CDFT failed to optimize multipliers...


Clicked A Few Times
Hello

Following in my input which I am trying to calculate, but I am facing repeated errors while performing this task.

start Cu-singlet
title "CDFT for Cu-hedp complex"
geometry units angstroms
Cu 2.367615000 6.820525000 16.141388000
P 4.972409000 5.723075000 14.942223000
P 4.955077000 6.653845000 17.779067000
C 5.991272000 6.496681000 16.264219000
C 7.205286000 5.630274000 16.540669000
O 5.791658000 5.640725000 13.672704000
O 5.742741000 7.317980000 18.896622000
O 6.475099000 7.785565000 15.813621000
O 3.751185000 6.630905000 14.748708000
O 3.754478000 7.537470000 17.384587000
O 1.131839000 6.670513000 17.648291000
O 1.179077000 5.772199000 14.973841000
O 1.955102000 8.965530000 15.723237000
H 5.771655000 8.317512000 15.626893000
H 7.832932000 6.128114000 17.166407000
H 6.968412000 4.884888000 16.951206000
H 7.651145000 5.501000000 15.792432000
H 2.592331483 9.318622247 15.098045469
H 5.822632476 8.260537906 18.732909004
H 6.557859053 6.215418780 13.737995690
Cu 3.341964000 4.181475000 16.966435000
P 0.737171000 5.278925000 18.165601000
P 0.754503000 4.348155000 15.328757000
C -0.281692000 4.505319000 16.843605000
C -1.495706000 5.371727000 16.567155000
O -0.033161000 3.684020000 14.211202000
O -0.765519000 3.216435000 17.294203000
O -0.082078000 5.361275000 19.435120000
O 1.958395000 4.371095000 18.359116000
O 1.955102000 3.464530000 15.723237000
O 3.754478000 2.036470000 17.384587000
O 4.577741000 4.331487000 15.459533000
O 4.530503000 5.229801000 18.133983000
H -1.258832000 6.117112000 16.156618000
H -1.941565000 5.501000000 17.315392000
H -2.123352000 4.873886000 15.941417000
H -0.115055032 4.512056413 19.881595426
H -0.062075000 2.684488000 17.480931000
H -0.163217462 2.747706899 14.378543248
H 4.687139641 1.847467802 17.258015156
end
basis
H library ahlrichs_vdz
C library ahlrichs_vdz
O library ahlrichs_vdz
P library ahlrichs_vdz
Cu library ahlrichs_tzv
end
driver
maxiter 50
end
dft
iterations 1000
xc b3lyp
convergence nolevelshifting
odft
mult 1
cdft 1 20 spin 0.5
cdft 21 40 spin -0.5
end
task dft


but I am facing the error as follows:

0:0:CDFT failed to optimize multipliers, cdft_maxiter=:: 999
(rank:0 hostname:dft pid:28716):ARMCI DASSERT fail. ../../ga-5-2/armci/src/common/armci.c:ARMCI_Error():208 cond:0
 0: ARMCI aborting 999 (0x3e7).
CDFT failed to optimize the multipliers.
multipliers:
1 -0.0089228345
2 0.0089228345
------------------------------------------------------------------------
CDFT failed to optimize multipliers, cdft_maxiter= 999
------------------------------------------------------------------------
An error occured in the Runtime Database
------------------------------------------------------------------------

I upgraded from version 6.0 to 6.3 but the issue has not been resolved. I suspect there must be a simple error somewhere which I am failing to notice... also there is another post in the forum with a similar error being reported but has not been answered and hence I am helpless...

I request you to kindly rectify problem and reply asap as I have submission deadline this weekend. Please help!!

Just Got Here
I don't recognise the basis set you use, but I found that CDFT would fail to converge if the basis wasn't sufficiently diffuse.

You need to have the ability for the CDFT micro iterations to be able to manipulate the potential to impose the constraint.

Clicked A Few Times
Hello Jarvist

Many thanks for the reply. Sorry for the delay as I had almost given up as I received no reply for quite some time. I am considering your suggestion but I want to make a point here. The input I gave here is through what I understood from a paper published in JCP in 2006. Rudra et al. at MIT. They have mentioned the use of Ahlrichs basis set as it is large and its combination with DFT has been shown to reliably approximate the complete basis set limit for exchange couplings. Here is the link: http://scitation.aip.org/content/aip/journal/jcp/124/2/10.1063/1.2145878 please go through it.
And regarding the second line of your reply, I did not understand it completely. Please elaborate it in the light of the paper. And yeah, I just realised my multiplicity might be wrong too, should be 3 instead of 1 for Cu.

Thanks again,

Clicked A Few Times
For a large molecular system it makes sense to converge the CDFT at a smaller basis set
first (e.g. 3-21G) and then project the MOs to the larger basis set. This is usually much
faster than applying the larger basis set right away, because the "initial guess" is way better.

Forum Vet
I have just discovered that the CDFT implementation had a pretty problematic bug since version 6.3.
I have found the fix for it, and the patch for 6.5 is available at

http://nwchemgit.github.io/images/Cdft.patch.gz

Clicked A Few Times
Hello Edo

Thank you for reply. I have just installed v. 6.5 and also added the patch. I am currently testing it and shall soon post my findings. Thanks,

Mic


Forum >> NWChem's corner >> General Topics