double check number of error estimates

src/signalsnap/spectrum_calculator.py

@@ -698,7 +698,7 @@ def c4_new(self, a_w, a_w_corr):
         s4 = m ** 2 / ((m - 1) * (m - 2) * (m - 3)) * (
                 (m + 1) * xyzw_mean -
                 (m - 1) * (
-                        xy_zw_mean + xz_yw_mean + xw_yz_mean
+                    xy_zw_mean + xz_yw_mean + xw_yz_mean
                 )
         )
 
@@ -733,43 +733,43 @@ def c4(self, a_w, a_w_corr):
 
         ones = to_gpu(np.ones_like(a_w.to_ndarray()[:, :, 0]))
 
-        sum_11c22c = af.matmulNT(a_w * a_w_conj, a_w_corr * a_w_conj_corr)  # d1
+        sum_11c22c = af.matmulNT(a_w * a_w_conj, a_w_corr * a_w_conj_corr) # d1
         sum_11c22c_m = mean(sum_11c22c, dim=2)
 
-        sum_11c2 = af.matmulNT(a_w * a_w_conj, a_w_corr)  # d2
+        sum_11c2 = af.matmulNT(a_w * a_w_conj, a_w_corr) #d2
         sum_11c2_m = mean(sum_11c2, dim=2)
-        sum_122c = af.matmulNT(a_w, a_w_corr * a_w_conj_corr)  # d3
+        sum_122c = af.matmulNT(a_w, a_w_corr * a_w_conj_corr) #d3
         sum_122c_m = mean(sum_122c, dim=2)
-        sum_1c22c = af.matmulNT(a_w_conj, a_w_corr * a_w_conj_corr)  # d4
+        sum_1c22c = af.matmulNT(a_w_conj, a_w_corr * a_w_conj_corr) #d4
         sum_1c22c_m = mean(sum_1c22c, dim=2)
-        sum_11c2c = af.matmulNT(a_w * a_w_conj, a_w_conj_corr)  # d5
+        sum_11c2c = af.matmulNT(a_w * a_w_conj, a_w_conj_corr) #d5
         sum_11c2c_m = mean(sum_11c2c, dim=2)
 
-        sum_11c = a_w * a_w_conj  # d6
+        sum_11c = a_w * a_w_conj # d6
         sum_11c_m = mean(sum_11c, dim=2)
-        sum_22c = a_w_corr * a_w_conj_corr  # d6
+        sum_22c = a_w_corr * a_w_conj_corr #d6
         sum_22c_m = mean(sum_22c, dim=2)
-        sum_12c = af.matmulNT(a_w, a_w_conj_corr)  # d7
+        sum_12c = af.matmulNT(a_w, a_w_conj_corr) #d7
         sum_12c_m = mean(sum_12c, dim=2)
-        sum_1c2 = af.matmulNT(a_w_conj, a_w_corr)  # d8
+        sum_1c2 = af.matmulNT(a_w_conj, a_w_corr) #d8
         sum_1c2_m = mean(sum_1c2, dim=2)
 
-        sum_12 = af.matmulNT(a_w, a_w_corr)  # d9
+        sum_12 = af.matmulNT(a_w, a_w_corr) #d9
         sum_12_m = mean(sum_12, dim=2)
-        sum_1c2c = af.matmulNT(a_w_conj, a_w_conj_corr)  # d9
+        sum_1c2c = af.matmulNT(a_w_conj, a_w_conj_corr) #d9
         sum_1c2c_m = mean(sum_1c2c, dim=2)
 
-        sum_1_m = mean(a_w, dim=2)  # d10
-        sum_1c_m = mean(a_w_conj, dim=2)  # d11
-        sum_2_m = mean(a_w_corr, dim=2)  # d10
-        sum_2c_m = mean(a_w_conj_corr, dim=2)  # d11
+        sum_1_m = mean(a_w, dim=2) # d10
+        sum_1c_m = mean(a_w_conj, dim=2) #d11
+        sum_2_m = mean(a_w_corr, dim=2) # d10
+        sum_2c_m = mean(a_w_conj_corr, dim=2) #d11
 
-        sum_11c_m = af.matmulNT(sum_11c_m, ones)  # d6'
-        sum_22c_m = af.matmulNT(ones, sum_22c_m)  # d6''
-        sum_1_m = af.matmulNT(sum_1_m, ones)  # d10'
-        sum_1c_m = af.matmulNT(sum_1c_m, ones)  # d11'
-        sum_2_m = af.matmulNT(ones, sum_2_m)  # d10''
-        sum_2c_m = af.matmulNT(ones, sum_2c_m)  # d11''
+        sum_11c_m = af.matmulNT(sum_11c_m, ones) # d6'
+        sum_22c_m = af.matmulNT(ones, sum_22c_m) # d6''
+        sum_1_m = af.matmulNT(sum_1_m, ones) # d10'
+        sum_1c_m = af.matmulNT(sum_1c_m, ones) # d11'
+        sum_2_m = af.matmulNT(ones, sum_2_m) #d10''
+        sum_2c_m = af.matmulNT(ones, sum_2c_m) #d11''
 
         s4 = m ** 2 / ((m - 1) * (m - 2) * (m - 3)) * (
                 (m + 1) * sum_11c22c_m - (m + 1) * (sum_11c2_m * sum_2c_m + sum_11c2c_m * sum_2_m +
@@ -948,14 +948,16 @@ def store_single_spectrum(self, single_spectrum, order):
                 else:
                     dim = 2
 
-                if order == self.orders[0]:
+                if order==self.config.orders[0]:
                     self.number_of_error_estimates += 1
 
-                # S_err_gpu_real = af.sqrt(self.config.m_var / (self.config.m_var - 1) * (
+
+
+                #S_err_gpu_real = af.sqrt(self.config.m_var / (self.config.m_var - 1) * (
                 #        af.mean(af.real(self.S_errs[order]) * af.real(self.S_errs[order]), dim=dim) -
                 #        af.mean(af.real(self.S_errs[order]), dim=dim) * af.mean(
                 #    af.real(self.S_errs[order]), dim=dim)))
-                # S_err_gpu_imag = af.sqrt(self.config.m_var / (self.config.m_var - 1) * (
+                #S_err_gpu_imag = af.sqrt(self.config.m_var / (self.config.m_var - 1) * (
                 #        af.mean(af.imag(self.S_errs[order]) * af.imag(self.S_errs[order]), dim=dim) -
                 #        af.mean(af.imag(self.S_errs[order]), dim=dim) * af.mean(
                 #    af.imag(self.S_errs[order]), dim=dim)))
@@ -1274,11 +1276,11 @@ def __store_final_spectra(self, orders, n_chunks, n_windows):
             self.S[order] = self.S_gpu[order].to_ndarray()
 
             if not self.config.turbo_mode:
-                # self.S_err[order] /= n_windows // self.config.m_var * np.sqrt(n_windows)
+                #self.S_err[order] /= n_windows // self.config.m_var * np.sqrt(n_windows)
 
                 print('number of error erstimates:', self.number_of_error_estimates)
                 print('n_windows // self.config.m_var:', (n_windows // self.config.m_var))
-                self.S_err[order] = 1 / (n_windows // self.config.m_var) * np.sqrt(self.S_err[order])
+                self.S_err[order] = 1/(n_windows // self.config.m_var) * np.sqrt(self.S_err[order])
 
                 if self.config.interlaced_calculation:
                     self.S_err[order] /= 2
@@ -1375,6 +1377,7 @@ def reset_and_backend_setup(self):
         """
         af.set_backend(self.config.backend)
         orders = self.process_order()
+        self.orders = orders
         self.__reset_variables(orders, f_lists=self.config.f_lists)
         return orders
 
@@ -1753,7 +1756,6 @@ def calc_spec_poisson_one_spectrum(self, sigma_t=0.14, exp_weighting=True, T_win
         """
 
         orders = self.reset_and_backend_setup()
-        self.orders = orders
 
         # -------data setup---------
         if self.config.data is None: