RI-CCSD calculation crashes without any error message

Questions
#1

I am trying to optimize some molecules containing ferrocene. At first I started optimizing the neutral ferrocene only and cationic ferrocene only. the neutral ferrocene calculation terminated normally but the cationic calculation just ends without any error. This is my input and output:
input
$molecule
1 2
C -2.27569358 1.25528625 1.06623924
C -1.73266905 0.92075215 -0.18161216
C -2.56665159 1.45370853 -1.17375773
C -3.62510578 2.11762741 -0.53908585
C -3.44528385 1.99499568 0.84530841
H -1.86874666 0.99212581 2.02019983
H -2.42086102 1.36864677 -2.23036060
H -4.42935651 2.62848202 -1.02602362
H -4.08813101 2.39577913 1.60096714
Fe -3.62259346 0.15814424 0.00223290
C -4.39068800 -1.31178643 -1.17860526
C -5.42450885 -0.64724848 -0.50516198
C -5.20446550 -0.78707131 0.87177388
C -4.03465031 -1.53802426 1.04932385
C -3.53170854 -1.86231672 -0.21788012
H -4.27713187 -1.38583616 -2.23998248
H -6.23888383 -0.12482637 -0.96207395
H -5.82133317 -0.39014729 1.65076402
H -3.60152390 -1.81513912 1.98767817
H -2.64715261 -2.43050489 -0.41693529
H -0.83831529 0.35732584 -0.34769081
$end

$rem
jobtype opt
METHOD ccsd
BASIS def2-tzvp
AUX_BASIS rimp2-def2-tzvp
CC_MAX_ITER 400
gui 2
GEOM_OPT_MAX_CYCLES=500
MAX_SCF_CYCLES = 800
$end

and the last part of the output file is -
…
70 1.46e-07 3.64e-09
71 1.46e-07 6.97e-10
72 1.48e-07 5.18e-10
73 1.46e-07 1.84e-09
74 1.42e-07 3.94e-09
75 1.30e-07 7.03e-09
76 1.13e-07 7.56e-10
77 9.82e-08 8.42e-09

                               CCSD Lambda converged.

Computing density matrices in double precision…

What is the cause of these calculation failures? How can I successfully run RICCSD calculations?
Thank you!

#2

can you please attach the entire output? Also, can you check your $QCSCRATCH directory, to see if you ran out of disk space.

#3

I ran this calculation on my cluster and successfully made one optimization cycle (I killed it afterwards). Note that the calculation is sensitive to memory available to it and can be speed-up using single precision execution. Here is rem section illustrating the setup:

$rem
jobtype opt
METHOD ccsd
BASIS def2-tzvp
AUX_BASIS rimp2-def2-tzvp
CC_MAX_ITER 400
gui 2
GEOM_OPT_MAX_CYCLES=100
MAX_SCF_CYCLES = 200
CC_MEMORY = 80000 use 80 GB
! Use SP for speedup
CC_SINGLE_PREC = 1
CC_SP_DM = 1
CC_SP_T_CONV = 4
CC_SP_E_CONV = 6
CC_ERASE_DP_INTEGRALS = 1
$end

I ran it on a older node with 12 cores and gave 80 GB of memory. It would be faster on a node with larger memory.