asteroid_spinprops 0.2.19__tar.gz → 0.2.21__tar.gz

{asteroid_spinprops-0.2.19 → asteroid_spinprops-0.2.21}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: asteroid_spinprops
-Version: 0.2.19
+Version: 0.2.21
 Summary: Collection of tools used for fitting sHG1G2 and ssHG1G2 photometric models to sparse asteroid photometry
 License: MIT
 Author: Odysseas
@@ -13,14 +13,12 @@ Classifier: Programming Language :: Python :: 3.12
 Classifier: Programming Language :: Python :: 3.13
 Classifier: Programming Language :: Python :: 3.14
 Requires-Dist: astropy (==7.0.0)
-Requires-Dist: fink-utils (==0.41.0)
+Requires-Dist: fink-utils (==0.43.0)
 Requires-Dist: matplotlib (==3.10.1)
-Requires-Dist: nifty-ls (==1.0.1)
-Requires-Dist: numpy (==2.2.5)
-Requires-Dist: pandas (==2.2.3)
-Requires-Dist: requests (==2.32.3)
-Requires-Dist: scipy (==1.15.2)
-Requires-Dist: space-rocks (==1.9.13)
+Requires-Dist: nifty-ls (==1.1.0)
+Requires-Dist: pandas (==2.3.3)
+Requires-Dist: scipy (==1.16.2)
+Requires-Dist: space-rocks (==1.9.14)
 Requires-Dist: tqdm (==4.67.1)
 Project-URL: Homepage, https://gitlab.com/odysseas_xenos/asteroid-spinprops
 Project-URL: Repository, https://gitlab.com/odysseas_xenos/asteroid-spinprops

{asteroid_spinprops-0.2.19 → asteroid_spinprops-0.2.21}/asteroid_spinprops/ssolib/periodest.py

@@ -86,6 +86,106 @@ def get_period_estimate(residuals_dataframe, p_min=0.03, p_max=2):
     )
 
 
+def get_multiband_period_estimate(
+    residuals_dataframe, p_min=0.03, p_max=2, k_free=True, k_val=None
+):
+    period_min, period_max = p_min, p_max
+    period_range = (period_min, period_max)
+    results = []
+    residuals = np.zeros(len(residuals_dataframe["filters"].values))
+    bands = np.unique(residuals_dataframe["filters"].values)
+    if k_free:
+        for k in range(1, 11):
+            model = LombScargleMultiband(
+                residuals_dataframe["jd"].values,
+                residuals_dataframe["residuals"].values,
+                residuals_dataframe["filters"].values,
+                residuals_dataframe["sigma"].values,
+                normalization="standard",
+                fit_mean=True,
+                nterms_base=k,
+                nterms_band=1,
+            )
+            frequency, power = model.autopower(
+                method="fast",
+                sb_method="fastnifty_chi2",
+                minimum_frequency=1 / period_range[1],
+                maximum_frequency=1 / period_range[0],
+                samples_per_peak=5,
+            )
+
+            f_best = frequency[np.argmax(power)]
+            y_model = model.model(
+                residuals_dataframe["jd"].values, f_best, bands_fit=bands
+            )
+
+            # y_model.shape is (n_bands, len(time))
+
+            for n, f in enumerate(bands):
+                bindex = np.where(residuals_dataframe["filters"].values == f)
+                residuals[bindex] = (
+                    residuals_dataframe["residuals"].values[bindex] - y_model[n][bindex]
+                )
+
+            rms = np.sqrt(np.mean(residuals**2))
+            n_params = 2 * k + 1 + 3 * len(bands)
+            dof = len(residuals) - n_params
+
+            results.append((k, f_best, rms, dof, n_params))
+
+        model_comparison = pd.DataFrame()
+
+        for i in range(len(results) - 1):
+            k, f_best, rss, dof, n_params = results[i]
+            k_next, f_best_next, rss_next, dof_next, n_params_next = results[i + 1]
+            F = ((rss - rss_next) / (dof - dof_next)) / (rss_next / dof_next)
+
+            # Here crit = Fstat value for which model_2 (more complex) is in fact better than model_1 (less complex)
+            crit = f.ppf(
+                q=0.99,
+                dfn=n_params_next - n_params,
+                dfd=len(residuals) - n_params_next,
+            )
+
+            model_comparison.loc[i, "k"] = k
+            model_comparison.loc[i, "k_next"] = k_next
+            model_comparison.loc[i, "f_best"] = f_best
+            model_comparison.loc[i, "Fstat"] = F
+            model_comparison.loc[i, "alpha_crit"] = crit
+
+        cond = model_comparison["Fstat"] > model_comparison["alpha_crit"]
+        model_comparison = model_comparison[
+            ~cond
+        ]  # don't go for the more complex model
+        f_chosen = model_comparison[
+            model_comparison.k == model_comparison.k.min()
+        ].f_best[0]  # Simplest model
+    if not k_free:
+        model = LombScargleMultiband(
+            residuals_dataframe["jd"].values,
+            residuals_dataframe["residuals"].values,
+            residuals_dataframe["filters"].values,
+            residuals_dataframe["sigma"].values,
+            normalization="standard",
+            fit_mean=True,
+            nterms_base=k_val,
+            nterms_band=1,
+        )
+        frequency, power = model.autopower(
+            method="fast",
+            sb_method="fastnifty_chi2",
+            minimum_frequency=1 / period_range[1],
+            maximum_frequency=1 / period_range[0],
+            samples_per_peak=5,
+        )
+
+        f_best = frequency[np.argmax(power)]
+        f_chosen = f_best
+    period_in = 2 * (1 / f_chosen)
+
+    return period_in
+
+
 def plot_periodograms(signal, window, name=None, axis="frequency"):
     if name is not None:
         r = rocks.Rock(name, datacloud="spins")
@@ -231,18 +331,30 @@ def plot_periodograms(signal, window, name=None, axis="frequency"):
         plt.tight_layout()
 
 
-def perform_residual_resampling(resid_df):
-    sg, w, _ = get_period_estimate(resid_df)
-    Pog = 48 / sg[2][0]  # in hours
+def perform_residual_resampling(resid_df, p_min, p_max, k=1):
+    if k == 1:
+        sg, w, _ = get_period_estimate(resid_df)
+        Pog = 48 / sg[2][0]  # in hours
 
-    # Bootstrap residuals:
-    Pbs = np.zeros(25)
-    for n in range(25):
-        BS_df = resid_df.sample(n=len(resid_df), replace=True)
-        sg, w, _ = get_period_estimate(BS_df)
-        Pbs[n] = 48 / sg[2][0]
-
-    cond = np.abs(Pog - Pbs) / Pog < 1e-2
-    Nbs = np.sum(np.ones(25)[cond])
+        # Bootstrap residuals:
+        Pbs = np.zeros(25)
+        for n in range(25):
+            BS_df = resid_df.sample(n=len(resid_df), replace=True)
+            sg, w, _ = get_period_estimate(BS_df)
+            Pbs[n] = 48 / sg[2][0]
 
+        cond = np.abs(Pog - Pbs) / Pog < 1e-2
+        Nbs = np.sum(np.ones(25)[cond])
+    if k > 1:
+        Pog = 24 / get_multiband_period_estimate(
+            resid_df, p_min=p_min, p_max=p_max, k_free=False, k_val=k
+        )
+        Pbs = np.zeros(25)
+        for n in range(25):
+            BS_df = resid_df.sample(n=len(resid_df), replace=True)
+            Pbs[n] = 24 / get_multiband_period_estimate(
+                BS_df, p_min=p_min, p_max=p_max, k_free=False, k_val=k
+            )
+        cond = np.abs(Pog - Pbs) / Pog < 1e-2
+        Nbs = np.sum(np.ones(25)[cond])
     return BS_df, Nbs

{asteroid_spinprops-0.2.19 → asteroid_spinprops-0.2.21}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "asteroid_spinprops"
-version = "0.2.19"
+version = "0.2.21"
 description = "Collection of tools used for fitting sHG1G2 and ssHG1G2 photometric models to sparse asteroid photometry"
 authors = [
     {name = "Odysseas",email = "odysseas.xenos@proton.me"}
@@ -10,15 +10,13 @@ readme = "README.md"
 requires-python = ">=3.11"
 dependencies = [
     "astropy (==7.0.0)",
-    "fink-utils (==0.41.0)",
+    "fink-utils (==0.43.0)",
     "matplotlib (==3.10.1)",
-    "nifty-ls (==1.0.1)",
-    "numpy (==2.2.5)",
-    "pandas (==2.2.3)",
-    "requests (==2.32.3)",
-    "scipy (==1.15.2)",
+    "nifty-ls (==1.1.0)",
+    "pandas (==2.3.3)",
+    "scipy (==1.16.2)",
     "tqdm (==4.67.1)",
-    "space-rocks (==1.9.13)"
+    "space-rocks (==1.9.14)"
 ]
 
 [tool.poetry]