Add integration test for SIS (particle filter).

We verify that, in a linear model, the likelihood value obtained using
the SIS algorithm is equal to the likelihood value calculated with a
standard linear Kalman filter.

meson.build

@@ -1767,6 +1767,9 @@ mod_and_m_tests = [
   { 'test' : [ 'ecb/pooled_fgls/test_param_names.mod' ] },
 
   # Particle files
+  { 'test' : [ 'particle/noisy_ar1.mod',
+               'particle/noisy_ar1_nlf_sis.mod'],
+    'extra' : [ 'particle/noisy_ar1_data.m'] },
   { 'test' : [ 'particle/dsge_base2.mod' ],
     'extra' : [ 'particle/risky.m',
                 'particle/extreme.m',

tests/particle/noisy_ar1_nlf_sis.mod

+var x y;
+varexo v;
+
+parameters r;
+
+r = .975;
+
+model;
+  y = x;
+  x = r*x(-1) + v;
+end;
+
+shocks;
+  var v; stderr sqrt(0.02);
+end;
+
+steady;
+
+check;
+
+estimated_params;
+  r,  .975, uniform_pdf,,,0,1;
+  stderr y, sqrt(.2), uniform_pdf,,,0,1;
+  stderr v, sqrt(.02), uniform_pdf,,,0,1;
+end;
+
+varobs y;
+
+
+options_.particle.status = 1; // Required because we intend to use a nonlinear filter with a linear model
+
+
+estimation(nobs=100,
+           mode_compute=0,
+           mh_replic=0,
+           cova_compute=0,
+           mode_file=noisy,
+           datafile=noisy_ar1_data,
+           order=1,
+           resampling=systematic,
+           number_of_particles = 5000000,
+           filter_algorithm=sis);
+
+load optimal_value_with_linear_kalman_filter
+
+if 100*abs((oo_.likelihood_at_initial_parameters-(-ovalue))/(-ovalue))>0.05
+    error('Difference between SIS and linear Kalman filter is bigger than 0.05%%')
+end