On 15.08.2011 23:42 Jason Resch said the following:
On Mon, Aug 15, 2011 at 1:17 PM, Evgenii Rudnyi<use...@rudnyi.ru>

On 15.08.2011 07:56 Jason Resch said the following:


Can we accurately simulate physical laws or can't we?  Before you
answer, take a few minutes to watch this amazing video, which
simulates the distribution of mass throughout the universe on
the largest scales:
each point of light represents a galaxy, not a star)

The answer on your question depends on what you mean by accurately
and what by physical laws. I am working with finite elements (more
specifically with ANSYS Multiphysics) and I can tell for sure that
if you speak of simulation of the universe, then the current
simulation technology does not scale. Nowadays one could solve a
linear system reaching dimension of 1 billion but this will not
help you. I would say that either contemporary numerical methods
are deadly wrong, or simulated equations are not the right ones.
In this respect, you may want to look how simulation is done for
example in Second Life.

Well, today numerical simulation is a good business
(computer-aided engineering is about a billion per year) and it
continues to grow. Yet, if you look in detail, then there are some
areas when it could be employed nicely and some where it better to
forget about simulation.

I understand that you speak "in principle".

Yes, this is why in my first post, I said consider God's Turing
machine (free from our limitations).  Then it is obvious that with
the appropriate tape, a physical system can be approximated to any
desired level of accuracy so long as it is predictable.  Colin said
such models of physics or chemistry are impossible, so I hope he
elaborates on what makes these systems unpredictable.

I have to repeat that the current simulation technology just does not scale. With it even God will not help. The only way that I could imagine is that God's Turing machine is based on completely different simulation technology (this however means that our current knowledge of physical laws and/or numerics is wrong).

Yet, I am not sure if extrapolation too far away from the current
knowledge makes sense, as eventually we are coming to
"philosophical controversies".

We're already simulating peices of brain tissue on the order of fruit
fly brains (10,000 neurons).  Computers double in power/price every
year, so 6 years later we could simulate mouse brains, another 6 we
can simulate cat brains, and in another 6 we can simulate human
brains. (By 2030)

But all of this is an aside from point that I was making regarding
the power and versatility of Turing machines.  Those who think
Artificial Intelligence is not possible with computers must show what
about the brain is unpredictable or unmodelable.

Why that? I guess that you should prove first that consciousness is predictable and could be modeled.


You received this message because you are subscribed to the Google Groups 
"Everything List" group.
To post to this group, send email to everything-list@googlegroups.com.
To unsubscribe from this group, send email to 
For more options, visit this group at 

Reply via email to