Hi Ronaldo,
Not yet, but it is planned to include multivariates in the AutoAnalyser in future. In short, you can already do multivariate conditionals directly in the code.
To do a multivariate conditional TE, generate the conditional TE code in the AutoAnalyser, then tweak a few of the lines to handle the multivariate conditional. That would include:
1. Telling the estimator to expect a multivariate conditional. This can be by calling one of the initialise() routines that supplies this information, e,g. initialise(int k, int l, int condEmbedDim), or by setting the property "COND_EMBED_LENGTHS" to have a comma-separated array value (e.g. "1,2,1") with a given embedding length for each dimension of the conditional.
2. Defining a 2D double array for the conditional variable, and supplying this in the setObservations() call.
Note that the above applies to the Conditional TE estimator (i.e. univariate source and destination, with a potentially multivariate conditional).
If you want multivariate source or destination instead, I don't think I have a conditional version of that implemented yet (just haven't needed it). You would be best to run a calculation for that by directly calling a conditional mutual information calculator (which handles multivariates for each variable).