[Wien] about mini

Jorissen Kevin Kevin.Jorissen at ua.ac.be
Wed Apr 21 22:05:11 CEST 2004


Hi,
 
look at the forces in the first iterations.  Ideally, the atom moves in the direction of its force, making it significantly smaller after each step, and getting near its equilibrium position after a small number of iterations (~10).
If the force gets smaller from iteration to iteration, increase the delta, so the atom will take larger steps in the direction of the force.
If the force changes sign and oscillates around 0 without actually getting there, decrease the deltas to make the atom take smaller steps.
 
If certain coordinates should be fixed, then put the corresponding deltas to zero : then these coordinates will not change at all.  (remember, it's the :FGL forces you should look at, and not :FOR, to determine the force in x,y,z direction).
 
Kevin.
 
 

	-----Original Message----- 
	From: ffffb4ffffe4ffffd3fffff1 ffffb9ffffa2 [mailto:cygeng77 at yahoo.com.cn] 
	Sent: Wed 4/21/2004 4:44 PM 
	To: wien at zeus.theochem.tuwien.ac.at 
	Cc: 
	Subject: Re: [Wien] about mini
	
	

		Dear prof Blaha:
		    There is Ni and Al in my structure. I input 200 k-points with a 0.02 mixing factor.
		    In file *.scf_mini, the forces on some atoms seems more larger, while the forces on the other atoms seems changed sign and the absolute value is large too. Does this means I should put different DELTA on atoms in *.inM file?
		    I attach the struct file in the attachment.I'm looking forward for your suggestion. Any comment is welcome.
		Best regards

	
	
  _____  

	Do You Yahoo!?
	惠普TT游戏剧,玩游戏,中大奖! <http://cn.rd.yahoo.com/mail_cn/tag/SIG=1402c0to2/**http%3A%2F%2Fhp.allyes.com%2Flaserjet%2Fgamestory%2Findex.html%3Fjumpid%3Dex_hphqapcn_MongooseLJ1010%2F201073CN407016%2FYahoo> 

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/ms-tnef
Size: 5582 bytes
Desc: not available
Url : http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20040421/d629177d/attachment.bin


More information about the Wien mailing list