2

There are tons and tons of physics simulators for classical/large scale physics. I'm interested in a simulator in which you feed an input as atoms and positions and it simulates the evolution of the system. That is, it must form bonds, ions, and so on as would happen naturally. But it should be fast enough that I could be able to actually feed some huge materials into it (a full scale transistor, perhaps?) and it would be able to simulate it accurately and under reasonable times.

In other words, what I do not want is a system based on higher models such as fluid dynamics, heat transfer and so on. I want it to let those phenomena emerge.

I'm aware maybe what I am asking is too much, but is there anything at least close to it?

Qmechanic
  • 220,844
MaiaVictor
  • 1,992

1 Answers1

3

Yes, I'm doing research with exactly such a concept right now. It is called molecular dynamics. It works exactly as you stated: you feed atom types and positions into it and get the time evolution of the system. I am also using what is known as a reactive potential, which allows chemical reactions, and the formation of bonds and ions (specifically, I am doing lithium ion simulations in carbon).

As for a transistor, those range in size. Experimental concepts have them as small as 1-7 atoms (at least according to the Wikipedia page; I know nothing about them); however, more conventional transistors could probably be simulated atomistically as well. My systems consist of 1-2 million atoms in a volume of 20x20x20 nanometers.

What I'm doing isn't quite quantum, but it's not quite classical either. It's in between. The potential (also known as a force field) that describes the interactions between the atoms was developed through quantum studies.

Nick
  • 3,039