The bug consisted in the following:
when run in both Serafin or tupac clusters, many schemes (particularly 1 to 6) failed to be loaded.
By inserting some lines in the code that tell us when the program reached different points, we were able to deduce that the problem was that the function extract_parameters_last never returned (eternal loop)
The candidate for this misbehaviour was extract_iter, If it returned true at the end of the file, then we will never leave extract_parameter_last.
What was the return value of extract_iter?
v_num_beta == data.walkers.size()
but v_num_beta was not initialized, so if it kept the value from previous loop it will be equal to data.walkers.size().
So the patch was just initializing v_num_beta (was uninitialized before)
The bug consisted in the following: when run in both Serafin or tupac clusters, many schemes (particularly 1 to 6) failed to be loaded. By inserting some lines in the code that tell us when the program reached different points, we were able to deduce that the problem was that the function extract_parameters_last never returned (eternal loop)
The candidate for this misbehaviour was extract_iter, If it returned true at the end of the file, then we will never leave extract_parameter_last. What was the return value of extract_iter? v_num_beta == data.walkers.size()
but v_num_beta was not initialized, so if it kept the value from previous loop it will be equal to data.walkers.size(). So the patch was just initializing v_num_beta (was uninitialized before)