?

Log in

No account? Create an account

Previous Entry | Next Entry

David Finkelstein is one of the greatest thinkers in physics today. I say thinker because there are those who simply "do" physics, and then there are those who also think about what they're doing. He's one of the best at both. (On a historical note, a paper he wrote in grad school on how to interpret black holes is what convinced John Wheeler they existed, who in turn convinced the rest of the world.) I was fortunate enough to have him for a professor in 1998 at Georgia Tech. And he's influenced my life and thoughts in countless ways, and given me something to aspire to. I have no doubt that some of the ideas he's had and some of the work he's done will have great importance in 21st-century physics, but for the past 3 or 4 years there have been certain growing objections in my mind to the way he talks about physics. I haven't been able to fully articulate them until now that they've solidified enough.

Since firmament, burdges, and ikioi are probably the only readers of this who are at all familiar with his ideas, I'll start with an (extremely brief) few sentences on what they are. (Unfortunately, it would take a very long time to do them justice.) He is a post-copenhagenist who believes that Heisenberg and Bohr were on the right track with their original interpretation of quantum mechanics, but he takes it a big step further and makes it far more consistant philosophically. To do this, he replaces classical logic with a more sophisticated quantum logic (and correspondingly, set theory with a quantum set theory, etc.). He believes that truth is found in processes, not in states. Truth doesn't just have a single mode (being), it has initial and final modes. He tells us we should give up our belief in objects and instead believe in actions. (You do X, and such and such results... there is no meaning to talking about what happened in between the initial and final actions.) Hence, he says reality is not ontological (consisting of states of being) but praxic (consisting of actions performed by experimentors). This is a vast simplification of his views, and it probably makes him sound crazy... but nevertheless it gives some idea of where he's coming from. So the following is an account of my objections to this type of process-based metaphysics. I would not be surprised at all if, after a few more years of thought, I changed my mind again, but right now this is the way I see it so I think it's worth writing down.

I agree with him that there is a strong, and fundamentally important, analogy between quantum mechanics and relativity. But I disagree with his contention that quantum mechanics has to do with giving up belief in states of being. What it does is relativizes states of being, it doesn't make them any less real. The way he treats it goes beyond what is necessary. As he says, relativity involves giving up belief in space and time. But what he doesn't mention which is very important, is that spacetime still exists absolutely. And from a bird's eye view it is just as "objective" as space and time were before. We re-interpret space and time as mere components of a larger more universal object. The solution in quantum mechanics should be the same. We should NOT give up our belief in states, we merely have to re-interpret things like position and momentum as basis-dependant components of a larger, more universal structure: the wavefunction. One of the primary objections people raise to Everett interpretations is the so-called "basis problem". They say that since the only way to talk about two split universes is to select a basis from within which to make the split, this amounts to picking out a preferred basis which is just as bad as a preferred frame in relativity. Certainly they are correct that there is no such thing as a preferred basis in quantum mechanics. The basis is always relative to the observer. To one person, the multiverse splits up one way; to another it may be split up another way. This is merely a matter of perspective, the structure is there and absolute. Just as with relativity where there is an absolute spacetime structure, there is an absolute object in quantum mechanics, the so-called universal wavefunction. Unfortunately, the term "wavefunction" for some conjures images of its representation in, say, the position basis of a single-particle-eigenstate. The wavefunction as a mathematical object has no preferred basis. It will appear different to different observers in the same way a car appears different when viewed from different angles. But it still exists. There is no strong justification for retreating into a process-based language. You can talk about objects which exist, it is okay to use the word "is". The only catch is, the objects which are there are much more abstract and bizarre1 and higher-dimensional and beautiful than anything we have direct contact with. Perhaps when unification is all finished2 there will only be one absolute "object", I find this idea aesthetically appealing (but would by no means bet the farm on it). That would mean that all of our conceptual boundries between objects are man-made. But it would not mean that reality is non-ontological. It would simply mean reality is holistic. Of course, it's also possible there will be multiple objects. Either way, I don't think processes are the way to go. The whole idea of an action or a process involves doing something IN TIME. Viewed from outside of spacetime, a process is just another noun, an object which exists in the spacetime manifold somewhere. Making processes more fundamental than states is singling out time as special when it's not. There are differences between space and time directions, of course (a change of sign in the metric), but neither is fundamentally more important. It's kitsch to say this, but ultimately he may just be confusing the means by which we do experiments with the system we're experimenting on. There's no convincing reason to expect these should be one and the same. Only if it is absolutely necessary should we support the banning of all nouns from language, and I'm thinking more and more that it's just not necessary. Objects are far too useful and universal for me to ignore. At least for the foreseeable future.

(1) by "abstract and bizarre" here I mean something very specific: I mean things which, in any human representation which we'd need for understanding them, have enough dualities that it's impossible to visualize them--they are not geometric shapes you could look at directly with reflected light. The number of things you have to "mod out by" in order to strip them down to their naked reality is large.

(2) some may find the idea that unification will one day be "finished" dubious. But at each stage, we reduce the number of objects (meaning: the number of properties, qualities, or parameters reality has). Currently there are less than a hundred distinct interacting objects in the Standard Model, and AFAIK a similarly small set of objects in General Relativity and cosmology. The process of unification cannot continue indefinitely because there is a lower limit to the number of objects (one). So there are only two possibilities in my opinion: either we will get stuck with a small number of objects and not be able to simplify it any further; or we'll get all the way down to one object and know that we're done. Either way, I think my usage of "finished" is appropriate. After it's finished, of course, there will still be plenty of emergent phenomena to investigate, both at the particle level and higher up, which is what will keep physics from dying out.

Comments

( 13 comments — Leave a comment )
dankamongmen
Dec. 14th, 2004 05:19 am (UTC)
i take physics of stars and galaxies with him next semester, as a free elective, to round out my CS education!
spoonless
Dec. 14th, 2004 08:28 am (UTC)
cool, ask him how his bud the Dalai Lama is doing these days!
firmament
Dec. 14th, 2004 07:21 pm (UTC)
Rad post. It definitely hits some fundamental points, and it demands some reflection. I'll play the orthodox and come back with a longer reply after I think about it a bit.

But for the moment, a few pot-shot points:

It seems like the 3N dimensional wavefunction is a bad candidate for ultimate reality, since we're kicking around in a world that looks pretty damn low-D. One might use flatland arguments to convince me to add a dew more dimensions, but 3N...

Also, process metaphysics is totally compatible with special relativity.

Finally, why is the One so much better than the Many? Is it just an aesthetic thing? Do we think aesthetics should track reality?
spoonless
Dec. 14th, 2004 10:18 pm (UTC)
Glad you liked it, I figured you'd be interested.

It seems like the 3N dimensional wavefunction is a bad candidate for ultimate reality, since we're kicking around in a world that looks pretty damn low-D. One might use flatland arguments to convince me to add a dew more dimensions, but 3N...

First, not 3N... but I'll get back to that later. For now:
Spacial dimensions are only one type of dimension. A dimension in general is simply a degree of freedom, and even classically reality has more than 3 degrees of freedom (there is a lot more complexity than just a single thing moving in three directions). Just because time happens to be interconnected with and mix with the three spacial dimensions doesn't mean all degrees of freedom would. I would not call these other things spacial dimensions, they are just degrees of freedom... not the dimensionality of spacetime, more like the dimensionality of what's living in spacetime.

Now back to the 3N. 3N usually refers to a Fock space... N non-interacting particles in 3 dimensions using non-relativistic quantum mechanics. The real story has to be much different. The 3 needs to be a 4, because you don't want to treat time separately. And N particles only makes sense in the non-interacting limit or at velocities much less than c. Having N particles if they are identical just means the vacuum was excited into the Nth level in that mode. Having N different particles means it was excited in N different ways. The N you get is not fixed, since particles pop up and disappear all the time. There may be problems with field theory as an ultimate theory, but at least it got this much right. Once those N particles start interacting, you get weird mixed states which so far we've only been able to talk about using perturbation theory which is messy stuff. Basically, in field theory the 3N is replaced with a 4*infinity. Presumably, once spacetime is quantized successfully, either by Finkelstein's nets, or by loop quantum gravity, or by string theory, that infinity will have a cutoff related to the Planck energy. But whatever it is, it's not really the same N as in the 3N.

Also, process metaphysics is totally compatible with special relativity.

Yes, of course. But he draws much of his motivation for giving up states from the idea that quantum should be done analogously to relativity. And I'm saying "yes, it should. except the type of relativizing done in relativity is not the same as giving up belief in something. relative != non-existent." That's my central criticism, really.

Finally, why is the One so much better than the Many? Is it just an aesthetic thing? Do we think aesthetics should track reality?

This was really just a side comment. I consider it an entirely open question. So yes, it's just an aesthetic thing. I have noticed nature is beautiful, and therefore I expect any new layers of it to be equally beautiful. But the way in which that beauty manifests is highly debatable and not easily predictable, which is why we should treat any sheerly aesthetic arguments as hunches at best. Good for giving you ideas, but not good for proving anything or making convincing arguments.
spoonless
Dec. 15th, 2004 03:24 am (UTC)
Oh and I should have mentioned, lest I make things seems a lot smaller than they are... that if we're talking about the dimensionality of the Hilbert space, which I think is what's really relevant here, then it's really infinity^(3N) and infinity^(4*infinity) for the two cases I discussed. Of course in a final theory, both of these infinities should be finite. The first one is sort of "# of locations" which would be on the order of L/hL where L is the width of the universe and hL is the Planck length. The infinity in the exponent (next to the 4) is the number of possible energies, so this as I mentioned would have something to do with the Planck energy... roughly, the ratio of that to the inverse-width of the universe which would determine the minimum energy something can have. So all this could be used to give an order of magnitude estimate for the number of dimensions in the Hilbert space. The actual number of states is another huge number raised to that power, which is larger still! So you're right in saying it's not small. But again, counting how many directions you can move around in gives you only an estimate of spacial dimensions (which could also be higher if something like string theory were right) but is not the same as the overall dimensionality of state-space... nor should it be.
gustavolacerda
Dec. 15th, 2004 01:33 am (UTC)
Do you read or write anything on philosophy of physics? quantum mechanics? This is probably what I would need before taking on QM again.
spoonless
Dec. 15th, 2004 03:03 am (UTC)

Do you read or write anything on philosophy of physics? quantum mechanics?

A more common question I get is: do I ever read or write about anything else? ;)

This whole entry was intended to be on the philosophy of physics, and in particular quantum mechanics. Did it come across as something else? Anyway, yes... I read, write, and think about it incessantly.
romanarce
Dec. 15th, 2004 03:35 am (UTC)
"Just as with relativity where there is an absolute spacetime structure, there is an absolute object in quantum mechanics, the so-called universal wavefunction."
Every part of spacetime can be tested (an experiment at some point in spacetime) at any time and multiple tests will give the same result. The wave function can be tested only once which is a huge difference. I could add that different tests with the same wavefunction could give different answers because all you have is probabilities but probability only makes sense mathematically after multiple tests and a wavefunction can only accept one. And if you think about repeating the tests with the same wavefunction to give probabilities a meaning then there's no way to know if a repeated experiment which looks the same has the same wavefunction or not. In principle a repeated experiment has the same wavefunction, but all you get is probabilities, which again need identical repetition, or the wavefunction is different and besides probabilities you have hidden variables (not Born's ones). Basically probability and being able to test it only once is contradictorious so you can't say it's something so absolute and there's nothing else, although there's probably nothing else that can be tested with experiments making this unphysical (which is fine as long as we keep the line between physics and philosophy).

"Making processes more fundamental than states is singling out time as special when it's not."
No, events are just points in 4D spacetime. Now we can only make experiments with before and after something so the idea of process comes from relating to spacetime points, but that's all. He should have written "relations between events" instead of processes.
spoonless
Dec. 15th, 2004 06:02 pm (UTC)
I'm on my way to catch a plane flight, so I'll briefly respond to some of these... after that, it might be a few days before I get a chance if there's more to say.

You raise some good points.

Every part of spacetime can be tested (an experiment at some point in spacetime) at any time and multiple tests will give the same result. The wave function can be tested only once which is a huge difference.

You're right that the wavefunction is not nearly as directly testable as spacetime. But even spacetime, you can't test every point. For instance, we cannot test points in the past. We can hope to test points in the future, as long as they are inside our light-cone. The rest of spacetime we assume exists because it's nice and simple and consistant and makes sense, just like the wavefunction.

which again need identical repetition, or the wavefunction is different and besides probabilities you have hidden variables (not Born's ones).

I hope you're not saying you believe in local hidden variables. I assume by Born you mean Bohm? I'm not sure I followed this whole sentence.

Basically probability and being able to test it only once is contradictorious so you can't say it's something so absolute and there's nothing else

I think there's an important thing you can test which is the wave-interference. This is an indication that something is there interfering. And most physicists now would agree, the wave-function never fully collapses. "Collapse" happens only in the limit of a large number of particles becoming correlated and decherent. You sort of get more than one trial, in a sense, with things like quantum computers. There's a lot of structure there going on in the intermediate stage, and you can either acknowledge its existence or ignore it. I think it makes more sense to acknowledge it.

events are just points in 4D spacetime. Now we can only make experiments with before and after something so the idea of process comes from relating to spacetime points, but that's all. He should have written "relations between events" instead of processes.

I admit, this is the weaker of my two main complaints with Finkelstein's view. He actually uses the term "actions" more than processes. In a way, I think actions are even worse because they seem to imply an actor, some sort of conscious being. I see your point about processes merely connecting events, and it's something I thought of as I was writing this. But the thing is, there are two types of connections between spacetime points--timelike intervals, and spacelike intervals. Spacelike intervals are more like states and timelike are more like processes. It definitely doesn't make sense to call a spacelike interval a process, because no process can happen faster than light. I guess I'm just saying that there shouldn't be a distinction between these two. But I guess "relation between events" would fix that.
gustavolacerda
Dec. 18th, 2004 09:18 pm (UTC)
What kind of work do modern theoretical physicists really do? Do they ever have to touch a data set?
If not, it seems like their role is one of knowledge engineering: reformulating, tweaking theories, unifying theories, interpreting theories in different ways. Which is very much like philosophy, isn't it? with the difference that theoretical physicists can make empirical predictions.

I'd like to know what is the day-to-day routine of a theoretical physics researcher.
spoonless
Dec. 22nd, 2004 02:41 am (UTC)

What kind of work do modern theoretical physicists really do? Do they ever have to touch a data set?

Sure, they have to make sure that anything they propose is consistant with any known experiments to date. This sometimes could take a lot of work, or it could take very little work, or no work at all depending on the situation. To publish a paper about a new theory which contradicts something which has already been measured by experimentalists would be a great embarrasment... if there's any chance this might happen, it's the responsibility of the theorist to consult the data and make sure it doesn't... usually, these get caught before being published, but sometimes it doesn't come out until later. Fitting the theory to constraints in the form of "data sets" is a part of some theoretical physics work, but there are many other parts as well (see below).

If not, it seems like their role is one of knowledge engineering: reformulating, tweaking theories, unifying theories, interpreting theories in different ways. Which is very much like philosophy, isn't it? with the difference that theoretical physicists can make empirical predictions.

Knowledge engineering is indeed very much like philosophy, in fact I think most would agree it is philosophy. But the ramifications of new theories is not always clear, and if they are also to be classified as "physics" (in addition to philosophy) then they need to at least have the hope of making new predictions which could in principle be tested. This is the subject of some degree of controversy within the field.

I'd like to know what is the day-to-day routine of a theoretical physics researcher.

Mostly, lots and lots of calculations. Professors try to pass as much of this as possible on to graduate students, but it's still a part of the job and they do them too. The things done in theory include trying to explain new experimental observations by modifying or extending old theories and proposing new theories; trying to figure out whether those new theories are contradiction-free; trying to find other predictions (and retrodictions) from the new theories; trying to propose new experiments to test the theories.
gustavolacerda
Dec. 22nd, 2004 08:17 am (UTC)
Mostly, lots and lots of calculations. Professors try to pass as much of this as possible on to graduate students, but it's still a part of the job and they do them too.

Hm... I wonder how much of this could be automated or tremendously helped by current technology in computer algebra? You know programs like Mathematica have integration, diffeqs and much much more?

They're even starting to do proofs.
spoonless
Dec. 26th, 2004 02:58 am (UTC)

Hm... I wonder how much of this could be automated or tremendously helped by current technology in computer algebra? You know programs like Mathematica have integration, diffeqs and much much more?

Most theorists are heavy mathematica users. If anything can be reduced to just an integral or a differential equation, then mathematica (or equivalent software) is the way to go. However, the math used for most of this stuff is far more abstract than what mathematica can handle... in short, the technology doesn't keep up with the math we use.

That said, I agree, that a lot of it can probably be automated with computers much more than is currently done. And that's one of the areas I'd definitely be willing to get into. However, there are limitations. Physics is somewhat unique in that it's easy to ask clear meaningful questions which are very difficult to answer. Solving some problems requires multiple stages of creativity. It requires dropping the right terms at the right times, and always keeping in mind which factors are the most relevant and which factors can be ignored. We use lots of tricks, and even have to come up with new ones sometimes, in order to get around intractible computations. Often, human minds can be more powerful than computers (are right now) because it can skip steps, be more creative, and use analogies to other related problems. What would be a really nice breakthrough is to automate that kind of thinking, but it's by no means an easy task!
( 13 comments — Leave a comment )

Profile

blueshirt
spoonless
domino plural

Latest Month

May 2017
S M T W T F S
 123456
78910111213
14151617181920
21222324252627
28293031   

Tags

Powered by LiveJournal.com
Designed by Lizzy Enger