virtual chemistry

[butterysteez]

hi my name's Collin I'm a Computer Science student at UAA AlaskA. I'm in the process of making a 3d engine using opengl which uses rigid body dynamics and the Finite element method to simulate extremely interactive enviornments. (wood burns, steel melts, glass shatters, gass's compress)

 

I need a method for modeling the physical properties of atoms and molecular compounds.

 

given an atom or a molecule and it's temperature I need to be able to determine:

 

-the strength of the material (compression and tension)

-surface-tension / flexibility

-mass

-density

-elastisity

-cleavage

-appearance (color, reflecitions etc...)

 

also Givin a set of atoms and molecules I need to be able to determine:

 

-which new molecules are created

-what are the properties of the new molecule

-an idea of what the chemical reaction looks like

 

-----------------------------------------------------------------------------------

 

does something like this already exist in open-source software?

 

tables of physical properties are not sufficient. I need formulas for specific elements or for subatomic particles to determine the physical properties.

 

is there a methodical way for determining the physical properties of an atom based on it's subatomic makeup and temperature?

 

is there a methodical way for determining the physical properties of compounds based on it's atomic makeup?

 

-----------------------------------------------------------------------------------

[fafalone]

You can't really apply rigid body dynamics to individual atoms, since they're not really rigid bodies, as 99% of the space an atom occupies is electrons, which are standing waves.

 

But as far as simulating molecular interactions, ChemCAD is the way to go. Hopefully your university already has it, or it's going to cost $20,000... and it's not open source.

[blike]

eek!! I'm not sure, I'm sure you've dove into google looking for this stuff, so I wouldn't be of much help. How much background in chemistry do you have? Some of that sounds like physical chemistry. Some of it is available on the periodic table (mass). You'd have to have pretty extensive knowledge of chem to be able to predict complex compounds. I'll ask my chem prof. and see what he says.

[Radical Edward]

Originally posted by butterysteez

is there a methodical way for determining the physical properties of an atom based on it's subatomic makeup and temperature?

 

is there a methodical way for determining the physical properties of compounds based on it's atomic makeup?

 

Quantum Mechanics and Quantum Mechanics.

[butterysteez]

I don't care about the rigid body dynamics of the atoms themselfs. I need to know the physical properties of objects composed of elements. and they don't need to be exact.

 

my vision is the ability to interact with a 3d enviornment with a molecular background ie: wood burns, iron melts, ice cracks....

 

ie: iron melts at x degrees,

iron's strenth/flexibility/elastisity etc... are relitive to heat. what is the relation.

 

this is what I need for a large number of elements.

I don't have any chemistry background. I am taking a chem class this summer to help me out with it.

 

there are only 5 variables though why is it so complicated?

 

this is what I don't get:

----------------------------------------

atoms only have 5 variables: protons, nutrons, electrons, architecture and heat/energy. (although acrchitecture is much more complicated than a single variable)

 

Are there formulas relating these variables to the physical properties of large tangible objects which are composed of them?

 

I need either formulas or open source sofware because I'm trying to integrate them into a 3d physics engine. it doesen't need to be exact or perfect, engineers are not going to depend on it.

 

I could even get away with just predefining specific elements if somone would be kind enough to point me to where these properties are located or how I can derive them myself.

 

 

thanks so much for your help my software will be open-source upon completion and I'll post it here when I'm done.

[Radical Edward]

Originally posted by butterysteez

atoms only have 5 variables: protons, nutrons, electrons, architecture and heat/energy. (although acrchitecture is much more complicated than a single variable)

 

Are there formulas relating these variables to the physical properties of large tangible objects which are composed of them?

 

 

your view on atomic structure is rather simplistic. I will be kind and ignore isotopes for now, as they are largely irrelevant. when looking at individual atoms, you are not too far off, although each one of those variables has sub variables - the arrangement of the electrons in th"shells" is dictated by quantum mechanical rules, which get consecutively more complex as you get to larger and larger atoms. Something like hydrogen is really easy to describe, but something like Iron isn't. as soon as you start allowing intercactions between different sorts of molecules, you then start having to consider all the energy levels and interactions, and competition between different, atoms trying to interact with one another. once you start to get to large molecules it gets even more complicated, as you have to take molecular size, bond strength, and other niceties such as hydrophilia/phobia. This isn't even looking at different material properties such as plasticity, metallic, crystalline and so on.

 

In short, I think you are approaching the simulation from entirely the wrong direction. The only way to make a program of the sort you describe, is to actually have a table of material properties, and then assign each ting to a particular material type. so glass would be an amorphous brittle solid at room temp, diamond a crystal, copper a metal, and so on. Generally speaking you can give a particular material a class which will give it a bunch of properties that are the same as all others in that class, and all you need to tweak is things like the colour and so on, and the numerical material properties such as melting point and so on. Of course you would still be faced with some oddities, sand/dust/grains being one, as they have some solid properties, and some liquid properties, but these could be put in a misc category.

 

Atomic Physics and material Science are huge fields, and if it was so easy to simulate them, someone would have already done it. Even something like protein folding is hideously complicated.

[butterysteez]

thanks alot. I figured there was some pretty complicated stuff burried under there. I appreciate your help. I think I am going to just end up with a table of elements and their changing properties relitive to temperature.

 

even if I did wanna study chemistry for a while to get to understand it I think that the resulting mathmatical ops would take too long on the computer to warrant real time interaction.

[Radical Edward]

almost certainly. I looked into it with regards to writing a protein folding simulator, and that would have been extremely difficult. If I could get it to work though, I would become a millionaire very very quickly.

[butterysteez]

ya the more I research the more I'm finding out that it would be easyiest to just build up a table for elements and compounds. aparantly there are very few common naturally occuring minerals and elements. all I really need is for things to burn, melt, change colors, change their friction coeficcients, change their density, and perhapps react with a few other substances.

 

so do you think I could just take two elements (whose properties I have predefined) and if their extra/missing electrons match up (or come close) I could bond them? then the new mass would be equal to the sum and the density/strength/state would be relitive to the elegance of the molecular bond and origonal physical properties?

 

in other words: the relitive percentage of each elements' electrons which it is sharing will determine the percentage of volume loss and density increase from the molecular bond.

 

I know it would only be 1 line of code to determine if there is an electron bond then the density and volume determination would only be 3 or 4 interger multiplications which is nothing for a computer.

 

hell it wouldn't be exact enough for a chemist but it might be exact enough for determining erosion or solubility etc...

 

this is pretty helpful:

http://www.ce.berkeley.edu/~paulmont/CE60/atomic%20bonds/index.htm