Simulation

This module is to simulate whole data sets.

You can chose to simulate with the same parameters for the whole set or provide different parameters by subject. In the simulate_sample function we sample from PyDream posteriors.

Example Simulation

Simulation Notebook

• Simulate Data from the SceneWalk Model

Full Simulation Doc

Simulate data

scenewalk.simulation.simulate_dataset.simulate(dur_dat, im_dat, densities_dat, sw_model, params=None, start_loc='center', x_path=None, y_path=None, resample_durs=False, verbose=False, save_to=None, custom_id=None)

simulate and save dataset given durations and images

Parameters

dur_datarray

duration vector Subjects[Images[Scanpath[]]]

im_datarray

image numbers vector (starts at 1) - densities_dat : 128x128px densities in the order of number reference

sw_modelscenewalk model object

scenewalk object (if using only one param combination for the whole simulation, pass object with preset param values)

params{None, dict}

if None, just use whatever is in the scenewalk object. Otherwise pass a dictionary of all the parameters your model needs

start_loc{“center” or “data”}

start location of each trial

x_path{None, Array}

if start_loc is set to data, provide x positions of the data

y_path: {None, Array}

if start_loc is set to data, provide y positions of the data

resample_dursbool

set to true if you want durations to be sampled from a fitted gamma distribution for each subject. If False, it uses empirical fixation durations.

verbosebool

set to true for basic progress printing

save_tostr

folder to save the simulation to. If none, it makes a folder in the working directory, named with the simulation id.

custom_idstr

simulation id. if none it gets a timestamp.

Returns

str

simulation id number

scenewalk.simulation.simulate_dataset.simulate_sample(dur_dat, im_dat, densities_dat, sw_model, chains_dict, sample_level, start_loc='center', x_path=None, y_path=None, resample_durs=False, verbose=False, save_to=None, custom_id=None)

simulates dataset given durations and images but sample from the posterior parameter distribution.

Parameters

dur_datarray

duration vector Subjects[Images[Scanpath[]]]

im_datarray

image numbers vector (starts at 1) - densities_dat : 128x128px densities in the order of number reference

sw_modelscenewalk model object

scenewalk object (if using only one param combination for the whole simulation, pass object with preset param values)

chains_dictdict

pass estimated posteriors in a dictionary to sample from

sample_level{“vp”, “trial”, “fix”}

how often do we sample? every subject, every trial, or every fixation?

params{None, dict}

if None, just use whatever is in the scenewalk object. Otherwise pass a dictionary

start_loc{“center” or “data”}

start location of each trial

x_path{None, Array}

if start_loc is set to data, provide x positions of the data

y_path: {None, Array}

if start_loc is set to data, provide y positions of the data

resample_dursbool

set to true if you want durations to be sampled from a fitted gamma distribution for each subject. If False, it uses empirical fixation durations.

verbosebool

set to true for basic progress printing

save_tostr

folder to save the simulation to. If none, it makes a folder in the working directory, named with the simulation id.

custom_idstr

simulation id. if none it gets a timestamp.

Returns

str

simulation id number