Unexpected behavior with EDA + INPUT_BOHR

Questions
#1

Hi,
I’m trying to perform ALMO-EDA-2 calculation with geometry input in Bohr (INPUT_BOHR TRUE). However it seems that the unit conversion of isolated fragments might be unexpected, while the super system SCF provides correct result. I wonder if there’s any keyword I’m missing.

I’m using Q-Chem version 6.1.1, and here are my two inputs, one in Bohr and one in Angstrom:

$molecule
0 1
--
0 1
O      1.0478417977145795  0.2293957439870774  0.3437411820583848
O      1.755049123119559   0.2054774804284574  2.938563807947287 
H      2.6336867333667073 -0.2189776838623502 -0.4927725431460838
H      0.152124842753612   0.4699427618352133  3.7657706319882727
--
0 1
H      6.687572569210668  -0.432031076324189   1.9307482992771203
C      7.57460245317701   -0.0204090421453027  0.1216681268039969
H      7.976486728636012   2.004128254420106   0.0128898218956583
H      9.343612872904854  -1.077701360208832  -0.029218945338025 
O      5.83617725077525   -0.7691960114690873 -1.763747532470932 
H      6.515764339143586  -0.3730527239765134 -3.4038370565287   
$end

$rem
JOBTYPE                     eda
EDA2                        1
METHOD                      HF
BASIS                       def2-svp
MAX_SCF_CYCLES              100
SCF_CONVERGENCE             10
THRESH                      14
MEM_STATIC                  8000
MEM_TOTAL                   80000
SYMMETRY                    FALSE
SYM_IGNORE                  TRUE
INPUT_BOHR                  TRUE
$end

and

$molecule
0 1
--
0 1
O      0.554494    0.121391    0.181900
O      0.928732    0.108734    1.555021
H      1.393687   -0.115878   -0.260764
H      0.080501    0.248683    1.992760
--
0 1
H      3.538911   -0.228621    1.021708
C      4.008307   -0.010800    0.064384
H      4.220975    1.060539    0.006821
H      4.944427   -0.570295   -0.015462
O      3.088372   -0.407041   -0.933335
H      3.447994   -0.197411   -1.801233
$end

$rem
JOBTYPE                     eda
EDA2                        1
METHOD                      HF
BASIS                       def2-svp
MAX_SCF_CYCLES              100
SCF_CONVERGENCE             10
THRESH                      14
MEM_STATIC                  8000
MEM_TOTAL                   80000
SYMMETRY                    FALSE
SYM_IGNORE                  TRUE
$end

Ideally they should provide almost identical results, but the first input provides:

Simplified EDA Summary (kJ/mol)
--------------------
 PREPARATION      -0.0000
 FROZEN           -10649.2527 (ELEC + PAULI + DISP)
[ELEC + PAULI = -10649.2527, DISP = -0.0000]
 POLARIZATION    -8247.7363
 CHARGE TRANSFER -13.9000
 TOTAL           -18910.8890     (PRP + FRZ + POL + CT)
--------------------

and the second input provides:

Simplified EDA Summary (kJ/mol)
--------------------
 PREPARATION      0.0000
 FROZEN           -11.9578 (ELEC + PAULI + DISP)
[ELEC + PAULI = -11.9578, DISP = -0.0000]
 POLARIZATION    -6.5171
 CHARGE TRANSFER -13.9000
 TOTAL           -32.3749     (PRP + FRZ + POL + CT)
--------------------

Thanks!

#2

Thanks for your post. If you think it’s working when you manually convert the $molecule into Angstroms then this might be a bug. If so I can report. A possible(?) workaround is to include

$rem_frgm
input_bohr true
$end

in your input file, which may force the fragment jobs to recognize that the $molecule coordinates are in Bohr.