TDDFT restart in NWChem 6.1 - bgj get scf dens: could not read mo vectors


Clicked A Few Times
Hi,

I'm trying to restart a TDDFT calculation in NWChem 6.1 as I believe that should be now possible. To do so I have modified my input file to:

restart min
basis
C library 6-31G**
H library 6-31G**
end
dft
xc xcamb88 1.00 lyp 0.81 vwn_5 0.19 hfexch 1.00
cam 0.33 cam_alpha 0.19 cam_beta 0.46
direct
iterations 100
end
TDDFT
CIS
NROOTS 6
NOTRIPLET
END
TASK TDDFT ENERGY

When I run this input file in the directory where there also the files from the previous run of the same TDDFT calculation that ran out of time then NWChem starts, proceeds to the TDDFT part of the code and starts the iterative procedure but then crashes with an error message:

 Entering Davidson iterations
Restricted singlet excited states

 Iter   NTrls   NConv    DeltaV     DeltaE      Time   
---- ------ ------ --------- --------- ---------
1 6 0 0.86E-01 0.10+100 1202.0
2 12 0 0.37E-01 0.12E-01 1202.7
3 18 0 0.28E-01 0.84E-02 1202.1
4 24 0 0.29E-01 0.25E-02 1204.7
movecs_read_header: failed to open ./min.movecs IERR =
43
bgj_get_scf_dens movecs output =
./min.movecs

------------------------------------------------------------------------
bgj_get_scf_dens: could not read mo vectors 911
------------------------------------------------------------------------
------------------------------------------------------------------------
current input line :
17: TASK TDDFT ENERGY
------------------------------------------------------------------------
bgj_get_scf_dens movecs output =
------------------------------------------------------------------------
./min.movecs

An error occured while trying to read or write to disk space
------------------------------------------------------------------------
bgj_get_scf_dens movecs output =
For more information see the NWChem manual at
bgj_get_scf_dens movecs output =
./min.movecs

Now I have two questions:

1) What could cause this crash and is it in anyway related to me trying to do a restart?

and

2) If this is a restart and in the previous job it had already found 4 converged vectors why does it start again with 0 converged vectors and stays there until it crashes.

Any suggestions would be more than welcome.

Thanks in advance,

Martijn

Forum Regular
Hi Martijn,

Can you send me your complete input file, if possible ? I would like to investigate the error you reported above. Btw, TDDFT jobs cannot be restarted.

Best wishes,
-Niri
niri.govind@pnnl.gov

Clicked A Few Times
Hi Niri,

Sorry, the files have since been overwritten when I managed to successfully run the same job again. A bit confused though about the lack of restartability of TDDFT calculations. The NWChem manual says that one of the new things in NWchem 6.1 is a "Proper restart for Davidson procedure in TDDFT", which I assumed meant that TDDFT calculations can now be restarted.

Cheers,

Martijn

Forum Regular
Hi Martijn,

It's tricky to work out a true restart for TDDFT because the right-hand-side of the working equations have to be recomputed regardless and this is the crucial bit. I apologize for the documentation. That restart refers to the Davidson solver. We reworked the solver so that it doesn't abort when storage runs out with the expanding subspace. We'll correct the documentation to clarify this.

Please let me know if the job fails and send the files along.

Best wishes,
-Niri
niri.govind@pnnl.gov


Forum >> NWChem's corner >> Running NWChem