Compiling Stan components from Matlab

Before any model fitting can be done, Stan models must be translated to C++ code and compiled. This can be done automatically when using the stan function. In addition, the StanModel class exposes a method for compiling models, as well as the underlying Stan components. For example, to build the stan compiler:

model = StanModel('verbose',true);
model.compile('stanc');

The other main components of CmdStan can also be built (libstan.a, libstanc.a, stansummary, and print (deprecated)). Note that Matlab will be busy until compilation is finished.

We could then define a model and compile that too:

code = {
'data {'
'    int<lower=0> N;'
'    int<lower=0,upper=1> y[N];'
'}'
'parameters {'
'    real<lower=0,upper=1> theta;'
'}'
'model {'
'for (n in 1:N)'
'    y[n] ~ bernoulli(theta);'
'}'
};

model.set('model_code',code,'model_name','bernoulli');
model.compile();

Now we can fit a model, either by calling the sampling method directly

data = struct('N',10,'y',[0, 1, 0, 0, 0, 0, 0, 0, 0, 1]);
fit = model.sampling('data',data);

or by using the stan function and passing in the defined model

fit2 = stan('fit',model,'data',data);