When trying to run manually after this installation error, it says:
File “/home/leonid/psi4conda/bin/psi4”, line 177, in
import psi4
File “/home/leonid/psi4conda/lib//python3.7/site-packages/psi4/init.py”, line 82, in
from .driver import endorsed_plugins
File “/home/leonid/psi4conda/lib//python3.7/site-packages/psi4/driver/init.py”, line 33, in
from qcelemental import constants
ModuleNotFoundError: No module named ‘qcelemental’
I defer to @loriab on any errors during a conda installation.
Normally whenever you see a ModuleNotFoundError, a good first idea is to pip install the module it can’t find. qcelemental may be special because the version requirements seem to change frequently.
any chance you weren’t explicitly bashing the installer (i.e., bash Psi4conda-1.3.2...)? there’s a post-install script that moves “noarch” packages like qcelemental into the correct location (details and why upon request), and it’s looking like something very fundamental is going wrong. afaik this post-install step has been working fine for months, so your environment or execution are my first inquires.
I was doing exactly as written here https://admiring-tesla-08529a.netlify.com/installs/v132/
the installation was successful till the final step. I managed to install an earlier version
psi4conda-1.2.1-py27-Linux-x86_64 as the temporary solution which works fine.
The pip version was “You are using pip version 10.0.1”, so I updated it to 19.2.1, though haven’t tried to install the py37 version of psi4 yet.
let’s get a which conda python psi4, conda list, and psi4 --version, please. did you already have psi4 present by any means (installer/conda/clone) before starting the 1.3.2 installation? I’m wondering if your environment was set up for an older psi and the new psi is trying to run against that older (and for new psi4, incomplete) set of dependencies.
Maybe executing this little script failed. Haven’t seen that before, but I’d believe anything of csh. If you cd /home/401orca/ly/sourcefolders/psi4conda && ls and there’s a site-packages/ folder sitting there, try cp -r site-packages/* lib/python3.6/site-packages . Report back which conda python psi4, please.
gdb --args python psi4 share/psi4/samples/tu1-h2o-energy/test.in
(gdb) run
Program received signal SIGSEGV, Segmentation fault.
0x00002aaace047012 in __kmp_register_root () from /home/401orca/ly/sourcefolders/psi4conda/lib//python3.6/site-packages/psi4/../../../libiomp5.so
(gdb) bt
#0 0x00002aaace047012 in __kmp_register_root () from /home/401orca/ly/sourcefolders/psi4conda/lib//python3.6/site-packages/psi4/../../../libiomp5.so
#1 0x00002aaace05251b in _INTERNAL_25_______src_kmp_runtime_cpp_6206a45a::__kmp_do_serial_initialize() ()
from /home/401orca/ly/sourcefolders/psi4conda/lib//python3.6/site-packages/psi4/../../../libiomp5.so
#2 0x00002aaace046560 in __kmp_get_global_thread_id_reg () from /home/401orca/ly/sourcefolders/psi4conda/lib//python3.6/site-packages/psi4/../../../libiomp5.so
#3 0x00002aaace034e49 in omp_init_lock@OMP_3.0 () from /home/401orca/ly/sourcefolders/psi4conda/lib//python3.6/site-packages/psi4/../../../libiomp5.so
#4 0x00002aaab360d2ad in psi::timer_init() () from /home/401orca/ly/sourcefolders/psi4conda/lib//python3.6/site-packages/psi4/core.so
#5 0x00002aaab3489326 in psi4_python_module_initialize() () from /home/401orca/ly/sourcefolders/psi4conda/lib//python3.6/site-packages/psi4/core.so
#6 0x00002aaab34870e6 in _ZZN8pybind1112cpp_function10initializeIRPFbvEbJEJNS_4nameENS_5scopeENS_7siblingEEEEvOT_PFT0_DpT1_EDpRKT2_ENUlRNS_6detail13function_callEE_4_FUNESL_ ()
from /home/401orca/ly/sourcefolders/psi4conda/lib//python3.6/site-packages/psi4/core.so
#7 0x00002aaab2f3c046 in pybind11::cpp_function::dispatcher(_object*, _object*, _object*) () from /home/401orca/ly/sourcefolders/psi4conda/lib//python3.6/site-packages/psi4/core.so
#8 0x00002aaaaabba9e4 in _PyCFunction_FastCallDict ()
#9 0x00002aaaaac47dfc in call_function ()
#10 0x00002aaaaac6c94a in _PyEval_EvalFrameDefault ()
#11 0x00002aaaaac42cb9 in PyEval_EvalCodeEx ()
#12 0x00002aaaaac43a4c in PyEval_EvalCode ()
#13 0x00002aaaaac6a26d in builtin_exec ()
#14 0x00002aaaaabbdae1 in PyCFunction_Call ()
#15 0x00002aaaaac71aac in _PyEval_EvalFrameDefault ()
#16 0x00002aaaaac41206 in _PyEval_EvalCodeWithName ()
#17 0x00002aaaaac421cf in fast_function ()
#18 0x00002aaaaac47ed5 in call_function ()
#19 0x00002aaaaac6c94a in _PyEval_EvalFrameDefault ()
#20 0x00002aaaaac41f8b in fast_function ()
#21 0x00002aaaaac47ed5 in call_function ()
#22 0x00002aaaaac6c94a in _PyEval_EvalFrameDefault ()
#23 0x00002aaaaac41f8b in fast_function ()
#24 0x00002aaaaac47ed5 in call_function ()
#25 0x00002aaaaac6c94a in _PyEval_EvalFrameDefault ()
#26 0x00002aaaaac41f8b in fast_function ()
#27 0x00002aaaaac47ed5 in call_function ()
#28 0x00002aaaaac6c94a in _PyEval_EvalFrameDefault ()
#29 0x00002aaaaac425db in _PyFunction_FastCallDict ()
#30 0x00002aaaaabbadaf in _PyObject_FastCallDict ()
#31 0x00002aaaaabfe0e0 in _PyObject_CallMethodIdObjArgs ()
#32 0x00002aaaaabb16e0 in PyImport_ImportModuleLevelObject ()
#33 0x00002aaaaac6f38e in _PyEval_EvalFrameDefault ()
#34 0x00002aaaaac42cb9 in PyEval_EvalCodeEx ()
#35 0x00002aaaaac43a4c in PyEval_EvalCode ()
#36 0x00002aaaaacbfc44 in run_mod ()
#37 0x00002aaaaacc0041 in PyRun_FileExFlags ()
#38 0x00002aaaaacc0244 in PyRun_SimpleFileExFlags ()
#39 0x00002aaaaacc3d24 in Py_Main ()
#40 0x00002aaaaab8b75e in main ()
I think the libiomp5.so may link a wrong libgcc_s.so.1 which is not in /home/401orca/ly/sourcefolders/psi4conda/lib/.I tried to compile from source code, but found that it needs to be networked. Do you know how to compile without network?
If you don’t mind, how about blowing away the broken psi4conda shell script and installation and trying again from this page (https://admiring-tesla-08529a.netlify.com/installs/v132/). Switch to the bash shell before doing anything. And please paste me the full terminal session.
(I recommend giving the 1.3.2 another try, but you can get 1.2.1 off that same page. Some browsers changed the rules about renaming files during a download, so some psi4conda/Psi4conda conventions got messed up. I suspect that’s what’s going on with the 1.2.x page you linked to.)