Jump to content
Search In
  • More options...
Find results that contain...
Find results in...
GoatLord

How do you define solidity in computer simulations?

Recommended Posts

About 20 years ago, as a child, I had the following thought after seeing "Toy Story": What would happen if you were to slice one of the character models in half? It quickly occurred to me that the models are shells; infinitely flat polygons that have been stitched together into a mesh. The model, then, would have nothing inside of it. I pictured an algorithm attempting to fill in the gaps by instantly drawing another face where the slice was made, but since any given object can be assumed to be a complex arrangement of various material elements, the algorithm would have to differentiate between say, Woody's plastic skin and his denim pants. And since the denim pants are stitched together, the algorithm would have to understand the implied geometry of the pant's interior, since it would not be a smooth, continuous surface like his plastic skin.

Essentially, I don't like the idea of even a clever algorithm attempting to simulate what the interior of any given object looks like, because it's just a way of disguising the fact that the object is a mesh. One could try to solve this by creating elaborate layers of additional meshes to be revealed when sliced, but this is still phony as well. Filtered voxels could be used to represent individual atoms, but animating a data set of that size--hell, just rendering an unanimated data set of that size--would require far too much memory.

There is also the additional problem of the illusion of solidity. Quantum mechanics tells us that solidity is a manifestation of various subatomic forces, which our extremely limited sensory inputs interpret as continuous, smooth reality. This implies that neither triangles nor voxels are sufficient for representing what we perceive to be as solidity. So I ask: Should we just stick to these phony approximations, or are there rendering methods that could give us a better representation of the interior of any given object?

Share this post


Link to post
GoatLord said:

So I ask: Should we just stick to these phony approximations, or are there rendering methods that could give us a better representation of the interior of any given object?


Sure, wait until molecular-level, full-physics simulation will be feasible. Which brings us to the next question....can you completely simulate a molecule by using less than one physical molecule to represent its status, and perform the necessary computations?

If we're talking just about bits, then you can theoretically store even more than one bit in one atom, but even that is extremely wasteful if you think of all the molecules and atoms that you need to simulate even the simplest physical system, since you certainly need way more than one bit for status storage.

Share this post


Link to post

I was thinking about this, and I suspect that a very small, limited (but accurate) physical simulation is possible, if we use quantum computers, since they can be in multiple states simultaneously. In binary, you're either 1 or 0, but with quantum computing, you can be 00, 01, 10 and 11 all at once. That would definitely save space on computation. Problem is, we don't fully understand the physical or quantum world, so the simulation will still not be accurate, even if we simulate every single atom. You would also run out of atoms in the universe to use as quibits (quantum bits) once the simulation becomes so large.

Share this post


Link to post

I thought that the ultimate result would be that the only perfect simulation of a physical system, is to build a similar physical system. No perfect atom-level simulation could ever possibly take less physical space and less energy than the actual system itself.

E.g. to perfectly simulate a power plant (down to the last brick and speck of dust, not just saying "its current power output is 1 MW") you'd need nothing short of the power plant itself ;-) And if you simulate everything but the power output, then in reality you're simulating a real power plant in testing mode.

Since most simulations focus only on some aspects, they are simply finetuned to the point that they are "good enough" for a specific purpose.

Share this post


Link to post

What about a Matrix-style simulation, where there are limits in place, but the simulation is otherwise believable?

Share this post


Link to post

The advancements in the applied sciences are continually blurring the definition of what matter is, so I don't see why you'd need to represent molecules and atoms as individual physical objects. It'd be more prudent to make more comprehensive models for the four fundamental forces.

That's still the distant future, but maybe less so.

Share this post


Link to post

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×