Floating exception in tree_node_black_hole_mass_rate_compute_standard

[Andrew Benson]

Reported by Markus Haider:

>> I ran a new tree and got the following problem:
>>
>> 11: Output tree data at t= 1.25 Gyr
>> 7: Output tree data at t= 0.90 Gyr
>> Galacticus experienced a floating point exception - will try to flush
>> data before exiting.
>> => Error occurred in thread 3
>>
>> A fatal error occurred! Backtrace for this error:
>> HDF5-DIAG: Error detected in HDF5 (1.8.7) thread 0:
>> #000: ../../../../src/H5Dio.c line 228 in H5Dwrite(): not a dataset
>> major: Invalid arguments to routine
>> minor: Inappropriate type
>> 5: Output tree data at t= 12.40 Gyr
>> 0: Output tree data at t= 7.91 Gyr
>> Fatal error in IO_HDF5_Write_Dataset_Double_1D():
>> unable to write dataset 'hotHaloUnaccretedMass ' in object 'nodeData'
>> => Error occurred in thread 10
>> HDF5-DIAG: Error detected in HDF5 (1.8.7) thread 0:
>> #000: ../../../../src/H5T.c line 1721 in H5Tclose(): not a datatype
>> major: Invalid arguments to routine
>> minor: Inappropriate type
>>
>> A fatal error occurred! Backtrace for this error:
>> #0 0x2B9734678897
>> #0 0x2B9734678897
>> #1 0x2B9734679DE6
>> #1 0x2B9734679DE6
>> #2 0x2B9734742BB8
>> #2 0x2B9734742BB8
>> #3 0x461B76 in __galacticus_error_MOD_galacticus_signal_handler_sigfpe
>> at galacticus.error.F90:186
>> #3 0x4620FE in __galacticus_error_MOD_galacticus_error_report_char at
>> galacticus.error.F90:114
>> #4 0x3DFF83648F
>> #4 0x462198 in __galacticus_error_MOD_galacticus_error_report_varstr at
>> galacticus.error.F90:92
>> #5 0x4D9E94 in
>> __tree_node_methods_black_hole_MOD_tree_node_black_hole_mass_rate_compute_st
>> andard at objects.tree_node.methods.black_hole.standard.F90:490
>> #6 0x4C99F6 in __merger_trees_MOD_tree_node_compute_derivatives at
>> objects.tree_node.derivatives.inc:1
>> #7 0x4CB933 in tree_node_odes at objects.merger_trees.F90:629
>> #8 0x2B97340DF4BA
>> #9 0x2B97340F3A4E
>> #10 0x4B1557 in __fodeiv2_MOD_fodeiv2_driver_apply at
>> numerical.ODE_solver.ODEIV2.wrapper.F90:487
>> #11 0x4B0D28 in __odeiv2_solver_MOD_odeiv2_solve at
>> numerical.ODE_solver.ODEIV2.F90:143
>> #12 0x4CB351 in __merger_trees_MOD_tree_node_evolve at
>> objects.merger_trees.F90:567
>> #13 0x4A1282 in __merger_trees_evolve_MOD_merger_tree_evolve_to at
>> merger_trees.evolve.F90:263
>> #14 0x475CEE in
>> __galacticus_tasks_evolve_tree_MOD_galacticus_task_evolve_tree._omp_fn.0
>> at galacticus.tasks.evolve_tree.F90:200
>> #15 0x2B973497F829
>> #16 0x3DFFC07CD0
>> #17 0x3DFF8DFD3C
>> #18 0xFFFFFFFFFFFFFFFF
>> #5 0x5B6155 in __io_hdf5_MOD_io_hdf5_write_dataset_double_1d at
>> utility.IO.HDF5.F90:5471
>> #6 0x46EFDC in __galacticus_output_merger_tree_MOD_double_buffer_dump
>> at galacticus.output.merger_tree.F90:364
>> #7 0x470353 in
>> __galacticus_output_merger_tree_MOD_galacticus_merger_tree_output at
>> galacticus.output.merger_tree.F90:218
>> #8 0x475E7E in
>> __galacticus_tasks_evolve_tree_MOD_galacticus_task_evolve_tree._omp_fn.0
>> at galacticus.tasks.evolve_tree.F90:206
>> #9 0x2B973497F829
>> #10 0x3DFFC07CD0
>> #11 0x3DFF8DFD3C
>> #12 0xFFFFFFFFFFFFFFFF
>> /sge-root/sge/AMD64/spool/astro28/job_scripts/99113: line 32: 13836
>> Aborted (core dumped) ./Galacticus.exe
>> ../galacticusStart.xml

[Andrew Benson (abenson)]

Problem occurs because a spheroid component gets a very small radius. When the black hole component attempts to compute the spheroid density (dividing by spheroid radius cubed) this causes a floating overflow. The density is in fact only needed as a ratio with the critical density for coupling to the black hole wind. The critical density has a factor of 1/radius^2 in it. Fix is to compute r^2*density in both cases (which does not affect their ratio).

[Andrew Benson (abenson)]

Fixed by revision 911 http://bazaar.launchpad.net/~abenson/galacticus/v0.9.1/revision/911