Starting a run with Bad Walkers

From Zori Wiki

Bad Walkers: Restarting a calculation from a previous run

Sometimes, especially if running without a correlation function, a random walk generates bad walkers (ie walkers that have an undefined or infinite observable such as energy or who antisymmetric wave function value is zero) and running of a Variational or Diffusion Monte Carlo simulation fails immediately prior to performing the walk. In this case it is essential to look at the file bad-walker.xml and based on its description delete it from the walker files that you are reading in.


The tail of the output screen of such as case looks as following:

Adding random walk to the internal run queue...
Born-Oppenheimer potential is: 20.2821398553
----------------------------------------------------
Beginning of RANDOM WALK and/or OPTIMIZATION Stage!
----------------------------------------------------
Random walk # 1
Diffusion Monte Carlo (One electron at a time with substeps)
------------------------------------------------------------
DMC will walk for 20 steps divided into blocks of 1 steps
Timestep: 0.002000    Number of walkers 8000
_LINE_ WARNING: Determinant failed, and rescaling is turned off...
Calculation of determinant failed at the level of fill_dets.
Can't continue composing the wave function. Determinants failed.
Possible error: Processor 0: Non-finite [ local energy of Psi_T ] value nan
Del2_Psi_T 0.000000e+00  Psi_T 0.000000e+00  Potential 7.329486e+00
OrbScale Alpha 1.000000e+00, OrbScale Beta 1.000000e+00,  CorrScale 0.000000e+00
Antisymmetric Wavefunction data (might not be relevant to this walker):
Det Psi -1.033672e-06 Det Lap 2.220562e-04 Corr Psi 1.000000e+00  Corr Lap 0.000000e+00
Writing bad walker to file: bad-walker.xml
Writing walkers to disk...
 
** ERROR **: Exiting due to non-finite observable in walker
aborting...
Aborted


A core file may be generated and may contain a similar to the message that follows:

Core was generated by `/home/software/bin/zori -i dmc.xml -p psi.xml -r distrib-walkers'.
Program terminated with signal 6, Aborted.
Reading symbols from /usr/lib/libglib-2.0.so.0...done.
Loaded symbols for /usr/lib/libglib-2.0.so.0
Reading symbols from /usr/lib/libgsl.so.0...done.
Loaded symbols for /usr/lib/libgsl.so.0
Reading symbols from /lib/tls/libm.so.6...done.
Loaded symbols for /lib/tls/libm.so.6
Reading symbols from /usr/lib/libz.so.1...done.
Loaded symbols for /usr/lib/libz.so.1
Reading symbols from /lib/tls/libpthread.so.0...done.
Loaded symbols for /lib/tls/libpthread.so.0
Reading symbols from /usr/lib/libpopt.so.0...done.
Loaded symbols for /usr/lib/libpopt.so.0
Reading symbols from /home/software/lib/libhdf5_hl.so.0...done.
Loaded symbols for /home/software/lib/libhdf5_hl.so.0
Reading symbols from /home/software/lib/libhdf5.so.0...done.
Loaded symbols for /home/software/lib/libhdf5.so.0
Reading symbols from /usr/lib/libaio.so.1...done.
Loaded symbols for /usr/lib/libaio.so.1
Reading symbols from /lib/libutil.so.1...done.
Loaded symbols for /lib/libutil.so.1
Reading symbols from /lib/tls/libc.so.6...done.
Loaded symbols for /lib/tls/libc.so.6
Reading symbols from /lib/ld-linux.so.2...done.
Loaded symbols for /lib/ld-linux.so.2
Reading symbols from /lib/libnss_files.so.2...done.
Loaded symbols for /lib/libnss_files.so.2
#0  0x4040ccdf in __libc_sigaction () from /lib/tls/libc.so.6
(gdb) bt
#0  0x4040ccdf in __libc_sigaction () from /lib/tls/libc.so.6
#1  0x4040e4e5 in bsearch () from /lib/tls/libc.so.6
#2  0x40044d38 in g_logv () from /usr/lib/libglib-2.0.so.0
#3  0x40044d74 in g_log () from /usr/lib/libglib-2.0.so.0
#4  0x080d1a7f in validate_energy (w=0x9498d80, s=0x87e9918, observable=2.6527607910806133e-314, observable_name=0x87ab4a0 "[ local energy of Psi_T ]")
    at energy.c:74
#5  0x080d1ae9 in elocal_psi_t (w=0x9498d80, s=0x87e9918) at energy.c:109
#6  0x080de293 in gather_weighted_stat (s=0x87e9918, avg=0x96b1dc0, avg_func=0x80d1a98 <elocal_psi_t>, weight_func=0x8064ee8 <get_walker_weight>,
    GET_MEAN=1, GET_VARIANCE=1, GET_STD_DEV=1, GET_ABS_DEV=0, GET_SKEW=0, GET_KURTOSIS=0) at stats.c:225
#7  0x080d25e7 in elocal_weighted_average (s=0x87e9918, avg=0x96b1dc0) at energy.c:492
#8  0x080d9b9d in walk_dmc_single_substeps (s=0x87e9918, params=0x881bc90) at dmc.c:2624
#9  0x080dd661 in walk (s=0x87e9918) at walk.c:129
#10 0x0805c637 in main (argc=7, argv=0xbfffaec4) at main.c:27


This is perfectly normal and should be taking care of by deleting the offending walkers.

Back to the Zori manual