Docstrings are altered to reflect that certain parameters are bounds rather than approximations, and other minor corrections. The bug is a documentation error, timeout only sets an upper bound on the time used, actual time used is controlled by a combination of hidden parameters, num_restarts and num_reads.
I have exposed three new parameters. By setting an effectively infinite value for zmax (number of spin updates), the timeout can now be used to better (though still imperfectly) control wallclock runtime.
samples = tabu_sampler.sample_ising(h = {'a': -0.5, 'b': 1.0}, J = {'ab ': -1},num_restarts = 0,num_reads=100,timeout=1000, lower_bound_z = 2147483648)
This satisfies my use case and solves the bug, but I would like to a couple of other things in a similar vein: expose lambda and alpha parameters through the wrapper; and allow timeout control of the STS subroutines (complementing zmax control).
Pull request closes bug: https://github.com/dwavesystems/dwave-samplers/issues/25
Docstrings are altered to reflect that certain parameters are bounds rather than approximations, and other minor corrections. The bug is a documentation error, timeout only sets an upper bound on the time used, actual time used is controlled by a combination of hidden parameters, num_restarts and num_reads.
I have exposed three new parameters. By setting an effectively infinite value for zmax (number of spin updates), the timeout can now be used to better (though still imperfectly) control wallclock runtime.
samples = tabu_sampler.sample_ising(h = {'a': -0.5, 'b': 1.0}, J = {'ab ': -1},num_restarts = 0,num_reads=100,timeout=1000, lower_bound_z = 2147483648)This satisfies my use case and solves the bug, but I would like to a couple of other things in a similar vein: expose lambda and alpha parameters through the wrapper; and allow timeout control of the STS subroutines (complementing zmax control).