Why are you so sure that the memory is sufficent? Do you have a simultaneous abservation of the memory usage? If not so, at least you could try this. Reduce k points size is indeed not suitable, but this could help you to find the reason that you job unexpectly terminated. Another way may be worth to try is use use less cores to run this job, this also can reduce the memory consuming.
Hope this helps. /Zhang 2015-01-27 Guangping Zhang 发件人：ali lahmer <scienc...@gmail.com> 发送时间：2015-01-27 09:22 主题：Re: [SIESTA-L] LDAU calculation run with problem 收件人："siesta-l"<siesta-l@uam.es> 抄送： Dear Zhang thank you very much for your help, you suggest to use large memory or reduce the k points. yes this a good solution when the used memory is less than what will be needed in optical calculations. however, i have sufficient memory in my computer to do this calculations and i will add that i have performed in last optical calculation with the same system size by using 5x5x5 kgrid . in the same time, i have used the siesta code (3.2 version ) to run LDA and GGA calculation and i have used the same k grid without any problem. the other solution which is k grid size reducing is not appropriate since the optical calculation need a dense K grid to get best results. so i think that it exist some other problem thank you and if any one have a suggestion please let me know best regards 2015-01-26 13:27 GMT+01:00 Guangping Zhang <zgp...@126.com>: Tryto use computers with large memory. Optical calculation needs much more memory than scf. Or reduce the k size for optical calculation. This would most solve your problem. -------- 原始邮件 -------- 发件人：ali lahmer <scienc...@gmail.com> 时间：周一 1月26日 12:46 收件人：siesta-l@uam.es 主题：[SIESTA-L] LDAU calculation run with problem >Dear siesta user's >i would calculate the optical properties of ZnO using LDAU method. for this >reason i have instaled the siesta ldau 308 version under cygwin and linux. >i have tried to run the optical prroperties of the unitcell ZnO. >the optical properties was calculated in normal manner with 3x3x3 optical >kgrid >however, when i tried to use more dense kgrid (4x4x4 or 5x5x5) the code run >in scf part only and don't calculate the optical properties with the next >error message: > >$ /home/ldau/Obj/siesta <ZnO.fdf> ZnO.out > >Program received signal SIGABRT: Process abort signal. > >Backtrace for this error: >#0 0x55416812 >#1 0x55496EEB >#2 0x6103048D >Aborted (core dumped) > >please if any one have a suggestion or a solution let me know >best regards > >M Lahmer >Department of physics >university of boumerdes >algeria
