Am running a DFT-RB3LYP/6-31G(3d) optimization which terminated with the following error?
"Integral accuracy reduced to 1.0D-05 until final iterations.
Problem detected with inexpensive integrals.
Switching to full accuracy and repeating last cycle.
Spurious integrated density or basis function:
NE= 124 NElCor= 0 El error=2.30D-01 rel=1.85D-03 Tolerance=1.00D-03
Shell 27 absolute error=1.21D-04 Tolerance=1.20D-02
Shell 27 signed error=1.21D-04 Tolerance=1.00D-01
Inaccurate quadrature in CalDSu.
Error termination via Lnk1e in /usr/gaussian/g03/l502.exe at Fri Aug 8 22:36:09 2014.
Job cpu time: 0 days 0 hours 13 minutes 29.8 seconds.
File lengths (MBytes): RWF= 102 Int= 0 D2E= 0 Chk= 5 Scr= 1"
How can I troubleshoot?
Hi,
The error termination is not due to "problem detected with inexpensive integrals", but "Inaccurate quadrature in CalDSu", which implies that the integration grids for DFT functional is not good enough for present case.
Try to use int=fine or int=ultrafine keyword to employ better integration grids.
Dear Maria,
I cant get your problem completely, but still can guess that it may be due to memory problem too. Have you tried your run with increased memory?
kindly reply..
Hi,
The error termination is not due to "problem detected with inexpensive integrals", but "Inaccurate quadrature in CalDSu", which implies that the integration grids for DFT functional is not good enough for present case.
Try to use int=fine or int=ultrafine keyword to employ better integration grids.
Thank you for your answers. I will have a go with the keywords.
p.s. I tried increased memory and number of processors but didn't help.
Dear maria,
I understood ur problem, even if the problem is not solved with increased memory, as tian pointed out use int=ultrafine keyword and try with M06-2x and M05-2x keys.
It may help.....
Another little tip that has worked for me in some ADMP trajectories is to use pure D and F functions (5d 7f). If your current basis set already defaults to these then I guess you're still SOL...
Thanks, Oleg B. Gadzhiev!
I tried adding more processors and memory, Int=UltraFine, but that didn't work. Seems that the problem is really in initial guess wavefunction, so "Guess=INDO" solves the issue (on G09).
http://www.gaussian.com/g_tech/g_ur/k_guess.htm
Sorry, I just noticed that I didn't even include the best way that I have been able to fix the problem (if the ultrafine grid doesn't help), which I have found to be simply using a smaller/different basis set.
It is a bit late to comment the issue, but I have just arrived to such a problem and ready to share my case.
It was a molecule with symmetry which would result in exactly zero integrals. However, my input was not symmetrized, so the program did not recognize that certain integrals cannot be calculated numerically when approaching its zero value.
I recently solved the problem of "Inaccurate quadrature in CalDSu" by adding the statement (integral=NoXCTest) to the input command line. This statement stops any checking to the accuracy of quadrature calculations.
I have also solved this problem using integral=NoXCTest in a calculation with Uranium (SDD basis set)
Venkatesan Ragavendran , Antonio Frontera , Tian Lu, I am facing this kind of problem using m062x/cc-pvtz theoretical level on organic molecules having C,H,O,N atoms only. I also have used int=ultrafine, still the following error appears.
Integral accuracy reduced to 1.0D-05 until final iterations.
Problem detected with inexpensive integrals.
Switching to full accuracy and repeating last cycle.
Spurious integrated density or basis function:
NE= 314 NElCor= 0 El error=8.42D-01 rel=2.68D-03 Tolerance=1.00D-03
Shell 17 absolute error=7.99D-05 Tolerance=1.20D-02
Shell 17 signed error=7.99D-05 Tolerance=1.00D-01
Inaccurate quadrature in CalDSu.
Error termination via Lnk1e.
Any help?
I have only faced this problem once and it was solved by using Integral=NoXCTest as recommended before by Mudar Ahmed Abdulsattar. By using this, the program does not check the accuracy of the quadrature. Good luck
Prasanta Bandyopadhyay Try "scf=novaracc", if it still doesn't work, try "int=superfine". In fact, I strongly suggest changing the basis set from cc-pVTZ to def2-TZVP, because the latter is more suitable for DFT calculations with normal functional and meantime it is more expensive. The problem may also be solved due to change the basis set.
Dear Antonio Frontera
Can you please provide an example of input for using the SDD basis set for Uranium calculation? especially If anyone want to use the relativistic effect also?
e.g. ECP60
Thanks in advances.
Md Al Mamunur Rashid
I used sdd basis set for U as implemented in gaussian-09 without any other specification, I mean using the program default options.
Sample Gaussian input file is below:
%nprocshared=2
%chk=C:\test.chk
#p B3LYP/gen Pseudo=Cards Pop=reg gfinput gfprint scf=(maxcycle=500) Integral(Grid=Ultrafine)
Title Card Required
0 1
molecular coordinates
H, N, O, C 0
6-31g(d)
****
U 0
SDD
****
U 0
SDD
- Badges
- Science topic