Skip to main content Link Search Menu Expand Document (external link)
MamboLab Software Docs

Simulated Data Generator

The simulated data generator it used to create source signals that are coupled according to a brain state dynamics, in which the brain state can be seen as a particular connectivity pattern that remains stable.

The simulated data generator can be found in the /simulator folder in the mambolab repository, and is the simulatedata.m function.

The function needs a cfg structure as input, which has the following elements:

  • cfg.fsample : int, the sampling frequency (default = 256)
  • cfg.LengthTS : float, the approximative length of the simulated timeseries (default : 1)
  • cfg.NumBCS : int, number of brain states to be simulated
  • cfg.RangeLenBCS : cell (n_states x 2), range of the length of brain states occurence (min, max) in sec (default:lengthTS)
  • cfg.ProbTransMat : default uniform distributed transition matrix
  • cfg.NumParcel : number of parcels
  • cfg.NumNodesBCS : vector (1 x n_states) whose i-th component codes the number of parcels involved in the i-th BCS
  • cfg.C : adjacency matrix e.g. cfg.C{i}(1,2):1 denotes setting an interaction from parcel 1 to parcel 2 in the i-th BCs
  • cfg.FreqBandInter : vector (2), band in which the frequency-specific interactions occur (in Hz)
  • cfg.ModelType : string, model used to generate data values: ‘PhaseDelayed’, ‘TimeDelayed’ or ‘AR’ (default:’PhaseDelayed’)
  • cfg.PhaseDelayInter : cell (n_states x 2), range of the phase delay of brain states (only for ‘PhaseDelayed’)
  • cfg.TimeDelayInter : cell (n_states x 2), range of the time delay of brain states (only for ‘TimeDelayed’)
  • cfg.IntARorder : int, order of the AR model (only for ‘AR’)
  • cfg.SeqBCS : timeseries of the BCS occurence (e.g. cfg.SeqBCS:[1 3] means that BCS-1 is the first occurring state, followed by BCS-3)
  • cfg.LenBCSloc : length in seconds of the BCS duration for each occurring state (e.g. cfg.LenBCSloc: [1.5 0.3] means that occurrence of first occuring state is 1.5 sec, while the one of the second occuring state is 0.3 sec.

To run the simulator you need to execute data = simulatedata(cfg), and the output will be a FieldTrip compatible structure with the trial field that contains the simulated timeseries, while the time field contains the timepoints used and the cfg contains the passed parameters.

Example

The file examples/example_simulator.m contains an example for the data generator.

cfg                     = [];
cfg.fsample             = 256;  % sampling frequency
cfg.LengthTS            = 20;   % approximative length of the simulated time series (in sec)
cfg.NumBCS              = 4;    % number of brain connectivity states (BCS) to be simulated
cfg.RangeLenBCS         = {[1 2], [0.1 1], [0.1 1], [0.1 1], [0.1 1]}; % range of the length of BCS occurrence (in sec.)

cfg.NumParcel           = 87;   % number of parcels (or channels)
cfg.NumNodesBCS         = randi([2 5], 1, cfg.NumBCS); % number parcels involved in each BCSs

cfg.C                   = {}; % adjacency matrix