Open stenczelt opened 3 years ago
this is a more general problem, and Albert said he would code up skipping descriptors which have no data. this is needed for the gap-in-castep as well.
-- Gábor
On 10 Jul 2021, at 10:51, Tamas Stenczel @.***> wrote:
If you specify a gp_coordinate for certain species, explicitly ie. add_species=F, and provide no data for it, then the fitting program exits with a segementation fault much later than when the descriptor counts were printed to be zero for that one.
This happened with turbo_soap in particular now, @mcaroba
I was fitting on data of Fe+H and the descriptors were for C+H:
stderr:
Program received signal SIGSEGV: Segmentation fault - invalid memory reference.
Backtrace for this error:
0 0x7fdc3e8a632a
libAtoms/GAP#1 0x7fdc3e8a5503 libAtoms/QUIP#497 0x7fdc3dad503f libAtoms/QUIP#516 0x560e31daf2b3 libAtoms/QUIP#496 0x560e31df247f libAtoms/GAP#5 0x560e31a35f5f libAtoms/QUIP#517 0x560e31a1d2ed libAtoms/GAP#7 0x560e31a1cc8e libAtoms/GAP#8 0x7fdc3dab7bf6 libAtoms/QUIP#495 0x560e31a1ccc9 libAtoms/GAP#10 0xffffffffffffffff
stdout:
libAtoms::Hello World: 10/07/2021 10:45:15 libAtoms::Hello World: git version https://github.com/libAtoms/QUIP.git,818ca4e-dirty libAtoms::Hello World: QUIP_ARCH linux_x86_64_gfortran_openmp libAtoms::Hello World: compiled on Jul 6 2021 at 16:16:33 libAtoms::Hello World: OpenMP parallelisation with 16 threads WARNING: libAtoms::Hello World: environment variable OMP_STACKSIZE not set explicitly. The default value - system and compiler dependent - may be too small for some applications. libAtoms::Hello World: Random Seed = 38715882 libAtoms::Hello World: global verbosity = 0
Calls to system_timer will do nothing by default
================================ Input parameters ==============================
atoms_filename = //MANDATORY// at_file = train.xyz gap = " soap_turbo central_index=1 n_sparse=400 delta=0.0002460893103680562 n_species=2 species_Z={{1 6}} rcut_hard=4.5 rcut_soft=3.5 alpha_max={{10 10}} l_max=6 atom_sigma_r={{0.3 0.3}} atom_sigma_t={{0.3 0.3}} atom_sigma_r_scaling={{0.10 0.10}} atom_sigma_t_scaling={{0.10 0.10}} amplitude_scaling={{1. 1.}} radial_enhancement=1 basis=poly3gauss scaling_mode=polynomial central_weight={{1. 1.}} f0=0.0 covariance_type=dot_product zeta=4 sparse_method=cur_points add_species=F : soap_turbo central_index=2 n_sparse=400 delta=0.0002460893103680562 n_species=2 species_Z={{1 6}} rcut_hard=4.5 rcut_soft=3.5 alpha_max={{10 10}} l_max=6 atom_sigma_r={{0.3 0.3}} atom_sigma_t={{0.3 0.3}} atom_sigma_r_scaling={{0.10 0.10}} atom_sigma_t_scaling={{0.10 0.10}} amplitude_scaling={{1. 1.}} radial_enhancement=1 basis=poly3gauss scaling_mode=polynomial central_weight={{1. 1.}} f0=0.0 covariance_type=dot_product zeta=4 sparse_method=cur_points add_species=F : distance_Nb order=2 n_sparse=20 cutoff=4.5 cutoff_transition_width=1.0 compact_clusters covariance_type=ard_se theta_uniform=1.0 sparse_method=uniform f0=0.0 add_species=T delta=0.0002460893103680562" e0 = 0.0 local_property0 = 0.0 e0_offset = 0.0 e0_method = average default_kernel_regularisation = //MANDATORY// default_sigma = " 0.005 0.050 0.1 1.0" default_kernel_regularisation_local_property = 0.001 default_local_property_sigma = 0.001 sparse_jitter = 1.0e-8 hessian_displacement = 1.0e-2 hessian_delta = 1.0e-2 baseline_param_filename = quip_params.xml core_param_file = quip_params.xml baseline_ip_args = core_ip_args = energy_parameter_name = QM_energy local_property_parameter_name = local_property force_parameter_name = QM_forces virial_parameter_name = QM_virial stress_parameter_name = stress hessian_parameter_name = hessian config_type_parameter_name = config_type kernel_regularisation_parameter_name = sigma sigma_parameter_name = sigma force_mask_parameter_name = force_mask parameter_name_prefix = config_type_kernel_regularisation = config_type_sigma = kernel_regularisation_is_per_atom = T sigma_per_atom = T do_copy_atoms_file = T do_copy_at_file = F sparse_separate_file = F sparse_use_actual_gpcov = F gap_file = gap_new.xml gp_file = GAP.xml verbosity = NORMAL rnd_seed = -1 openmp_chunk_size = 1 do_ip_timing = F template_file = template.xyz sparsify_only_no_fit = F
======================================== ======================================
============== Gaussian Approximation Potentials - Database fitting ============
Initial parsing of command line arguments finished. Found 3 GAPs. Descriptors have been parsed XYZ file read Unchanged GAP: { soap_turbo central_index=1 n_sparse=400 delta=0.0002460893103680562 n_species=2 species_Z={1 6} rcut_hard=4.5 rcut_soft=3.5 alpha_max={10 10} l_max=6 atom_sigma_r={0.3 0.3} atom_sigma_t={0.3 0.3} atom_sigma_r_scaling={0.10 0.10} atom_sigma_t_scaling={0.10 0.10} amplitude_scaling={1. 1.} radial_enhancement=1 basis=poly3gauss scaling_mode=polynomial central_weight={1. 1.} f0=0.0 covariance_type=dot_product zeta=4 sparse_method=cur_points add_species=F} Unchanged GAP: { soap_turbo central_index=2 n_sparse=400 delta=0.0002460893103680562 n_species=2 species_Z={1 6} rcut_hard=4.5 rcut_soft=3.5 alpha_max={10 10} l_max=6 atom_sigma_r={0.3 0.3} atom_sigma_t={0.3 0.3} atom_sigma_r_scaling={0.10 0.10} atom_sigma_t_scaling={0.10 0.10} amplitude_scaling={1. 1.} radial_enhancement=1 basis=poly3gauss scaling_mode=polynomial central_weight={1. 1.} f0=0.0 covariance_type=dot_product zeta=4 sparse_method=cur_points add_species=F} Old GAP: { distance_Nb order=2 n_sparse=20 cutoff=4.5 cutoff_transition_width=1.0 compact_clusters covariance_type=ard_se theta_uniform=1.0 sparse_method=uniform f0=0.0 add_species=T delta=0.0002460893103680562} New GAP: { distance_Nb order=2 n_sparse=20 cutoff=4.5 cutoff_transition_width=1.0 compact_clusters covariance_type=ard_se theta_uniform=1.0 sparse_method=uniform f0=0.0 delta=0.0002460893103680562 Z={1 1 }} New GAP: { distance_Nb order=2 n_sparse=20 cutoff=4.5 cutoff_transition_width=1.0 compact_clusters covariance_type=ard_se theta_uniform=1.0 sparse_method=uniform f0=0.0 delta=0.0002460893103680562 Z={1 26 }} New GAP: { distance_Nb order=2 n_sparse=20 cutoff=4.5 cutoff_transition_width=1.0 compact_clusters covariance_type=ard_se theta_uniform=1.0 sparse_method=uniform f0=0.0 delta=0.0002460893103680562 Z={26 26 }} Multispecies support added where requested
===================== Report on number of descriptors found ====================
Descriptor: soap_turbo central_index=1 n_sparse=400 delta=0.0002460893103680562 n_species=2 species_Z={1 6} rcut_hard=4.5 rcut_soft=3.5 alpha_max={10 10} l_max=6 atom_sigma_r={0.3 0.3} atom_sigma_t={0.3 0.3} atom_sigma_r_scaling={0.10 0.10} atom_sigma_t_scaling={0.10 0.10} amplitude_scaling={1. 1.} radial_enhancement=1 basis=poly3gauss scaling_mode=polynomial central_weight={1. 1.} f0=0.0 covariance_type=dot_product zeta=4 sparse_method=cur_points add_species=F Number of descriptors: 5 Number of partial derivatives of descriptors: 1485
Descriptor: soap_turbo central_index=2 n_sparse=400 delta=0.0002460893103680562 n_species=2 species_Z={1 6} rcut_hard=4.5 rcut_soft=3.5 alpha_max={10 10} l_max=6 atom_sigma_r={0.3 0.3} atom_sigma_t={0.3 0.3} atom_sigma_r_scaling={0.10 0.10} atom_sigma_t_scaling={0.10 0.10} amplitude_scaling={1. 1.} radial_enhancement=1 basis=poly3gauss scaling_mode=polynomial central_weight={1. 1.} f0=0.0 covariance_type=dot_product zeta=4 sparse_method=cur_points add_species=F Number of descriptors: 0 Number of partial derivatives of descriptors: 0
Descriptor: distance_Nb order=2 n_sparse=20 cutoff=4.5 cutoff_transition_width=1.0 compact_clusters covariance_type=ard_se theta_uniform=1.0 sparse_method=uniform f0=0.0 delta=0.0002460893103680562 Z={1 1 } Number of descriptors: 0 Number of partial derivatives of descriptors: 0
Descriptor: distance_Nb order=2 n_sparse=20 cutoff=4.5 cutoff_transition_width=1.0 compact_clusters covariance_type=ard_se theta_uniform=1.0 sparse_method=uniform f0=0.0 delta=0.0002460893103680562 Z={1 26 } Number of descriptors: 320 Number of partial derivatives of descriptors: 5760
Descriptor: distance_Nb order=2 n_sparse=20 cutoff=4.5 cutoff_transition_width=1.0 compact_clusters covariance_type=ard_se theta_uniform=1.0 sparse_method=uniform f0=0.0 delta=0.0002460893103680562 Z={26 26 } Number of descriptors: 6240 Number of partial derivatives of descriptors: 112320
======================================== ======================================
E0/atom = -0.84167321066128113E+003 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 -0.84167321066128113E+003 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000 0.00000000000000000E+000
========== Report on number of target properties found in training XYZ: ========
Number of target energies (property name: QM_energy) found: 5 Number of target local_properties (property name: local_property) found: 0 Number of target forces (property name: QM_forces) found: 735 Number of target virials (property name: QM_virial) found: 30 Number of target Hessian eigenvalues (property name: hessian) found: 0
================================= End of report ================================ — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.
great, I agree
If you specify a gp_coordinate for certain species, explicitly ie.
add_species=F
, and provide no data for it, then the fitting program exits with a segementation fault much later than when the descriptor counts were printed to be zero for that one.This happened with turbo_soap in particular now, @mcaroba
I was fitting on data of Fe+H and the descriptors were for C+H:
stderr:
stdout: