Dear Frederico Dutilh Novaes, Thanks for your advice. When I use the same fdf input files and put the "SolutionMethod" flag to "diagon", the segmentation fault error occur again. I have checked the memory by "top", it showed that the calculation stoped when 12.9% memory are used. It is same as the situation when the "SolutionMethod" flag is "Transiesta". it is a serial run on the redhat 3.0 with AMD Opteron 244. my systems have 20 atoms both in electrodes and scatter region. I do not have any idea about the error , could you help me?
yours JW Wei > Dear J Wei, > > Well, segmentation fault is never great news. If you start from scratch, > TranSIESTA will first compute a SIESTA density matrix (SystemLabel.DM) and > then the TranSIESTA density matrix (SystemLabel.TSDE), but only if you don't > have a segmentation fault at the beginning of the calculation, as it seems > to be your case. > You could get this error due to the fact that your calculation requires more > memory than what you have. As a first test, please use the same fdf input > files but put the "SolutionMethod" flag to "diagon". Also, is it a parallel > run ? > > Best, > > Frederico. > > On Fri, Nov 20, 2009 at 10:38 AM, jw wei <[email protected]> wrote: >> >> Dear SIESTA developers, >> >> I get the following error in a transiesta calculation with command >> transiesta <aa.fdf> aa.out: Segmentation fault. >> >> I have check the out file, it end with the following messages: >> New_DM. Step: 1 >> Initializing Density Matrix... >> TRANSIESTA: No TS-DensityMatrix file found >> TRANSIESTA: Initialization runs using diagon >> >> InitMesh: MESH = 240 x 240 x 120 = 6912000 >> InitMesh: Mesh cutoff (required, used) = 350.000 360.982 Ry >> >> >> Why not have TS-DensityMatrix file ? I use the parameters of Fast >> example in transiesta/Scat except some necessary correction. >> before this calculation, i have run the Fast examples successfully. >> Many thanks in advance for your help. >> >> yours >> >> J Wei > >
