CamCASP/Numerical - Revision history

import>Am592: /* Water Dimer */

2010-11-10T10:41:24Z

Water Dimer

← Older revision		Revision as of 10:41, 10 November 2010
Line 279:		Line 279:
	aDZ -29.0921 -5.0598 -6.8513 24.1837 2.0930 0.7356 -13.9909		aDZ -29.0921 -5.0598 -6.8513 24.1837 2.0930 0.7356 -13.9909
	aTZ -29.0949 -4.9782 -6.8502 24.1769 2.0598 0.7370 -13.9497		aTZ -29.0949 -4.9782 -6.8502 24.1769 2.0598 0.7370 -13.9497
			*10Nov10 -29.0948 -5.0648 -6.8502 24.1769 2.0955 0.7370 -14.0006 <-- SVD instead of LU in DF
	aQZ -29.0948 -5.0598 -6.8502 24.1761 2.0934 0.7372 -13.9982		aQZ -29.0948 -5.0598 -6.8502 24.1761 2.0934 0.7372 -13.9982
	JK-tzvpp -29.0948 -5.0986 -6.8474 24.1844 2.1095 0.7339 -14.0130		JK-tzvpp -29.0948 -5.0986 -6.8474 24.1844 2.1095 0.7339 -14.0130

import>Am592: /* Water Dimer */

2009-05-14T17:48:06Z

Water Dimer

← Older revision		Revision as of 17:48, 14 May 2009
Line 286:		Line 286:
	* This is a lot of data, but what (I believe) is important is the very small variation in the energies when we use the DC/switch=1 combination. The interaction energy varies by less than 0.5% for the four auxiliary basis sets when we use this combination.		* This is a lot of data, but what (I believe) is important is the very small variation in the energies when we use the DC/switch=1 combination. The interaction energy varies by less than 0.5% for the four auxiliary basis sets when we use this combination.
	* Compare this with the 15% variation if the MC/switch=0 combination is used. Even if we skip the aug-cc-pVDZ auxiliary basis (as it is small), the variation in E2int is still about 3% for this case.		* Compare this with the 15% variation if the MC/switch=0 combination is used. Even if we skip the aug-cc-pVDZ auxiliary basis (as it is small), the variation in E2int is still about 3% for this case.
			* But this still large error would be even larger if not for a cancellation in errors from the e-e and e-n integrals. There is no such cancellation in the DC/switch=1 case: here all integrals are accurate.

	====Conclusions?====		====Conclusions?====

import>Am592: /* Helium dimer */

2009-05-14T17:45:51Z

Helium dimer

← Older revision		Revision as of 17:45, 14 May 2009
Line 86:		Line 86:

	Now for SAPT(DFT) energies. Keep in mind that we have no SAPT(DFT) benchmark here. The Korona et al. benchmark is		Now for SAPT(DFT) energies. Keep in mind that we have no SAPT(DFT) benchmark here. The Korona et al. benchmark is
	''Eint(2) = -11.06 K''. Given the numerical accuracies in the SAPT(KS) energies, we should take the SAPT(DFT) energies calculated in the aQZ/Mb1 basis as our benchmark. ~~This is the first row.~~All calculations use ''Switch = 1''.		''Eint(2) = -11.06 K''. Given the numerical accuracies in the SAPT(KS) energies, we should take the SAPT(DFT) energies calculated in the aQZ/Mb1 basis as our benchmark. All calculations use ''Switch = 1''.
	<pre>		<pre>
	Limit E1elst E2ind E2disp E1exch E2exind E2exdisp E2int %Diff		Limit E1elst E2ind E2disp E1exch E2exind E2exdisp E2int %Diff

import>Am592: /* Helium dimer */

2009-05-14T17:42:24Z

Helium dimer

← Older revision		Revision as of 17:42, 14 May 2009
Line 88:		Line 88:
	''Eint(2) = -11.06 K''. Given the numerical accuracies in the SAPT(KS) energies, we should take the SAPT(DFT) energies calculated in the aQZ/Mb1 basis as our benchmark. This is the first row.All calculations use ''Switch = 1''.		''Eint(2) = -11.06 K''. Given the numerical accuracies in the SAPT(KS) energies, we should take the SAPT(DFT) energies calculated in the aQZ/Mb1 basis as our benchmark. This is the first row.All calculations use ''Switch = 1''.
	<pre>		<pre>
	Limit E1elst ~~E1exch~~ ~~E2ind~~ ~~E2exind~~ ~~E2disp~~ E2exdisp ~~E2int~~ %Diff		Limit E1elst E2ind E2disp E1exch E2exind E2exdisp E2int %Diff
	========================================================================================		========================================================================================
	Ref: aQZ/Mb1 G -1.82034 -0.28699 -22.56953 12.79005 0.23734 0.64182 -11.00765 0.0		Ref: aQZ/Mb1 G -1.82034 -0.28699 -22.56953 12.79005 0.23734 0.64182 -11.00765 0.0

import>Am592: /* Water Dimer */

2009-05-14T17:40:27Z

Water Dimer

← Older revision		Revision as of 17:40, 14 May 2009
Line 283:		Line 283:
	=======================================================================================		=======================================================================================
	</pre>		</pre>

			* This is a lot of data, but what (I believe) is important is the very small variation in the energies when we use the DC/switch=1 combination. The interaction energy varies by less than 0.5% for the four auxiliary basis sets when we use this combination.
			* Compare this with the 15% variation if the MC/switch=0 combination is used. Even if we skip the aug-cc-pVDZ auxiliary basis (as it is small), the variation in E2int is still about 3% for this case.

			====Conclusions?====
			* So can we conclude that we have good evidence that the best way to calculate interaction energies is to use the DC auxiliary basis with ''Integral Switch = 1''?
			* I think we have good reason to believe that this is the case, but we need to use an energy scan with multiple orientations to make sure.
			* What are the consequences of this observation for our calculations of density overlaps? The density-fitted density must introduce a good bit of noise into the overlap model. And I cannot advocate calculating the first-order energies using the present ENERGY-SCAN module. We probably need to use the DC auxiliary basis and repeat DF at each geometry. In which case, CamCASP needs to know how to handle DC auxiliary bases efficiently.

import>Am592: /* Water Dimer */

2009-05-14T17:29:43Z

Water Dimer

@@ Line 190: / Line 190: @@
 ---------------------------------------------------------------------------------------
 aDZ        -26.6585   24.2225   -5.0145    2.0169   -9.1100    0.9630
 =======================================================================================
 </pre>
@@ Line 201: / Line 201: @@
 Ref.       -28.8435   24.1824   -5.0944    2.0529   -9.2025    0.9758
 ---------------------------------------------------------------------------------------
 =======================================================================================
 </pre>
@@ Line 213: / Line 213: @@
 Ref.       -28.8435   24.1824   -5.0944    2.0529   -9.2025    0.9758
 ---------------------------------------------------------------------------------------
 =======================================================================================
 </pre>
@@ Line 225: / Line 225: @@
 Ref.       -28.8435   24.1824   -5.0944    2.0529   -9.2025    0.9758
 ---------------------------------------------------------------------------------------
 =======================================================================================
 </pre>

import>Am592: /* Water Dimer */

2009-05-14T17:26:30Z

Water Dimer

@@ Line 183: / Line 183: @@
 First, SAPT(KS) energies. (kJ/mol).
-Amd now SAPT(DFT) energies: (kJ/mol)
+And now SAPT(DFT) energies: (kJ/mol)
 * Aux basis: MC type : Integral switch = 0

import>Am592: /* Water Dimer */

2009-05-14T17:20:04Z

Water Dimer

import>Am592: /* Water Dimer */

2009-05-14T17:18:13Z

Water Dimer

@@ Line 147: / Line 147: @@
 (A) DC+ type for main basis
-First, SAPT(KS) energies. Note that E1exch is calculated in the S2 approximation in CamCASP but not in the reference calculations from SAPT.
+First, SAPT(KS) energies. Note that E1exch is calculated in the S2 approximation in CamCASP but not in the reference calculations from SAPT. All energies in kJ/mol.
 <pre>
                    E1elst   E1exch    E2ind(UC)  E2exind(UC) E2disp(UC) E2exdisp(UC)
@@ Line 158: / Line 158: @@
 </pre>
 * Why the somewhat large error in E1exch? Error due to S2 approximation in CamCASP?
-* 0.7% error in E2ind(UC).
+* 0.7% error in E2ind(UC). Also keep in mind that the reference calculation used Cartesian GTOs in the main basis.
 * Some differences should arise because CamCASP uses the Weighted algorithm for the position of the mid-bond set. So the mid-bonds are located in a different position: (0.0000, 0.0000, 2.8345) in the reference calculation versus (0.0038, 0.0000, 3.6479) in CamCASP. This difference could account for all the small differences in energies.
 So let's say that there are no surprises here and that we have rather good energies.
-Here are the SAPT(DFT) energies:
+Here are the SAPT(DFT) energies: (kJ/mol)
 <pre>
                    E1elst   E1exch    E2ind      E2exind   E2disp    E2exdisp   E2int
@@ Line 180: / Line 180: @@
 (B) MC type for main basis

import>Am592: /* Water Dimer */

2009-05-14T16:35:28Z

Water Dimer

← Older revision		Revision as of 16:35, 14 May 2009
Line 143:		Line 143:
	* Main Basis: Sadlej pVTZ : DC+ and MC types		* Main Basis: Sadlej pVTZ : DC+ and MC types
	* Aux basis: aTZ Basis type as indicated. New mid-bond set used: ''3s3p3d2f1g''		* Aux basis: aTZ Basis type as indicated. New mid-bond set used: ''3s3p3d2f1g''
	* Reference calculations without DF are in /home/am592/DistProp/systems/water/sapt-dft/outputs/sadlej/ Look at geometry (3).		* Reference calculations without DF are in /home/am592/DistProp/systems/water/sapt-dft/outputs/sadlej/ Look at geometry (3).One problem: The main basis in these calculations used Cartesian GTOs! Will cause differences in all energies.

	(A) DC+ type for main basis		(A) DC+ type for main basis

← Older revision		Revision as of 17:20, 14 May 2009
Line 192:		Line 192:
	Ref. -28.8435 -4.8671 -6.8557 24.1824 1.9613 0.7270 -13.6956		Ref. -28.8435 -4.8671 -6.8557 24.1824 1.9613 0.7270 -13.6956
	---------------------------------------------------------------------------------------		---------------------------------------------------------------------------------------
	aDZ -26.6585 -4.8811 -6.7738 24.2225 1.9640 0.7160		aDZ -26.6585 -4.8811 -6.7738 24.2225 1.9640 0.7160 -11.4109
	aTZ -28.4528 -4.9476 -6.8365 24.1657 2.0360 0.7324		aTZ -28.4528 -4.9476 -6.8365 24.1657 2.0360 0.7324 -13.3027
	aQZ -28.6441 -4.9948 -6.8482 24.1766 2.0695 0.7355		aQZ -28.6441 -4.9948 -6.8482 24.1766 2.0695 0.7355 -13.5056
	JK-tzvpp -28.9813 -3.8415 -6.3363 24.1948 1.2562 0.5923		JK-tzvpp -28.9813 -3.8415 -6.3363 24.1948 1.2562 0.5923 -13.1158
	=======================================================================================		=======================================================================================
	</pre>		</pre>
Line 204:		Line 204:
	Ref. -28.8435 -4.8671 -6.8557 24.1824 1.9613 0.7270 -13.6956		Ref. -28.8435 -4.8671 -6.8557 24.1824 1.9613 0.7270 -13.6956
	---------------------------------------------------------------------------------------		---------------------------------------------------------------------------------------
	aDZ -36.8057 -4.8811 -6.7738 24.1126 1.2568 0.7136		aDZ -36.8057 -4.8811 -6.7738 24.1126 1.2568 0.7136 -22.3775
	aTZ -29.1432 -4.9476 -6.8365 24.1849 1.8300 0.7316		aTZ -29.1432 -4.9476 -6.8365 24.1849 1.8300 0.7316 -14.1808
	aQZ -29.5866 -4.9948 -6.8482 24.1755 2.0365 0.7351		aQZ -29.5866 -4.9948 -6.8482 24.1755 2.0365 0.7351 -14.4825
	JK-tzvpp -34.7426 -3.8415 -6.3363 24.1369 0.6432 0.5905		JK-tzvpp -34.7426 -3.8415 -6.3363 24.1369 0.6432 0.5905 -19.5499
	=======================================================================================		=======================================================================================
	</pre>		</pre>