[ghemical-devel] Constraint View Patch

[Tommi Hassinen]

On Fri, 31 Mar 2006, Donald Ephraim Curtis wrote:

> I got all this, and the engine IS getting re-created.  Look at this log
> output.  I drew 3 atoms, added a constraint of 1nm between two of them,
> optimized, changed the constraint to .2nm, then optimized and no graphic
> change occured.  (i'm just making sure this is the correct
> implementation).  If you use *min* bond length constraints, and he
> "natural" bond length is .01nm, then anything greater than that which is
> set as the minimum should be the one that gets optimized to?
>
>    |--1nm--|
> 1--2---------3
>
>    |-.2nm?-|
> 1--2---------3

Ok, I think I understand. And I think everything is OK, let me explain.

When we first set the minimum distance constraint to be 1 nm, and 
optimize, then the atom distance definitely should increase to the 1 nm 
(or larger).

When we later set the minimum distance constraint to be less than 1 nm, 
say 0.5 nm or 0.3 nm, and do geometry optimization, the following happens:

1) the current atom distance 1 nm is more than the set minimum distance 
(0.5 nm or 0.3 nm), so there is no energy "penalty" from the constraint ; 
the current structure satisfies the requirements set by constraint.

2) the structure where the atoms are closer together than 1 nm probably is 
slightly more favourable than the current structure (distance being 1 nm) 
but the energy difference is very small. Now the geometry optimization 
algorithm may or may not be able to find this slightly more stable 
structure where the atoms are closer together, but success is not 
really guaranteed.

Hopefully this clarifies the matter.

 	Tommi

PS: what should we do with this Ångström/nanometer issue?