Hi Geoff,
Thanks for moving this -- I should check autocompleted mailing list
addresses more closely ;-)
On Tue, Jan 3, 2012 at 11:51 AM, Geoffrey Hutchison
<ge...@geoffhutchison.net> wrote:
> On Jan 3, 2012, at 11:44 AM, David Lonie wrote:
>> which copies the data in _gradientPtr to _grad1. Trouble is,
>> _gradientPtr doesn't seem to be used in this function. Can this be
>> removed, or am I missing something?
>
> No, this object member variable is used by other code, e.g. GetCoordinates()
> as the forces on the atoms.
GetCoordinates etc read _gradientPtr, but where is it written?
>From my reading of the code, ConjugateGradientsInitialize calculates a
gradient and stores it into _gradientPtr, but then
ConjugateGradientsTakeNSteps calculates and stores a new gradient into
_grad1 for each step taken. After the linesearch, this call to memcpy
overwrites the gradient at the current step (_grad1) with the initial
gradient (_gradientPtr). The comment ("Save the direction") makes this
look like a bug to me, since it appears to actually reset the
gradient. Perhaps the memcpy args should be reversed?
I'm trying to track down a bug somewhere in the forcefields that is
causing this to happen:
C O N J U G A T E G R A D I E N T S
STEPS = 250
STEP n E(n) E(n-1)
--------------------------------
1 757722.040 5890899.081
2 179656.470 179656.470
CONJUGATE GRADIENTS HAS CONVERGED
The geometry is not reasonable, though it claims to be converged. Also
for step n=2, E(n-1) != E(1). This is not a result of the memcpy (I've
tried removing it and the problem persists), but that call does still
look buggy to me.
Dave
------------------------------------------------------------------------------
Write once. Port to many.
Get the SDK and tools to simplify cross-platform app development. Create
new or port existing apps to sell to consumers worldwide. Explore the
Intel AppUpSM program developer opportunity. appdeveloper.intel.com/join
http://p.sf.net/sfu/intel-appdev
_______________________________________________
OpenBabel-Devel mailing list
OpenBabel-Devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/openbabel-devel
