A roots_per_irrep value crashes CC3?

radical

memory 12 GB
molecule radical {
0 2
O 0.0000000000 0.0000000000 0.1081705869
H 0.0000000000 0.0000000000 -0.8651605869
units angstrom
symmetry C1
}
set {
basis “6-31+G(d)”
reference “uhf”
freeze_core true
roots_per_irrep [3]
}
energy(‘eom-cc3’)

Greetings,
When I run the input above, the job crashes.
Tail from the output file:
Collapsing to 6 vector(s).
Change in CC3 energy from last iterated value -0.0049162310
Iter=11 L=6
PBS logs the following error message:
Loading psi4/1.3.2
Loading requirement: intel-mkl/2019.3.199 python3/3.7.4
Intel MKL ERROR: Parameter 3 was incorrect on entry to DGEBAL.
Intel MKL ERROR: Parameter 2 was incorrect on entry to DGEHD2.
forrtl: severe (174): SIGSEGV, segmentation fault occurred
(plus some other stuff)

The same job runs fine when roots_per_irrep [1], roots_per_irrep [2] and roots_per_irrep [4].
oh_1.pof:EOM State 1 0.006 52.1 0.0002375165 -75.546000339049
oh_2.pof:EOM State 1 4.289 34590.9 0.1576076619 -75.388630193629
oh_4.pof:EOM State 1 8.868 71527.1 0.3259015478 -75.220336307723

What am I doing wrong in the case of roots_per_irrep [3]? All comments/feedback will be appreciated.

Many thanks.

I’m not seeing anything obviously wrong with your input file. I’ve done some preliminary debugging, but figuring this out will require expert attention. I’ve turned this into an issue on the Psi GitHub, which is as much as I can do.