- Contemporary messages sorted: [ by date ] [ by thread ] [ by subject ] [ by author ] [ by messages with attachments ]

From: Adrian Roitberg <roitberg.qtp.ufl.edu>

Date: Thu, 05 Jan 2006 21:38:51 -0500

David A. Case wrote:

*> On Wed, Jan 04, 2006, Magne Olufsen wrote:
*

*>
*

*>
*

*>>but why should the snapshots
*

*>>have to be minimized prior to nmode analysis?
*

*>
*

*>
*

*> Conventional normal mode analysis assumes that the structure is at a minimum;
*

*> none of the formulas for estimating vibrational entropies, etc. are correct
*

*> if this is not the case.
*

*>
*

*> Operationally, having an RMS gradient of about 10**-4 kcal/mol-A (or even
*

*> 10**-3) is usually enough to get results are are quite close to those you
*

*> would get from a more precise minimum.
*

*>
*

*> ...dac
*

*>
*

*> -----------------------------------------------------------------------
*

*> The AMBER Mail Reflector
*

*> To post, send mail to amber.scripps.edu
*

*> To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
*

*>
*

Just to add my two cents to Dave's comments:

When the structure is fully minimized, the main approximation of the

theory of normal modes hold. This means that when the potential energy

is expanded to second order in cartesian coordinates, the first order

term in the expansion is exactly zero (the gradient is zero). When that

happens, you get 6 normal modes of zero frequency, NONE with imaginary

frequencies, and 3N-6 positive freqs.

If you do not minimize enough, you will see negative freqs. The trick is

to minimize enough to get 6 'very small' freqs and everything else

separated from that group. It is up to you to see what 'minimized enough

means' but I agree with Dave 10**-3 (-4) might be enough.

Why is this important ? Because fluctuations for instance, depends like

1/freq^2 of the nmode frequency. So, errors in the small freqs could

give large errors in fluctuations.

One more caveat: there is a full body of literature talking about

instantaneous normal modes, where you do nmode without minimization in

snapshots from MD. See papers by Tom Keyes for instance. That is a

different set of ideas that 'regular nmodes, but share some background.

I hope this helps.

Date: Thu, 05 Jan 2006 21:38:51 -0500

David A. Case wrote:

Just to add my two cents to Dave's comments:

When the structure is fully minimized, the main approximation of the

theory of normal modes hold. This means that when the potential energy

is expanded to second order in cartesian coordinates, the first order

term in the expansion is exactly zero (the gradient is zero). When that

happens, you get 6 normal modes of zero frequency, NONE with imaginary

frequencies, and 3N-6 positive freqs.

If you do not minimize enough, you will see negative freqs. The trick is

to minimize enough to get 6 'very small' freqs and everything else

separated from that group. It is up to you to see what 'minimized enough

means' but I agree with Dave 10**-3 (-4) might be enough.

Why is this important ? Because fluctuations for instance, depends like

1/freq^2 of the nmode frequency. So, errors in the small freqs could

give large errors in fluctuations.

One more caveat: there is a full body of literature talking about

instantaneous normal modes, where you do nmode without minimization in

snapshots from MD. See papers by Tom Keyes for instance. That is a

different set of ideas that 'regular nmodes, but share some background.

I hope this helps.

-- Dr. Adrian E. Roitberg Associate Professor Quantum Theory Project and Department of Chemistry University of Florida PHONE 352 392-6972 P.O. Box 118435 FAX 352 392-8722 Gainesville, FL 32611-8435 Email adrian.qtp.ufl.edu ============================================================================ To announce that there must be no criticism of the president, or that we are to stand by the president right or wrong, is not only unpatriotic and servile, but is morally treasonable to the American public." -- Theodore Roosevelt ----------------------------------------------------------------------- The AMBER Mail Reflector To post, send mail to amber.scripps.edu To unsubscribe, send "unsubscribe amber" to majordomo.scripps.eduReceived on Fri Jan 06 2006 - 02:42:03 PST

Custom Search