FQXi grant awardee abstracts (novel time ideas)

[Careful]

I think I must have been unclear. You are right we need new matehmatics! That doesn't contradict my point. My points is WHAT new mathematics? The specific new mathematics needed is apparently program-dependent. This is why one cna easily imagine landscapes of new mathematics.

So the new mathematics needs to prove itself. Unless you want to say publish the new mathematics and get interest of mathematicians, but that's a different thing, and not my ambition.

So for example, the new matematics, required by MY ideas, is my problem to work out. I need to work it out, and show that it is viable for producing a physical theory that makes concrete predictions as this is the ultimate confirmation of the need for this new mathematics in the first place. For the single researcher, hunches or intuition is enough, but that's insufficient to convince your opponents.

I don't underestimate this at all. I full realize that to speka for myself, there odds at succeeding with this are slim! Anyone succeeding with this would certainly earned the place in the history books. But that fact should not discourage an intrigued mind! I am not the least discouraged by that. The journey there itself is also half the fun, even if I don't make it.

Sure the mathematics should be directly inspired by physics, but my point is that the physics will as a backreaction be inspired by that very same mathematics (and a closure has to occur at some point). So, the real difficulty is how to present this. Should one immediatly give the closure in all its natural but complicated glory (which might still consume 100 pages) or should one follow the historical route (which gives you at least 250 pages) ??
Bottom line is, you still need Planck If you choose the first route you need Planck to reassure people that it is brilliant (because they won't know) and in the second case you need Planck to inspire people to read it (the historical route enables for a better comprehension since it goes into smaller steps).

These are interesting questions on it's own, but I have something different in mind. A model that only makes predictions if you need something unknowable as input is not a good model. This is why I prefer evolutionary model. Knowing everything isn't possible. It's more about navigation in seas of ignorance.

Euhh, ALL our theories require input of something we don't know.

A proof of concept would be convergence of navigation schemes or learning models. The old scheme of eternal laws and initial conditions and boundary conditions is exactly what I think we need to replace. Smolin also mentioned this several times, in this evolution of law and reality of time.
/Fredrik

You will never get rid of boundary conditions, that's mathematically impossible. This is however not in contradiction with evolution of law.

Careful

 

[Fra]

Euhh, ALL our theories require input of something we don't know.

The discussing is sliding here but what I meant is that a prediction that is of any use, is a computation of an expectation based on what we know, that further determines the action. You need to have information of the current state of knowledge, and the inference model. Of course both of these are evolving during the process.

If we lack information of some things, we still have estimates or guesses. Information doesn't need to be definitie, it's more often probabilistic statistical or sometimes random.

One of the core questions I persoanlly ask is, the problem of choosing an action as to maximize your own advantage given incomplete decision base. This is how all learning works, and also according to my private conjecture how all physical interactions work.

I'm personally seeking to construct a new framwork for information process and and information representation that will contain replace continuum models and be a replacement of continuum based probability. All actions are depending only on available information. So a theory that needs input that isn't at hand is not viable. It must have the trait of beeing able to work even with one leg missing as there will always be missing parts. But I have plenty of work to do I have no intention of publishing anything premature. It takes new views on consistecy for example. Transient inconsistencies will exist, perfect consistency are more related to equilibrium conditions or expectations in my view. So I agree that this touches upon the foundations of logic and inference. So probably not only do we need new old style mathematics, we probably need a development of the foundations of logic and inference and it's relation to science. At least this is in the direction I'm walking, but my problem isn't to convince others at this point, my problem is to work out and implement the new physical framework and show that it is more fit tha nthe old framework because it has solved problems previously unsolved. Then the problem of convincing others should be easy in comparasion because I would have made the work, instead of asking others to make work for me.

/Fredrik

 

[Fra]

You will never get rid of boundary conditions, that's mathematically impossible. This is however not in contradiction with evolution of law.

INFORMATION about the boudary conditions are still encoded locally. This is all that is needed to produce the expectation and local action of the information processing agent. If there is no expectation at all, then something is really wrong in the way those complexions are introduced, because it means the complexion are just a ghost and isn't needed. Then it should collapse and fall out from the computation. The configuration space is not fixed. So poitns in the configuration space of which we are completely uninformed should go away. They aren't physical. This is one reason why the continuum models will be revised IMO if the the action doesn't depend on the full continuum.

So I think we DO have the information. That doesn't mean it's right, but that's a different question. Sometimes the action of an informatio nprocessing agent are based on say "false information" and that should produce physical effects that are observable as I see it.

/Fredrik

 

[Careful]

The discussing is sliding here but what I meant is that a prediction that is of any use, is a computation of an expectation based on what we know, that further determines the action. You need to have information of the current state of knowledge, and the inference model. Of course both of these are evolving during the process.

The point is that this is only an approximation. Any model based upon knowledge is incomplete; likewise are those theories with hidden variables. Ockham's razor is not applied in a simple way here because the latter option is still more accurate than the previous one and objective measures of complexity/predictive power do not exist.

One of the core questions I persoanlly ask is, the problem of choosing an action as to maximize your own advantage given incomplete decision base. This is how all learning works, and also according to my private conjecture how all physical interactions work.

I don't agree with you here since you still have to tell to your model what kind of incompleteness you are dealing with. Nobody knows how learning works: for example, I think communication between two people, why it is possible in the first place, is such a question which I believe has no answer in maximizing any profit function

 

[Careful]

INFORMATION about the boudary conditions are still encoded locally.

You mean quasi-locally...

This is all that is needed to produce the expectation and local action of the information processing agent.

Euhh, who says a physical entity can access this information ?

If there is no expectation at all, then something is really wrong in the way those complexions are introduced, because it means the complexion are just a ghost and isn't needed.

They might not be needed for observation, but they are mostly damn needed to calculate the probability amplitudes. See for example the Faddeev-Popov ghosts or zero norm states in Gupta Bleuler quantization ...

Then it should collapse and fall out from the computation. The configuration space is not fixed. So poitns in the configuration space of which we are completely uninformed should go away. They aren't physical.

The point is that in principle you will never be able to define the ''ultimate configuration space''. And why would those things which are beyond our knowledge go away? How would you create the Mona Lisa if the hidden seeds for it were not present in da Vinci's mind in the first place at some moment in his life?

This is one reason why the continuum models will be revised IMO if the the action doesn't depend on the full continuum.

I don't see what this has to do with what you told before. The continuum in my view is sacrosant and there is no compelling argument for discreteness I know of. Actually, this reminds me to wrap up my text for the FQXi contest.

 

[Fra]

I think now are different views are starting to shine through. I'll comment briefly, but the main point is that ultimately we do not really have to argue (unless we want to, or find it interesting to exchange ideas). Once more work is done, there would be no need for me to argue for my method or your for your method, since the result should speak for itself.

The point is that this is only an approximation. Any model based upon knowledge is incomplete; likewise are those theories with hidden variables. Ockham's razor is not applied in a simple way here because the latter option is still more accurate than the previous one and objective measures of complexity/predictive power do not exist.

Yes I agree there are no objective measures, but life need to continue.

What you call approximation is what I call all you have. If you call a theory incomplete because it's inferred from incomplete information, ok, but that is all you ever will have. Anything else is an illusion IMO. I have accepted this, and the questions I ask is to try to understand how such effective or observer dependent theories interact and evolve.

/Fredrik

 

[Fra]

I don't agree with you here since you still have to tell to your model what kind of incompleteness you are dealing with. Nobody knows how learning works: for example, I think communication between two people, why it is possible in the first place, is such a question which I believe has no answer in maximizing any profit function

These questions are things I'm working on. Let me get back to you in a far future and see if I have what it takes to be more explicit :) Maybe I won't, but I obviously have plenty of ideas on this. To characterize learning and learning in a way that connects to action and physics and information bounds is exactly what I'm workong on.

I do not have an answer at this point.

/Fredrik

 

[Careful]

These questions are things I'm working on. Let me get back to you in a far future and see if I have what it takes to be more explicit :) Maybe I won't, but I obviously have plenty of ideas on this. To characterize learning and learning in a way that connects to action and physics and information bounds is exactly what I'm workong on.

I do not have an answer at this point.

/Fredrik

There are nevertheless quite compelling arguments why you will never succeed. It appears to me you are a hard core positivist (as Hawking is one) while I am a mixture between a realist and positivist. I have no problem with hidden variables at all ...

Careful

 

[Fra]

Euhh, who says a physical entity can access this information ?

I picture it, the other way around: Only ACCESSIBLE information makes it into the map. But these maps are observer dependent. (Observer = physical information processing agent).

Thus, one observer can map things invisible to another observer; further the comparasion between maps is again observer dependent as it requires a third observer. There is no conflict here. But there are complications here also needing tobe sorted out. I consider that my problem. I have no ambition to try to be convinving at this point

They might not be needed for observation, but they are mostly damn needed to calculate the probability amplitudes.

This is also part of what I look to reconstruct. The computation of a probability amplitide. In current physics the observer dependnece if this computational framework is not analyzed to my satisfaction. All we have is the renormalisation group but it does not answer all of my questions.

A full theory scaling in what I picture is is bound to be more complicated than scaling coupling constants, which at it's simplest is just zooming in and out correcting for screening and antiscreening. I think there are more things going on when you scale not only the observational part, but also the computational system.

These are also things needing to be sorted out. I don't have the answers though.

/Fredrik

 

[Fra]

There are nevertheless quite compelling arguments why you will never succeed.

Well, if you look at history of science, and the massiveness of the questions I outline, you are certainly correct beyond doubt. I do realize that my mission is foolish, but nevertheless...

...as much as I fully see how hard this is, I am equally certain that it's the only rational path to walk, so the choice is easy.

/Fredrik

 

[Careful]

I picture it, the other way around: Only ACCESSIBLE information makes it into the map. But these maps are observer dependent. (Observer = physical information processing agent).

Thus, one observer can map things invisible to another observer; further the comparasion between maps is again observer dependent as it requires a third observer. There is no conflict here. But there are complications here also needing tobe sorted out. I consider that my problem. I have no ambition to try to be convinving at this point

Sure, I agree that access to reality is observer dependent (and actually this is quite easy to describe mathematically) but it still does not preclude the fact that there may be information which nobody can access (and therefore exists in an objective sense). The reason why you may not try to be convincing at this point is because you lack a rigorous framework; other people might already have this and are busy calculating some nontrivial consequences of it.

This is also part of what I look to reconstruct. The computation of a probability amplitide. In current physics the observer dependnece if this computational framework is not analyzed to my satisfaction. All we have is the renormalisation group but it does not answer all of my questions.

I agree.

 

[Fra]

it still does not preclude the fact that there may be information which nobody can access (and therefore exists in an objective sense).

But information that no one can access (meaning no physical system can interact with) is redundant, from the inside perspective in the sense that it does not affect anything.

These types of objective existence, while "possible", doesn't make much sense ot me. They seem irrational to me.

It's like trying to describe possible other universes in detail, and then add that no one can ever verify it. Although that may be true, it seems irrational to even raise these questions. I have more important questions to ask

The other thing that can and does happen, is that information inaccessible to us know, may become accessible in the future, in a way that was inherently inpredictable by anyone on the inside. This is why an evolving law, in darwinian style is needed also for the physical framework as I see it. The notion of timless law is then maybe a rational expectation in the sense that it rules the present action, but it's nevertheless not really timeless.

/Fredrik

 

[Careful]

These types of objective existence, while "possible", doesn't make much sense ot me. They seem irrational to me.

It's like trying to describe possible other universes in detail, and then add that no one can ever verify it. Although that may be true, it seems irrational to even raise these questions. I have more important questions to ask

You do not seem to grasp the meaning of objective hidden variables. The only thing they imply is that one cannot measure them directly but you can infer ''existence'' of them through direct ''measurements'' of relations between other entities. Science has always operated like that, even standard quantum mechanics contains hidden variables. I guess your mother should have slapped you more with Kierkegaard around the ears when you were young.

 

[Fra]

You do not seem to grasp the meaning of objective hidden variables. The only thing they imply is that one cannot measure them directly but you can infer ''existence'' of them through direct ''measurements'' of relations between other entities. Science has always operated like that, even standard quantum mechanics contains hidden variables. I guess your mother should have slapped you more with Kierkegaard around the ears when you were young.

But then the obvious question is, what variable do you actually measure "directly"?

Pretty much everything we know is a result of inference or abduction from data. At minimum it has to pass through again idirectly your senses.

Hidden variable as I know it means there is a variable (an objective degree of freedom; whatever that means in this context) whos values if known could restore determinism in QM. The objection I have against this notion, is that I fail to see how the complexion behind and observable whose value is totally unknown and ranom, could be distinguished and inferred and put on the map in the first place.

The closest thing I get to "hidden variables" is subjective or observer dependent variables. If a degree of freedom can be hidden from one observer, but seen by another. The failure of hidden variabl theories is the assumption that information from one observer, explains a problem pose relative to another observer, because if you try to parallelltransport or communicate that between them, it will be lost in transmission.

So the fallacy is the assumption that a complexion defined relative to one observer, enters the computations of expectations of another observer. But that's I guess parts of realist thinkging.

I'm even strongly opposing to what's usually referred to as the "least objectionable" form of realism: structural realism, that we indeed have also in QM! So you're right about that. But I disagree with that too. No wonder I want to reform alot.

/Fredrik

 

[Careful]

But then the obvious question is, what variable do you actually measure "directly"?

Pretty much everything we know is a result of inference or abduction from data. At minimum it has to pass through again idirectly your senses.

Hidden variable as I know it means there is a variable (an objective degree of freedom; whatever that means in this context) whos values if known could restore determinism in QM.

Nah, that's just what silly people make of it. A hidden variable is a (dynamical ?) entity which you cannot access through experiment, that is the superposition principle does not apply to them. So, yes, you need to be very specific about what you can directly measure and what not; that is, to which part of reality does the superposition principle apply and to what part does it not?

I'm even strongly opposing to what's usually referred to as the "least objectionable" form of realism: structural realism, that we indeed have also in QM! So you're right about that. But I disagree with that too. No wonder I want to reform alot.
/Fredrik

But that kind of realism is unavoidable, these are the limitations of a symbolic language and we shall never overcome it.

Careful

 

[Fra]

But that kind of realism is unavoidable, these are the limitations of a symbolic language and we shall never overcome it.

No it's not unavoidable, then we aren't talking about the same thing.

It is a choice you make, to think it's unavoidable, or try to see how you could do without that assumption.

An implication of my view is that even the language needs to evolve. You can't have a fixed language or fixed logical system, and expect that to last forever.

This is why ultimately the logical system itself evolves, and this generally unpredictable. This is unavoidable, but we can make the choice to think that there exists a static formal system which can encode what I am looking for. I am certain there is not. This is also why we need not only new mathematics, but probably even close to a new logical inference system.

/Fredrik

 

[Careful]

This is why ultimately the logical system itself evolves, and this generally unpredictable. This is unavoidable, but we can make the choice to think that there exists a static formal system which can encode what I am looking for. I am certain there is not. This is also why we need not only new mathematics, but probably even close to a new logical inference system.

/Fredrik

That is what I meant of course; there exists no language in which to describe the change of ''language'' in physical/logical laws. Change of the laws are moments of irrational creation which can not be ''predicted'' (in the stochastic or deterministic sense) at all.

 

[Fra]

That is what I meant of course; there exists no language in which to describe the change of ''language'' in physical/logical laws. Change of the laws are moments of irrational creation which can not be ''predicted'' (in the stochastic or deterministic sense) at all.

It may seem like we agree but I sense that we have different views:

You say "there exists no language in which..." while I say "there exists no STATIC language ...".

This also relate to things that need work, but my claim is that because there is no static objective language doesn't mean there is no language at all.

I think each observer has some limit beyond which things are simply truly undecidable at any moment. But a complex observer can possibly predict and produce expectations of how another observer evolves it's capabilities. This I believe is predictable.

I think the mere insight that the action of any physical systems is constrained by some horizon of undecidability is a key to understanding the origin and diversity of interactions.

I think of unification of forces this way: the smaller and less complex the observer (physical processing agent) is, the simpler is it's action. At some point a generic prediction is that the diversity of interactions aren't distinguishable, simply because the information processing agents participating in the interactions are unable to "formulate and represent" this complex things. An interesting "scaling" of this appears as one scales up the complexit of the observers (~ energy scale of observer, not the probing energy). New interactions emerge in a way analogous to development of the language of the internal players.

Here I am crazy enough to be convinced that predictions can be made. However, the situation is completely different for cosmological scale theories, and here the undecidability becomes more obvious. Somehow it would be a generalization of some kind of theory scaling where, when you put aggregtes of simple blocks following simple laws (the unified interaction) the larger systems spontaneously start to break this simplicity and diversity is formed.

A larger observer, that can encode large amunts of such inforamtion, of systems of smaller systems, can thus predict how a original interaction breaks up into varieties.

What I've done is trie to identify abstractions of the simplest starting points, and they try to map what happens to rational action when the systems complexity increases. But one also has to somehow explain the complexity aggregation as well, connecting to mass generation. I just feel that to start with the regular classical lagrangians and the continuum field theory stuff is missing the whole point already at square one. The continuum itself is nontrivial and it seems very superficial to bypass a more careful introdcuton of counting systems.

/Fredrik

 

[Careful]

You say "there exists no language in which..." while I say "there exists no STATIC language ...".

You make a logical error here: the ''ultimate'' language is always static, a language does not allow for real creation. So therefore a theory of everything will not exist.

This also relate to things that need work, but my claim is that because there is no static objective language doesn't mean there is no language at all.

There is a language, but it is not a well defined one as true languages are.

I think each observer has some limit beyond which things are simply truly undecidable at any moment. But a complex observer can possibly predict and produce expectations of how another observer evolves it's capabilities. This I believe is predictable.

No, it is not. This may work for elementary particles on reasonable timescale but it is never going to work for humans.

Careful

 

[Fra]

You make a logical error here: the ''ultimate'' language is always static, a language does not allow for real creation. So therefore a theory of everything will not exist.


There is a language, but it is not a well defined one as true languages are.


No, it is not. This may work for elementary particles on reasonable timescale but it is never going to work for humans.

Careful

I am not seeking a static theory of everything, I am seeking to understand evolution of physical law, and how it's coded and related to the microstructure of matter, and how thus further encodes the physical interactions we know.

So I agree that such a static TOE won't exists; that doesn't mean I can't acquire and excellent but still of course incomplete knowledge of how this evolution of theory works. This is my quest.

This does work, but not for too complex system, and the reason is clear. But there is not cut limit, there is a sliding scale from strong predictivity to complete undecidability. This scale is an important scale to consider in the abstraction.

It's known from single cell and bacterial systems that understanding of the evolutionary mechanism can really provide predictions that deterministic models can not. Due to computational complexity, as well as sensitivity of initial conditions (deterministic chaos) it's de facto impossible to write down chemical dynamics equations from the chemistry in a cell and predict how it's gene expressison and regulation evolve as dynamical entitires. However, just defining a state space based on known genes, and chemicaal pathways, without detailed knowneldge of the actual regulatory mechanisms, one can try to optimise certain life functions such as growht rateetc and find a prediction. I recall reading an interesting paper long time ago when I studied yeast cells where this was confirmed in an experiment with cultures of I think E coli bacteria. One observer transient disagreements, but once the culture equilibrated in the new environment, the prediction based on pathway ang gene expression expectations was really close. The ideas is that the overall goal is used as a shortcut, then the assumption is that "nature will find a way", and after equilibratrion it did.

Similarly the way to understand and predict humans to the limited extent possible, it's for the same reason impossible to set up the differential chemical euqations of a human, instead we know the action space of a human, we assume rationality and then we get close. We don't get dead on, but fortunately beeind dead on is not necessary at all.

I am seeking to apply similar reasoning to physics. Once you have a "DNA" of physical or it's equivalent, predictions can be made in the same manner. But even this DNA is evolving, just like in biology, so each level has it's own predictive scheme. The only think that's not possible is to cover all scales. We happened to live at a certain scale, and it's around this we pose questions and learn about our environment. That's good enough for me.

/Fredrik

 

[PAllen]

I don't know. Didn't Hilbert discover relativity around the same time ? What was the interaction between those two men ?

Careful

Einstein presented his developing GR ideas to Hilbert in several long visits, and they had extensive correspondence. Hibert got involved only after the main goals and approaches had been explored by Einstein.

Without Einstein, I would expect that one of the SR+gravity variants would have won out for quite a long time.

 

[Careful]

Einstein presented his developing GR ideas to Hilbert in several long visits, and they had extensive correspondence. Hibert got involved only after the main goals and approaches had been explored by Einstein.

Without Einstein, I would expect that one of the SR+gravity variants would have won out for quite a long time.

I suspected something like that, but I was not sure anymore though.

Careful

 

[Careful]

I am not seeking a static theory of everything, I am seeking to understand evolution of physical law, and how it's coded and related to the microstructure of matter, and how thus further encodes the physical interactions we know.

So I agree that such a static TOE won't exists; that doesn't mean I can't acquire and excellent but still of course incomplete knowledge of how this evolution of theory works. This is my quest.

This does work, but not for too complex system, and the reason is clear. But there is not cut limit, there is a sliding scale from strong predictivity to complete undecidability. This scale is an important scale to consider in the abstraction.

It's known from single cell and bacterial systems that understanding of the evolutionary mechanism can really provide predictions that deterministic models can not. Due to computational complexity, as well as sensitivity of initial conditions (deterministic chaos) it's de facto impossible to write down chemical dynamics equations from the chemistry in a cell and predict how it's gene expressison and regulation evolve as dynamical entitires. However, just defining a state space based on known genes, and chemicaal pathways, without detailed knowneldge of the actual regulatory mechanisms, one can try to optimise certain life functions such as growht rateetc and find a prediction. I recall reading an interesting paper long time ago when I studied yeast cells where this was confirmed in an experiment with cultures of I think E coli bacteria. One observer transient disagreements, but once the culture equilibrated in the new environment, the prediction based on pathway ang gene expression expectations was really close. The ideas is that the overall goal is used as a shortcut, then the assumption is that "nature will find a way", and after equilibratrion it did.

Similarly the way to understand and predict humans to the limited extent possible, it's for the same reason impossible to set up the differential chemical euqations of a human, instead we know the action space of a human, we assume rationality and then we get close. We don't get dead on, but fortunately beeind dead on is not necessary at all.

I am seeking to apply similar reasoning to physics. Once you have a "DNA" of physical or it's equivalent, predictions can be made in the same manner. But even this DNA is evolving, just like in biology, so each level has it's own predictive scheme. The only think that's not possible is to cover all scales. We happened to live at a certain scale, and it's around this we pose questions and learn about our environment. That's good enough for me.

/Fredrik

We more or less agree, I think my point of view is somewhat stronger than yours but this may be a matter of wording. Words are personal too and we can only do our best to make ourselves clear.

 

[Fra]

I think my point of view is somewhat stronger than yours but this may be a matter of wording.

/Fredrik

 

[Careful]

/Fredrik

 

[yoda jedi]

ST has proven its worth beyond any doubt.

But details aside, I do agree that FQXi is unlikely to fund a lone wolf.

I think our disagreement is that I think we don't need lone wolves, but you think we do.

Maybe Perelman is such a case. I don't understand the mathematics, but some have said he was after all preceded by Thurston and Hamilton.

Or if we do, then by definition they must be lone. Can we engineer society to create them? What really gave rise to Bach, Beethoven and Brahms? It is still disconcerting to me that the very same culture that gave us those things, also gave us Nazism. (I suppose you could argue it wasn't the "same" culture.)

ST ? superstring theory ?

 

[Max™]

aDS/CFT and Holographic developments are valuable, it is not certain that they would not exist without string work, but it is true that they are results of the effort which proved useful in other areas.