no results for ldos

[esp]

Ok I don't want to be a bummer here, ... I set up a Graphene p-i-n structure with a double gate, but as small as possible, width of (3,0) and length of around 4nm .... i ran the LDOS .. it has taken 24 hours already but looks like it might finish this time ... i am just wondering:  Has anyone out there done any LDOS calculations with graphene, and does it normally take this long?  I wanted to use this output as part of an iterative design process, but if it takes this long I can't see how I can .. i just need to know if i am doing something wrong or not ... i have a very powerful machine with tons of memory, hd space, and equiv 24 cores .... i cant imagine that this is normal ... some parameters below:

kPntsX = 1
kPntsY = 1
kPntsZ = 101
ldos_energies = numpy.linspace(-3,3,31)*eV
iteration_control_parameters = IterationControlParameters(number_of_history_steps=12, tolerance=1e-3)
numerical_accuracy_parameters = NumericalAccuracyParameters(k_point_sampling=(kPntsX, kPntsY, kPntsZ))

oh and by the way it has created two files both 4GB in size ... is this normal?
Both exactly the same size, one has .tmp and one does not .. but it still running ...

[esp]

I am wondering if this tells you anything that might help: the log file shows as below at the end, dont worry about the times those come from my script and they do not include time, but for at least 12 hours it has been on that last line stuck at 2.4eV .... not sure what is happening ... "top" command shows 2 processes at 100%, not much memory used about 8% total .. but it is just hanging ... any ideas?  Again, i can't send the nc file because it is 4GB

Code

|  79   H   [  12.000 ,  16.368 ,  30.708 ]    1.05825   0.05825               |
+------------------------------------------------------------------------------+
|  13 E = -93.1637 dE =  7.491420e-06 dH =  4.465593e-04                       |
+------------------------------------------------------------------------------+
| Calculation Converged in 13 steps                                            |
+------------------------------------------------------------------------------+
+------------------------------------------------------------------------------+
|                                                                              |
| Device DFT Calculation  [Finished Sat Mar 17 02:27:38 2012]                  |
|                                                                              |
+------------------------------------------------------------------------------+

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy -3.0 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy -2.8 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy -2.6 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy -2.4 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy -2.2 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy -2.0 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy -1.8 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy -1.6 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy -1.4 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy -1.2 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy -1.0 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy -0.8 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy -0.6 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy -0.4 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy -0.2 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy 0.0 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy 0.2 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy 0.4 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy 0.6 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy 0.8 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy 1.0 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy 1.2 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy 1.4 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy 1.6 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy 1.8 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy 2.0 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy 2.2 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================
energy 2.4 eV done 03-17-12 12:00 AM

                            |--------------------------------------------------|
Calculating LDOS           : ==================================================

[esp]

How about this, forget about what I am doing in code ... is it possible that you (ATK) or anyone out there can take this device in the nc file, and run an LDOS? with bias, without, i dont care, i just want to see an LDOS plot ... the object_id is "cfg" ... it is small device because i wanted to get the calc to finish but it never did .. this is yet again going to be another entire weekend waiting for LDOS on a tiny device on a huge powerful computer .. please anyone, run LDOS for this device with whatever parameters you want (but do it self-consistently) ... if you have success, then please let me know how you did it.  

[esp]

just fyi, i ran this yet another time ... same result .. it ran all the way through but stopped at the same point where it was doing LDOS for 2.5eV ... the output files were both 4GB, and it just hung there for 6 hours ... seemingly doing nothing .. i am rerunning now with half the k points to see if that helps.  Also note i have checked on the linux machine for any file size restrictions and i do not see any ...

[Nordland]

Are you storing the result in the same file every time?

[Nordland]

My laptop is running on crunching some numbers now - and I would surprised if it did not work, so while we wait for it to finish, let me elaborate a  little on the details of the problem with appending extensive to the NC file

NetCDF offers support for only one unlimited variable, and it must be the last variable in the file. Therefore appending to a netcdf file is technical impossible if you dont know the number of entries once you create the netcdf file. On the other hand one is not able to predict what and in what order
a user might desire to put things into a netcdf file.
In ATK 11.8 and the version before this, behind the scenes, when a object is appended to an existing file, a temporary file is created with the original content, and then the new object is added to this file, and then this new file is moved back to the original name. (This is the reason for the *.nc.tmp file you are seeing ).
While this approach is quite strong it has two limitations. First of all you need to be able to have all of the file in memory at one point time. The other thing is somewhat more severe. Due to this lack in netcdf were you have to append to the file by rewriting it from scratch every time you write to it, means it scales problematic with the number of appends. Imagine that you have to add 9 grids to a netcdf file on 9 seperate occasions, and each grid is 100 megabyte of data, it means that you have to do 5 GB disk-work (Gauss summation)

In ATK 12.2 we have developed an extension to netcdf framework, that allow you to append a netcdf file without rewriting the file from scratch every time, and without loading it into memory. It is still an valid netcdf file in all extend but it is "prepared" in a way such we can add data to it with excess IO work.

If you are using 11.8 ATK I would recommend putting the LDOS in seperate files if you are going to have many of them.

[Nordland]

I just took some random parameters, but they will not influence the calculation of the LDOS time ( which takes a couple of seconds.)

Here is your plot.

[esp]

Thank you I am not sure what you mean about appending to the nc file .. I clean the directory before each run ... but thank you I will look at what you have done here .. hopefully i can figure out what is wrong ... my latest run i started at 4am this morning .. it is 12 hours now it has been running so far .. this time i tried with only 50 k points in the z direction hoping that will help, but i need to look at what you did before i try again

[esp]

I noticed the basis set is different ... you used: 
    LDABasis.Carbon_SingleZeta,
    LDABasis.Boron_SingleZeta,
    LDABasis.Hydrogen_SingleZeta,
    LDABasis.Nitrogen_SingleZeta,

I though you were supposed to use Cerda.Carbon for graphene for best results?

[Nordland]

Thank you I am not sure what you mean about appending to the nc file .. I clean the directory before each run ... but thank you I will look at what you have done here .. hopefully i can figure out what is wrong ... my latest run i started at 4am this morning .. it is 12 hours now it has been running so far .. this time i tried with only 50 k points in the z direction hoping that will help, but i need to look at what you did before i try again

Appending means writing to the same nc file.

[Nordland]

I noticed the basis set is different ... you used: 
    LDABasis.Carbon_SingleZeta,
    LDABasis.Boron_SingleZeta,
    LDABasis.Hydrogen_SingleZeta,
    LDABasis.Nitrogen_SingleZeta,

I though you were supposed to use Cerda.Carbon for graphene for best results?

No and Yes. Cerda is a good basis set, but it is only Huckel basis set and can't be used for DFT. SingleZeta is of course the smallest basis set possible, but it was just to show it could be done.

[Nordland]

In your routine doLDOS(), try replacing:

Code

nlsave(nc_files + outFileName + '_ldos.nc', LDOS, object_id="LDOS at %s" % energy)

with

Code

nlsave(nc_files + outFileName + '_ldos_%s.nc'%energy, LDOS, object_id="LDOS at %s" % energy)

[esp]

thank you not a bad idea, then i would have multiple files of smaller size... good idea

[esp]

well it did complete now with less k points .. i guess i will just have to watch it and see what the limit is before it wont finish anymore .. at least on the machines i am using now ... thank you again

[Nordland]

The number of k-points in the z-axis will not have an effect on the calculation time of the local density of states, it might have one the LDOS, but not the calculation time of it.