The world as a neural network

[erek]

" We discuss a possibility that the entire universe on its most fundamental level is a neural network. We identify two different types of dynamical degrees of freedom: "trainable" variables (e.g. bias vector or weight matrix) and "hidden" variables (e.g. state vector of neurons). We first consider stochastic evolution of the trainable variables to argue that near equilibrium their dynamics is well approximated by Madelung equations (with free energy representing the phase) and further away from the equilibrium by Hamilton-Jacobi equations (with free energy representing the Hamilton's principal function). This shows that the trainable variables can indeed exhibit classical and quantum behaviors with the state vector of neurons representing the hidden variables. We then study stochastic evolution of the hidden variables by considering D non-interacting subsystems with average state vectors, x¯1, ..., x¯D and an overall average state vector x¯0. In the limit when the weight matrix is a permutation matrix, the dynamics of x¯μ can be described in terms of relativistic strings in an emergent D+1 dimensional Minkowski space-time. If the subsystems are minimally interacting, with interactions described by a metric tensor, then the emergent space-time becomes curved. We argue that the entropy production in such a system is a local function of the metric tensor which should be determined by the symmetries of the Onsager tensor. It turns out that a very simple and highly symmetric Onsager tensor leads to the entropy production described by the Einstein-Hilbert term. This shows that the learning dynamics of a neural network can indeed exhibit approximate behaviors described by both quantum mechanics and general relativity. We also discuss a possibility that the two descriptions are holographic duals of each other. "

https://futurism.com/physicist-entire-universe-neural-network

 

[travm]

Mushrooms in the basement?

Or architect of the matrix

 

[erek]

Mushrooms in the basement?

Or architect of the matrix

which is more believable?

 

[termite]

I guess neural net is cooler than shadows cast on a wall.

 

[THRESHIN]

Mushrooms in the basement?

Or architect of the matrix

Couldn't it be both?

 

[erek]

I guess neural net is cooler than shadows cast on a wall.

Heh, perhaps

 

[cdabc123]

Seems alot like trying to shove a circle into a square hole. I'm not arguing everything couldn't just be a simulation but don't try to find a simple simulation (like basic neural networks are) and say it must be that because we can't disprove it.

Disprove life isn't a game of higher dimensional sims city

 

[Axman]

I always thought this would have been a better premise for the Matrix than the stupid battery plot.

 

[serpretetsky]

Seems alot like trying to shove a circle into a square hole. I'm not arguing everything couldn't just be a simulation but don't try to find a simple simulation (like basic neural networks are) and say it must be that because we can't disprove it.

Disprove life isn't a game of higher dimensional sims city

I'm not a physicist, so I don't know the details of this hypothesis. However in my experience physics rarely asks questions that are not testable in some way (at the very least theoretically).

Therefore I do not believe this hypothesis is claiming what are outside the bounds of our universe. Whether or not our life is a sims city simulation in some other universe is a philosophy question, not a physics question.

From a physics point of view, EVERYTHING is always a simulation. The entire field of physics is basically asking how to model the universe (effectively, how to simulate it). You find the model that works best for your requirements.

What do you mean by "don't try to find a simple solution"? You always try to find the simplest solution which models your system to the accuracy that you need.
Simplicity can mean multiple things. Easy to understand. Fast to compute.
Hopefully your model is simple enough to be useful, but accurate enough to explain phenomena that could not be modeled before. (if the usefulness of this neural network theory depends on being able to run the simulation on 10^100 supercomputers to answer some of our questions then I would argue it's probably not simple enough...)

 

[erek]

I'm not a physicist, so I don't know the details of this hypothesis. However in my experience physics rarely asks questions that are not testable in some way (at the very least theoretically).

Therefore I do not believe this hypothesis is claiming what are outside the bounds of our universe. Whether or not our life is a sims city simulation in some other universe is a philosophy question, not a physics question.

From a physics point of view, EVERYTHING is always a simulation. The entire field of physics is basically asking how to model the universe (effectively, how to simulate it). You find the model that works best for your requirements.

What do you mean by "don't try to find a simple solution"? You always try to find the simplest solution which models your system to the accuracy that you need.
Simplicity can mean multiple things. Easy to understand. Fast to compute.
Hopefully your model is simple enough to be useful, but accurate enough to explain phenomena that could not be modeled before. (if the usefulness of this neural network theory depends on being able to run the simulation on 10^100 supercomputers to answer some of our questions then I would argue it's probably not simple enough...)

think quantum computing can get us over the hump with accurate modeling of the universe?

 

[Dayaks]

God imagine the latency.

 

[Dayaks]

Isn’t that convenient.

 

[Krenum]

Mushrooms in the basement?

No, Mushrooms in the Sky

View attachment atomic-bomb-1011738_960_720.webp

 

[cdabc123]

I'm not a physicist, so I don't know the details of this hypothesis. However in my experience physics rarely asks questions that are not testable in some way (at the very least theoretically).

Therefore I do not believe this hypothesis is claiming what are outside the bounds of our universe. Whether or not our life is a sims city simulation in some other universe is a philosophy question, not a physics question.

From a physics point of view, EVERYTHING is always a simulation. The entire field of physics is basically asking how to model the universe (effectively, how to simulate it). You find the model that works best for your requirements.

What do you mean by "don't try to find a simple solution"? You always try to find the simplest solution which models your system to the accuracy that you need.
Simplicity can mean multiple things. Easy to understand. Fast to compute.
Hopefully your model is simple enough to be useful, but accurate enough to explain phenomena that could not be modeled before. (if the usefulness of this neural network theory depends on being able to run the simulation on 10^100 supercomputers to answer some of our questions then I would argue it's probably not simple enough...)

Thats the problem I have with this theory its specifically trying to ask a question that cannot be proven to somehow validate itself. There is literally nothing basing the conclusions the author comes up with. Its the equivalent of declaring the universe was conceived by a flying spaghetti monster and each line of spaghetti is a string in string theory.

And I dont mine simple conclusions but thats not what science is about. Its about attempting to find and model truth as we perceive it.

For another analogy this is like me trying to model the universe in a excel spreadsheet and saying it can't be disproven.

think quantum computing can get us over the hump with accurate modeling of the universe?

Wouldnt that be neat. Now tell me if we reached that point would we spin up another model of a universe possibly containing conciousness?

 

[travm]

No, Mushrooms in the Sky

View attachment 278106View attachment 278107

Omg, Skynet, that also makes sense

 

[serpretetsky]

Thats the problem I have with this theory its specifically trying to ask a question that cannot be proven to somehow validate itself. There is literally nothing basing the conclusions the author comes up with. Its the equivalent of declaring the universe was conceived by a flying spaghetti monster and each line of spaghetti is a string in string theory.

And I dont mine simple conclusions but thats not what science is about. Its about attempting to find and model truth as we perceive it.
For another analogy this is like me trying to model the universe in a excel spreadsheet and saying it can't be disproven.

Correct me if I'm wrong:

I think you are saying that this physicist is saying "Wouldn't it be cool if our universe was just a simulation running in some neural network?". This is not a physics question or hypothesis. Assuming this article is valid, they did some research, and this is a serious researcher, then I will default to authority that is it not what it is saying (since this is a pointless question).

What I get from the article is the physicist saying this "There are inherent patterns in neural networks that begin to appear that seem to resemble certain behaviors in physics, it's worth studying if this can directly model the universe". Again, I dont know the details of this hypothesis, and the article is incredibly vague, but it seems like this idea needs more study in order to credit or discredit it (or perhaps even to formulate it more clearly!) .

Perhaps you have experience in these fields. I certainly don't.

 

[serpretetsky]

think quantum computing can get us over the hump with accurate modeling of the universe?

No clue. I'm a lowly firmware engineer. Quantuam computing goes way over my head. All I know is it somehow breaks RSA and ECC cryptography? And that all quantuam computers thus far have basically not been able outperform classical computers. shrug.

 

[noko]

think quantum computing can get us over the hump with accurate modeling of the universe?

No because it may decide to be in another universe.

 

[erek]

No because it may decide to be in another universe.

that'd be coo

 

[cdabc123]

Correct me if I'm wrong:

I think you are saying that this physicist is saying "Wouldn't it be cool if our universe was just a simulation running in some neural network?". This is not a physics question or hypothesis. Assuming this article is valid, they did some research, and this is a serious researcher, then I will default to authority that is it not what it is saying (since this is a pointless question).

What I get from the article is the physicist saying this "There are inherent patterns in neural networks that begin to appear that seem to resemble certain behaviors in physics, it's worth studying if this can directly model the universe". Again, I dont know the details of this hypothesis, and the article is incredibly vague, but it seems like this idea needs more study in order to credit or discredit it (or perhaps even to formulate it more clearly!) .

Perhaps you have experience in these fields. I certainly don't.

What I got from the article is closer to the first statement.

It appeared they they are using a standard neural network (with relatively simple design) as a computational machine and saying the universe is a simulation running on this computational machine. No reason it has to be a neural network and not a traditional computational method.

" In my theory, everything you see around you is a neural network and so to prove it wrong all that is needed is to find a phenomenon which cannot be modeled with a neural network. But if you think about it it is a very difficult task manly because we know so little about how the neural networks behave and how the machine learning actually works "

I also dislike this statement as the training neural networks he is using to model "everything" are not a big unknown they are pretty simple learning algorithms. we also have a good idea of how this simple form of machine learning works.

It appears to me as a theory of "its a simulation" weaving complexity on top of that with things the author apparently does not understand and using the degree of confusion to prove truth.

 

[juanrga]

" We discuss a possibility that the entire universe on its most fundamental level is a neural network. We identify two different types of dynamical degrees of freedom: "trainable" variables (e.g. bias vector or weight matrix) and "hidden" variables (e.g. state vector of neurons). We first consider stochastic evolution of the trainable variables to argue that near equilibrium their dynamics is well approximated by Madelung equations (with free energy representing the phase) and further away from the equilibrium by Hamilton-Jacobi equations (with free energy representing the Hamilton's principal function). This shows that the trainable variables can indeed exhibit classical and quantum behaviors with the state vector of neurons representing the hidden variables. We then study stochastic evolution of the hidden variables by considering D non-interacting subsystems with average state vectors, x¯1, ..., x¯D and an overall average state vector x¯0. In the limit when the weight matrix is a permutation matrix, the dynamics of x¯μ can be described in terms of relativistic strings in an emergent D+1 dimensional Minkowski space-time. If the subsystems are minimally interacting, with interactions described by a metric tensor, then the emergent space-time becomes curved. We argue that the entropy production in such a system is a local function of the metric tensor which should be determined by the symmetries of the Onsager tensor. It turns out that a very simple and highly symmetric Onsager tensor leads to the entropy production described by the Einstein-Hilbert term. This shows that the learning dynamics of a neural network can indeed exhibit approximate behaviors described by both quantum mechanics and general relativity. We also discuss a possibility that the two descriptions are holographic duals of each other. "

https://futurism.com/physicist-entire-universe-neural-network

Lots of giberish in a single paragraph!

 

[cdabc123]

This is fascinating!! Taco was trying to simulate hunger yesterday(to keep weight in check and look sexey) and I feel a whole lot better! Maybe if we all try and change our routines, we cn break th system and prove that it isnt reality, but simulation after all!! These feelings and other senses we have are all coded and work as a switching station that gets transferred(like wifi) to our brain neural receptors.

Fascinating stuff!!

Your looking at food wrong. The 3d food you know and love is just a dot in the multi demintinal reality that is life. Why would you find a dot tasty? Are you pacman?