Repeating creating nodes of same structure causes growing memory consumption and creation time

[ChenK19]

Hello, I hope to create the same nest structure and adjust some connection weight in the structure for simulation, i accomplish that in a loop, for every iteration, create the nodes and connect them with newly weight, which shows growing memory consumption and creation time for nodes.

maybe i need some help to solve the problem.

The problem is similar to the pynest example in
https://nest-simulator.readthedocs.io/en/v3.3/auto_examples/brunel_alpha_evolution_strategies.html

replace the 'N' with 2000 in simulation_parameters and 'max_generation' to 100 in optimization_parameters at the start of main function

` simulation_parameters = {
'seed': 123,
'dt': 0.1, # (ms) simulation resolution
'sim_time': 1000., # (ms) simulation duration
'warmup_time': 300., # (ms) duration ignored during analysis
'delay': 1.5, # (ms) synaptic delay
'g': None, # relative ratio of excitation and inhibition
'eta': None, # relative strength of external drive
'epsilon': 0.1, # average connectivity of network
'N': 400, # number of neurons in network
'gamma': 0.8, # relative size of excitatory and
# inhibitory population
'N_rec': 40, # number of neurons to record activity from
}

optimization_parameters = {
    'verbosity': 1,             # print progress over generations
    'max_generations': 20,      # maximal number of generations
    'target_rate': 1.89,        # (spikes/s) target rate
    'target_corr': 0.0,         # target correlation
    'target_cv': 1.,            # target coefficient of variation
    'mu': [1., 3.],             # initial mean for search distribution
                                # (mu(g), mu(eta))
    'sigma': [0.15, 0.05],      # initial sigma for search
                                # distribution (sigma(g), sigma(eta))

    # hyperparameters of the fitness function; these are used to
    # compensate for the different typical scales of the
    # individual measures, rate ~ O(1), cv ~ (0.1), corr ~ O(0.01)
    'fitness_weight_rate': 1.,    # relative weight of rate deviation
    'fitness_weight_cv': 10.,     # relative weight of cv deviation
    'fitness_weight_corr': 100.,  # relative weight of corr deviation
}`

my environment is:

• OS: Ubuntu 22.04
• Python-Version: 3.10.5
• NEST-Version: 3.3
• Installation: conda without MPI

[heplesser]

Closing as fix is merged.