. I don't want
to have an explosion. That is, if you say I want to explain this much physics,
I can do it exactly and I need a certain-sized computer. If doubling the
volume of space and time means I'll need an
exponentially
larger computer,
I consider that against the rules (I make up the rules, I'm allowed to do
that). Let's start with a few interesting questions.
2. SIMULATING TIME
First I'd like to talk about simulating time. We're going to assume it's
discrete. You know that we don't have infinite accuracy in physical mea-
surements so time might be discrete on a scale of less than 10 -27 sec. (You'd
have to have it at least like to this to avoid clashes with experiment--but
make it
10 -41 sec.
if you like, and then you've got us!)
One way in which we simulate timewin cellular automata, for example
--is to say that "the computer goes from state to state." But really, that's
using intuition that involves the idea of time--you're going from state to
state. And therefore the time (by the way, like the space in the case of
cellular automata) is not simulated at all, it's imitated in the computer.
An interesting question comes up: "Is there a way of simulating it,
rather than imitating it?" Well, there's a way of looking at the world that is
called the space-time view, imagining that the points of space and time are
all laid out, so to speak, ahead of time. And then we could say that a
"computer" rule (now computer would be in quotes, because it's not the
standard kind of computer which cperates in time) is: We have a state s~ at
each point i in space-time. (See Figure 1.) The state
s i
at the space time
point i is a given function F,(sj, s k .... ) of the state at the points j, k in some
neighborhood of i:
s,= .... )
470
Feynman
• •
s~
"
~i
"
• " "$k
oto
~pac~
Fig. 1.
You'll notice immediately that if this particular function is such that the
value of the function at i only involves the few points behind in time, earlier
than this time i, all I've done is to redescrib6 the cellular automaton,
because it means that you calculate a given point from points at earlier
times, and I can compute the next one and so on, and I can go through this
in that particular order. But just let's us think of a more general kind of
computer, because we might have a more general function. So let's tlaink
about whether we could have a wider case of generality of interconnections
of points in space-time. If F depends on a//the points both in the future and
the past, what then? That could be the way physics works. I'll mention how
our theories go at the moment. It has turned out in many physical theories
that the mathematical equations are quite a bit simplified by imagining such
a thing--by imagining positrons as electrons going backwards in time, and
other things that connect objects forward and backward. The important
question would be, if this computer were laid out, is there in fact an
organized algorithm by which a solution could be laid out, that is, com-
puted? Suppose you know this function F, and it is a function of the
variables in the future as well. How would you lay out numbers so that they
automatically satisfy the above equation? It may not be possible. In the case
of the cellular automaton it is, because from a given row you get the next
row and then the next row, and there's an organized way of doing it. It's an
interesting question whether there are circumstances where you get func-
tions for which you can't think, at least right away, of an organized way of
laying it out. Maybe sort of shake it down from some approximation, or
something, but it's an interesting different type of computation.
Question: "Doesn't this reduce to the ordinary boundary value, as
opposed to initial-value type of calculation?"
Answer: "Yes, but remember this is the computer itself that I'm
describing."
It appears actually that classical physics is causal. You can, in terms of
the information in the past, if you include both momentum and position, or
Simulating Physics with Computers
471
the position at two different times in the past (either way, you need two
pieces of information at each point) calculate the future in principle. So
classical physics is
local, causal, and reversible,
and therefore apparently
quite adaptable (except for the discreteness and so on, which I already
mentioned) to computer simulation. We have no difficulty, in principle,
apparently, with that.
3. SIMULATING PROBABILITY
Turning to quantum mechanics, we know immediately that here we get
only the ability, apparently, to predict probabilities. Might I say im-
mediately, so that you know where I really intend to go, that we always have
had (secret, secret, close.the doors!) we always have had a great deal of
difficulty in understanding the world view that quantum mechanics repre-
sents. At least I do, because I'm an old enough man that I haven't got to the
point that this stuff is obvious to me. Okay, I still get nervous with it. And
therefore, some of the younger students ... you know how it always is,
every new idea, it takes a generation or two until it becomes obvious that
there's no real problem. It has not yet become obvious to me that there's no
real problem. I cannot define the real problem, therefore I suspect there's no
real problem, but I'm note sure there's no real problem. So that's why I like
to investigate things. Can I learn anything from asking this question about
computers--about this may or may not be mystery as to what the world
view of quantum mechanics is? So I know that quantum mechanics seem to
involve probability--and I therefore want to talk about simulating proba-
bility.
Well, one way that we could have a computer that simulates a prob-
abilistic theory, something that has a probability in it, would be to calculate
the probability and then interpret this number to represent nature. For
example, let's suppose that a particle has a probability
P(x, t)
to be at x at a
time t. A typical example of such a probability might satisfy a differential
equation, as, for example, if the particle is diffusing:
al,(x, t) _
v P(x,t)
Ot
Now we could discretize t and x and perhaps even the probability itself and
solve this differential equation like we solve any old field equation, and
make an algorithm for it, making it exact by discretization. First there'd be
a problem about discretizing probability. If you are only going to take k
digits it would mean that when the probability is less that 2 -k of something
happening, you say it doesn't happen at all. In practice we do that. If the
剩余21页未读,继续阅读
资源评论
jubincn
- 粉丝: 459
- 资源: 41
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功