Solution LAB1 kpt convergence - Revision history

Andrea Ferretti: /* Convergence wrt k-points */

2020-12-09T15:36:08Z

Convergence wrt k-points

@@ Line 92: / Line 92: @@
    $>  grep ! scf_nk*out | sed 's/scf_nk//' | sed 's/.out:!.*=//'
    $>  grep ! scf_nk*out | sed 's/scf_nk//' | sed 's/.out:!.*=//' | sort -n > conv_kpt.dat
 At this point we obtain
@@ Line 115: / Line 116: @@
           1728            72      -22.78078456
           4096           145      -22.78074173
 '''Plotting.'''

2020-12-09T12:54:53Z

← Older revision		Revision as of 12:54, 9 December 2020
Line 1:		Line 1:


	* Back to the previous page: [[Structural and electronic properties of semiconductors and metals #~~Exercises~~]]		* Back to the previous page: [[Structural and electronic properties of semiconductors and metals #Convergences]]

2020-12-08T18:44:20Z

2020-12-08T16:42:22Z

← Older revision		Revision as of 16:42, 8 December 2020
Line 1:		Line 1:


	* ~~Prev~~: [[Structural and electronic properties of semiconductors and metals#Exercises]]		* Back to the previous page: [[Structural and electronic properties of semiconductors and metals #Exercises]]
	* Next: [[LabQSM#Link to Selected Laboratories]]

2020-12-08T16:34:01Z

Convergence wrt k-points

← Older revision		Revision as of 16:34, 8 December 2020
Line 32:		Line 32:
	# generate the input file		# generate the input file
	#		#
			cat > $filein <<EOF
			&CONTROL
			[...] # note here you can use shell variables
			/
	[...]		[...]
			EOF
	#		#
	# running		# running

2020-12-02T13:11:53Z

Convergence wrt k-points

← Older revision		Revision as of 13:11, 2 December 2020
Line 110:		Line 110:
	16 4096 145 -22.78074173		16 4096 145 -22.78074173

	'''Plotting'''		'''Plotting.'''

	We are now in the condition to plot the data.		We are now in the condition to plot the data.

Line 120:		Line 121:
	→ data → import → ascii...		→ data → import → ascii...

	Some documentation concerning plotting can be found:		Some '''documentation''' concerning plotting can be found:
	* [https://github.com/max-centre/LabQSM/tree/main/tools ./LabQSM/tools] (GitHub / training material) or		* [https://github.com/max-centre/LabQSM/tree/main/tools ./LabQSM/tools] (GitHub / training material) or
	* [[Plotting and visualization tools]] (this wiki)		* [[Plotting and visualization tools]] (this wiki)

2020-12-02T13:11:06Z

Convergence wrt k-points

2020-12-02T13:08:47Z

Convergence wrt k-points

← Older revision		Revision as of 13:08, 2 December 2020
Line 1:		Line 1:


			* Prev: [[Structural and electronic properties of semiconductors and metals#Exercises]]
			* Next: [[LabQSM#Link to Selected Laboratories]]


	=== Convergence wrt k-points ===		=== Convergence wrt k-points ===
Line 119:		Line 124:
	* [[Plotting and visualization tools]] (this wiki)		* [[Plotting and visualization tools]] (this wiki)

	The ~~output~~		The plot should look like:
			File:Plot conv kpt.jpg

			[[File:Plot conv kpt.jpg\|left\| 180px]]

	'''Comments''':		'''Comments''':

2020-12-02T12:59:24Z

Convergence wrt k-points

← Older revision		Revision as of 12:59, 2 December 2020
Line 105:		Line 105:
	16 4096 145 -22.78074173		16 4096 145 -22.78074173

			'''Plotting'''
			We are now in the condition to plot the data.

			In particular we can plot the total energy (col 4) as a function of the grid generator parameter (col 1)
			$ gnuplot
			> plot "conv_kpt.dat" u 1:4 w lp
			Or
			$ xmgrace
			→ data → import → ascii...

			Some documentation concerning plotting can be found:
			* [https://github.com/max-centre/LabQSM/tree/main/tools ./LabQSM/tools] (GitHub / training material) or
			* [[Plotting and visualization tools]] (this wiki)

			The output

	'''Comments''':		'''Comments''':

2020-12-02T12:44:40Z

Convergence wrt k-points

← Older revision		Revision as of 12:44, 2 December 2020
Line 114:		Line 114:
	'''How to choose such threshold''' ? It depends on what we want to compute. K-point grids are a numerical parameter, and we		'''How to choose such threshold''' ? It depends on what we want to compute. K-point grids are a numerical parameter, and we
	want the convergence wrt such parameter not to impact the description or accuracy of computed physical quantities.		want the convergence wrt such parameter not to impact the description or accuracy of computed physical quantities.
	E.g. If we are interested in the binding energy of a molecule adsorbed on a surface, we want the numerical error bars to be		E.g. If we are interested in the binding energy of a molecule adsorbed on a surface, we want the numerical error bars related to numerical parameters to be
	smaller (one to 2 orders of magnitude, at least) than the physical binding energies.		smaller (one to 2 orders of magnitude, at least) than the physical binding energies (the physical quantities of interest).

			Only in this way we can say to have computed the physical quantity of interest within a given level of theory (e.g. KS-DFT using LDA, implemented using
			plane0waves and pseudopotentials).