calcOpenMMNonbondedAndGBSAForces(): OpenMM has not been initialized

[spirilaurentiu]

OMM::System::getDefaultPeriodicBoxVectors 2.000000 0.000000 0.000000
OMM::System::getDefaultPeriodicBoxVectors 0.000000 2.000000 0.000000
OMM::System::getDefaultPeriodicBoxVectors 0.000000 0.000000 2.000000
OPENMM_EXPLORE Platform::createKernel CalcHarmonicBondForce
OPENMM_EXPLORE HarmonicBondForceImpl::initialize created kernel CalcHarmonicBondForce
OPENMM_EXPLORE Platform::createKernel CalcHarmonicAngleForce
OPENMM_EXPLORE HarmonicAngleForceImpl::initialize created kernel CalcHarmonicAngleForce
OPENMM_EXPLORE Platform::createKernel CalcPeriodicTorsionForce
OPENMM_EXPLORE Platform::createKernel CalcNonbondedForce
OPENMM_EXPLORE Platform::createKernel ApplyAndersenThermostat
OPENMM_EXPLORE Platform::createKernel IntegrateVerletStep
ERROR: OpenMM error during initialization: Error launching CUDA compiler: -1
WARNING: DuMM: failed to initialize OpenMM
NOTE: DuMM: using OpenMM platform ''
Traceback (most recent call last):
File "simulate.rebas.py", line 111, in
context.getWorld(worldIx).addSampler(sampler, robosample.IntegratorType.VERLET, thermostat, True)
RuntimeError: SimTK Exception thrown at Integrator.cpp:431:
Integrator initialization failed apparently because:
ERROR: calcOpenMMNonbondedAndGBSAForces(): OpenMM has not been initialized.

[cezieu]

Error occurs with GfcD 121Å nanodisc with LPSA outerleaflet and POPE:POPG:PVCL 18:1:1 innerleaflet when using flexes
edit:Error also occurs with GfcD 94Å nanodisc with LPSA outerleaflet and POPE:POPG:PVCL innerleaflet 18:1:1 when using flexes

Error does not occur with LPSA-POPC 94Å nanodisc or LPSA-POPC "bilayer" of 1 LPSA molecule+2 POPC molecules
Error does not occur when running only cartesian simulation

Error does not occur even if invalid atom ids ( atoms that have no bonds between them ) are supplied as input for flexes

Will try other systems and update issue

[cezieu]

Ran tests on same 94Å system but without :
POPE: Error
POPG:Error
LPSA: Error
POPE+POPG: No error

PVCL: wierd related error:
ERROR: OpenMM error during initialization: Error launching CUDA compiler: -1
initialTss set to 0.000700 0.005000 0.005000
initialMDSs set to 357 200 200
Partitioning for ThermodynamicState:
N1_wCnt (first non-equil world): -1
N2_wCnt (equilibrium world before N1): 2
Nof equilibrium worlds: 3
Nof nonequilibrium rounds: 0
REX, replicaIx, thermoIx, wIx, T, boostT, ts, mdsteps, DISTORT_OPTION, NU, nofSamples, pe_o, pe_n, pe_set, ke_prop, ke_n, fix_o, fix_n, logSineSqrGamma2_o, logSineSqrGamma2_n , etot_n, etot_proposed, JDetLog , acc , MDorMC
REXdetails, thermoState_C, thermoState_H, replica_X, replica_Y, beta_C, beta_H, uC_X0, uH_Y0, uH_X0, uC_Y0, uC_Xtau, uH_Ytau, uH_Xtau, uC_Ytau, ref_uC_X0, ref_uH_Y0, ref_uH_X0, ref_uC_Y0, ref_uC_Xtau, ref_uH_Ytau, ref_uH_Xtau, ref_uC_Ytau, lnJac_X, lnJac_Y, W_X, W_Y, s_X, s_Y, s_X_1, s_Y_1, qC_s_X, qH_s_Y, qH_s_X_1, qC_s_Y_1, ETerm_equil, WTerm, correctionTerm, acc, unif
ERROR: OpenMM Context is not initialized!
ERROR: OpenMM Context is not initialized!
ERROR: OpenMM Context is not initialized!
RSM, 0
ERROR: OpenMM Context is not initialized!
(etc - these 5 last lines are each written once per production step - if the simulation runs succesfully with prod_steps=10 i get 10 lines with RSM, 0)

Will do a bit more testing