Insufficient temporary storage in EvlDn0

Questions
#1

Hello,

I am running a single-point calculation with cdft, and it crashes for a reason I don’t understand.
The input file is:

$molecule
0 9
[106 atoms] 
$end


$basis
O 0
cc-pvtz
****
N 0
cc-pvtz
****
C 0
cc-pvtz
****
H 0
cc-pvtz
****
Cl 0
cc-pvtz
****
Ni   0
[user defined basis] 
****
$END


$rem
exchange 		pbe0       
BASIS2                  6-31G**
BASIS                   gen
xc_grid                 1
sym_ignore              true
max_scf_cycles          2000
SCF_CONVERGENCE         6
thresh                  14
scf_final_print         1
purecart               2222
mem_total              50000
mem_static             25000
basis_lin_dep_thresh		4
s2thresh		14
SCF_algorithm		RCA
cdft                    true 
PLOT_SPIN_DENSITY             true
IANLTY              200
MAKE_CUBE_FILES              true
$end


$cdft
8
1 1 20 spin
$end

$plots
plot orbitals
300 0 19.3319
300 0 34.90702
300 0  41.5513
0 1 0 0 
0 
$end

I see that the small basis SCF calculation successfully converges. However, the large basis calculation ends this way:

---------------------------------------
  Cycle       Energy         DIIS Error
 ---------------------------------------
    1  -10251.3172505226      7.52E-03
[…]
   93  -10386.3093529033      1.59E-06

 Q-Chem fatal error occurred in module 0, line  119:

 Insufficient temporary storage in EvlDn0


 Q-Chem fatal error occurred in module forms2/ShlPairs.C, line 55:

 In ~ShlPairs:  s2DFT != 0.


 Please submit a crash report at q-chem.com/reporter 
 ---

I don’t understand what the issue is?
Any advice appreciated, thanks in advance.

#2

This is an internal buffer overflow error triggered in the DFT quadrature code. If this is latest Q-Chem, and the same job can be run successfully without constrained DFT, then there is probably not much that can be done at this point unfortunately except perhaps to use a different basis set or quadrature grid. Could you please email your full output file to support@q-chem.com for further review? Looking at the full output and reproducing the problem may enable us to provide better suggestions. Thank you.

#3

I repeated the run in QChem 5.3 (I previously used 5.2), and again it crashed the same way.
The same job can be successfully run without cdft.
I will email the file.
Thank you!

#4

Note that it’s the size of the basis set that’s causing problems, XC quadrature grid is already as small as you can safely go. (You are using SG-1; SG-0 is no longer recommended.)

The PBE0 functional is not especially demanding in terms of basis set so I might suggest as a workaround that you try something smaller than cc-pVTZ.

#5

This issue was identified and fixed. The fix will be available with the Q-Chem 5.4 release this May.

1 Like