CamCASP/Programming/0 - Revision history

import>Ajs1: /* Spherical versus Cartesian auxiliary bases */

2011-12-08T15:20:44Z

Spherical versus Cartesian auxiliary bases

← Older revision		Revision as of 15:20, 8 December 2011
Line 188:		Line 188:
	* The Cartesian d-functions include an s-function. But looking at the basis exponents it is not immediately obvious why this is an advantage as the most diffuse s- and d-functions have similar exponents. Can it be that the extra s-function in the Cartesian d-function, because it is fof the form <math> r^2 e^{-\alpha r^2}</math> is effectively even more diffuse than a standard s-function which is of the form <math> e^{-\alpha r^2}</math>?		* The Cartesian d-functions include an s-function. But looking at the basis exponents it is not immediately obvious why this is an advantage as the most diffuse s- and d-functions have similar exponents. Can it be that the extra s-function in the Cartesian d-function, because it is fof the form <math> r^2 e^{-\alpha r^2}</math> is effectively even more diffuse than a standard s-function which is of the form <math> e^{-\alpha r^2}</math>?
	If so, including even more diffuse s-functions may help. It's worth a try.		If so, including even more diffuse s-functions may help. It's worth a try.

			The problem probably arises because the basis includes p functions such as <math> x e^{-\alpha r^2}</math>, so the density includes substantial contributions from <math> x^2 e^{-2\alpha r^2}</math> etc. and needs <math> (x^2+y^2+z^2) e^{-2\alpha r^2}</math> to describe these properly. Just adding more diffuse s functions won't do it.

	* It occurred to me that we could pinpoint the reason for the loss in accuracy by using the water dimer in a configuration where the penetration part of the electrostatic energy is negligible, but the electrostatic energy is still large because of the permanent moments of water. In this case, both the spherical and Cartesian bases should result in the same energy particularly if we impose the charge constraint.		* It occurred to me that we could pinpoint the reason for the loss in accuracy by using the water dimer in a configuration where the penetration part of the electrostatic energy is negligible, but the electrostatic energy is still large because of the permanent moments of water. In this case, both the spherical and Cartesian bases should result in the same energy particularly if we impose the charge constraint.

import>Am592: /* Propagator */

2010-05-05T17:42:05Z

Propagator

← Older revision		Revision as of 17:42, 5 May 2010
Line 3:		Line 3:
	==Propagator==		==Propagator==
	There have been major changes to the propagator modules. See [[ CamCASP/Programming/7 \| Propagator]] for details.		There have been major changes to the propagator modules. See [[ CamCASP/Programming/7 \| Propagator]] for details.
			--[[User:am592\|alston]] 18:42, 5 May 2010 (BST)

	== ENERGY-SCAN ==		== ENERGY-SCAN ==

import>Am592 at 17:41, 5 May 2010

2010-05-05T17:41:57Z

← Older revision		Revision as of 17:41, 5 May 2010
Line 1:		Line 1:
	[[CamCASP]] => [[CamCASP/Programming \| Programming]] => Status		[[CamCASP]] => [[CamCASP/Programming \| Programming]] => Status

			==Propagator==
			There have been major changes to the propagator modules. See [[ CamCASP/Programming/7 \| Propagator]] for details.

	== ENERGY-SCAN ==		== ENERGY-SCAN ==

import>Am592: /* Test: water dimer */

2009-05-01T16:39:12Z

Test: water dimer

← Older revision		Revision as of 16:39, 1 May 2009
Line 189:		Line 189:
	We need a good reference, but this is now quite hard because the old scripts that could run SAPT are no longer supported. These could and probably should be re-activated.		We need a good reference, but this is now quite hard because the old scripts that could run SAPT are no longer supported. These could and probably should be re-activated.

	But we do have a water scan at a single orientation and this should be used for the present.		But we do have a water scan at a single orientation and this should be used for the present. In addition, I will do a scan with about 50 random geometries.

	Some questions that need answering:		Some questions that need answering:
	* Is MC+ always more accurate than MC?		* Is MC+ always more accurate than MC?
			** Sadlej/MC versus Sadlej/MC+
	* What aux basis do we use for the mid-bonds in the MC+ type?		* What aux basis do we use for the mid-bonds in the MC+ type?
			** Do variations in this basis have a significant impact on energies?
	* Which functions should we add to the MC bases to improve accuracy?		* Which functions should we add to the MC bases to improve accuracy?
	* How sensitive is the '''electrostatic self-energy''' to the auxiliary basis? This quantity is a monomer quantity and should be less sensitive.		* How sensitive is the '''electrostatic self-energy''' to the auxiliary basis? This quantity is a monomer quantity and should be less sensitive.

import>Am592: /* Spherical versus Cartesian auxiliary bases */

2009-04-30T12:08:31Z

Spherical versus Cartesian auxiliary bases

← Older revision		Revision as of 12:08, 30 April 2009
Line 185:		Line 185:
	If so, including even more diffuse s-functions may help. It's worth a try.		If so, including even more diffuse s-functions may help. It's worth a try.
	* It occurred to me that we could pinpoint the reason for the loss in accuracy by using the water dimer in a configuration where the penetration part of the electrostatic energy is negligible, but the electrostatic energy is still large because of the permanent moments of water. In this case, both the spherical and Cartesian bases should result in the same energy particularly if we impose the charge constraint.		* It occurred to me that we could pinpoint the reason for the loss in accuracy by using the water dimer in a configuration where the penetration part of the electrostatic energy is negligible, but the electrostatic energy is still large because of the permanent moments of water. In this case, both the spherical and Cartesian bases should result in the same energy particularly if we impose the charge constraint.

			===Test: water dimer===
			We need a good reference, but this is now quite hard because the old scripts that could run SAPT are no longer supported. These could and probably should be re-activated.

			But we do have a water scan at a single orientation and this should be used for the present.

			Some questions that need answering:
			* Is MC+ always more accurate than MC?
			* What aux basis do we use for the mid-bonds in the MC+ type?
			* Which functions should we add to the MC bases to improve accuracy?
			* How sensitive is the '''electrostatic self-energy''' to the auxiliary basis? This quantity is a monomer quantity and should be less sensitive.

import>Am592: /* Water dimer */

2009-04-14T17:15:37Z

Water dimer

@@ Line 103: / Line 103: @@
 ** E1elst is always correct (apart from issues because of the basis). '''Why is E1exch wrong for Geom3?????'''
 ** Second-order energies are always wrong. '''Why?'''
 ==Are the JK-tzvpp Aux bases any good?==

import>Am592: /* ENERGY-SCAN */

2009-04-09T14:07:49Z

ENERGY-SCAN

← Older revision		Revision as of 14:07, 9 April 2009
Line 6:		Line 6:
	** E1elst, E1exch, E2ind(UC),E2exind(UC),E2exdisp(UC),E2disp(UC), Overlap		** E1elst, E1exch, E2ind(UC),E2exind(UC),E2exdisp(UC),E2disp(UC), Overlap
	* '''REDO_DF_ON_ROTATION = .FALSE.'''		* '''REDO_DF_ON_ROTATION = .FALSE.'''
	** E1elst~~, E1exch~~		** E1elst

	What's not working and why:		What's not working and why:
Line 101:		Line 101:
	** Second-order energies wrong once molecule is rotated. This was expected. See argument above.		** Second-order energies wrong once molecule is rotated. This was expected. See argument above.
	* Spherical/NO-REDO-DF:		* Spherical/NO-REDO-DF:
	** E1elst ~~& E1exch are~~ always correct.		** E1elst is always correct (apart from issues because of the basis). '''Why is E1exch wrong for Geom3?????'''
	** Second-order energies are always wrong. '''Why?'''		** Second-order energies are always wrong. '''Why?'''

import>Am592: /* Spherical versus Cartesian auxiliary bases */

2009-04-07T11:02:07Z

Spherical versus Cartesian auxiliary bases

← Older revision		Revision as of 11:02, 7 April 2009
Line 139:		Line 139:
	* The Cartesian d-functions include an s-function. But looking at the basis exponents it is not immediately obvious why this is an advantage as the most diffuse s- and d-functions have similar exponents. Can it be that the extra s-function in the Cartesian d-function, because it is fof the form <math> r^2 e^{-\alpha r^2}</math> is effectively even more diffuse than a standard s-function which is of the form <math> e^{-\alpha r^2}</math>?		* The Cartesian d-functions include an s-function. But looking at the basis exponents it is not immediately obvious why this is an advantage as the most diffuse s- and d-functions have similar exponents. Can it be that the extra s-function in the Cartesian d-function, because it is fof the form <math> r^2 e^{-\alpha r^2}</math> is effectively even more diffuse than a standard s-function which is of the form <math> e^{-\alpha r^2}</math>?
	If so, including even more diffuse s-functions may help. It's worth a try.		If so, including even more diffuse s-functions may help. It's worth a try.
			* It occurred to me that we could pinpoint the reason for the loss in accuracy by using the water dimer in a configuration where the penetration part of the electrostatic energy is negligible, but the electrostatic energy is still large because of the permanent moments of water. In this case, both the spherical and Cartesian bases should result in the same energy particularly if we impose the charge constraint.

import>Am592: /* What's going on with the Hessians? */

2009-03-08T22:59:08Z

What's going on with the Hessians?

← Older revision		Revision as of 22:59, 8 March 2009
Line 48:		Line 48:

	<math>		<math>
	{R^{A'A}_{k'~~k}}^{-1~~} \langle k' \| \hat{C} \| l' \rangle {R^{A'A}_{l'~~l}}^{-1~~}		R^{AA'}_{kk'} \langle k' \| \hat{C} \| l' \rangle R^{AA'}_{ll'}
	</math>		</math>

import>Am592: /* What's going on with the Hessians? */

2009-03-08T22:57:03Z

What's going on with the Hessians?

← Older revision		Revision as of 22:57, 8 March 2009
Line 42:		Line 42:

	<math>		<math>
	\langle \hat{R} k \| \hat{C} \| \hat{R} l \rangle =		\langle \hat{R} k \| \hat{C} \| \hat{R} l \rangle = {R^{A'A}_{k'k}}^{-1} \langle k' \| \hat{C} \| l' \rangle {R^{A'A}_{l'l}}^{-1}
			</math>

			or

			<math>
			{R^{A'A}_{k'k}}^{-1} \langle k' \| \hat{C} \| l' \rangle {R^{A'A}_{l'l}}^{-1}
	</math>		</math>