Merge pull request #130 from PyFE/heston-mc

OUSV updates

Author: jaehyukchoi

pyfeng/data/ousv_benchmark.xlsx

@@ -253,16 +253,6 @@ def vol_paths(self, tobs):
 
         return var_path
 
-    def cond_states_old(self, var_0, texp):
-
-        tobs = self.tobs(texp)
-        n_dt = len(tobs)
-        var_paths = self.vol_paths(tobs)
-        var_final = var_paths[-1, :]
-        var_avg = spint.simps(var_paths, dx=1, axis=0) / n_dt
-
-        return var_final, var_avg
-
     def cond_states(self, var_0, texp):
 
         tobs = self.tobs(texp)

pyfeng/heston_mc.py

@@ -457,14 +457,14 @@ def cond_states_step(self, vol_0, dt, zn=None):
 
             if self.antithetic:
                 n_path_half = int(n_path//2)
-                z_sin = self.rng_spawn[2].standard_normal(size=(n_sin, n_path_half))
+                z_sin = self.rng_spawn[2].standard_normal(size=(n_path_half, n_sin)).T
                 z_sin = np.stack([z_sin, -z_sin], axis=2).reshape((n_sin, n_path))
 
-                z_gpqr = self.rng_spawn[1].standard_normal(size=(4, n_path_half))
-                z_gpqr = np.stack([z_gpqr, -z_gpqr], axis=2).reshape((4, n_path))
+                z_gpqr = self.rng_spawn[1].standard_normal(size=(n_path_half, 4)).T
+                z_gpqr = np.stack([z_gpqr, -z_gpqr], axis=-1).reshape((4, n_path))
             else:
-                z_sin = self.rng_spawn[2].standard_normal(size=(n_sin, n_path))
-                z_gpqr = self.rng_spawn[1].standard_normal(size=(4, n_path))
+                z_sin = self.rng_spawn[2].standard_normal(size=(n_path, n_sin)).T
+                z_gpqr = self.rng_spawn[1].standard_normal(size=(n_path, 4)).T
 
             z_g = z_gpqr[0, :]
             z_p = z_gpqr[1, :]

pyfeng/ousv.py

@@ -75,9 +75,9 @@ def init_benchmark(cls, set_no=None):
             return df_param
         else:
             df_val = pd.read_excel(file, sheet_name=str(set_no))
-            param = df_param.loc[set_no]
-            args_model = {k: param[k] for k in ("sigma", "theta", "vov", "rho", "mr", "intr", "divr")}
-            args_pricing = {k: param[k] for k in ("texp", "spot")}
+            param = df_param.loc[set_no].to_dict()
+            args_model = {k: param[k] for k in param.keys() & {"sigma", "theta", "vov", "rho", "mr", "intr", "divr"}}
+            args_pricing = {k: param[k] for k in param.keys() & {"texp", "spot"}}
 
             assert df_val.columns[0] == "Strike"
             args_pricing["strike"] = df_val.values[:, 0]