add fit method

src/polykin/physprops/property_equation.py

@@ -192,13 +192,13 @@ def plot(self,
                 Trange = (273.15, 373.15)
 
         try:
-            shape = getattr(self, '_shape')
+            shape = self._shape
         except AttributeError:
             shape = None
         if shape is not None:
             print("Plot method not yet implemented for array-like equations.")
         else:
-            TK = np.linspace(Trange[0], Trange[1], 100)
+            TK = np.linspace(*Trange, 100)
             y = self.__call__(TK, 'K')
             TC = TK - 273.15
             if Tunit == 'C':
@@ -218,47 +218,71 @@ def plot(self,
         if return_objects:
             return (fig, ax)
 
-    # def fit(self,
-    #         T: FloatVector,
-    #         Y: FloatVector,
-    #         sigmaY: Optional[FloatVector] = None,
-    #         fitonly: list[str]=[],
-    #         log: bool = False,
-    #         plot: bool = True,
-    #         ):
-
-    #     # select parameters to be fitted
-    #     params = self._pnames[0]
-    #     if fitonly:
-    #         params = set(fitonly) & set(params)
-    #     args = {p: getattr(self, p) for p in params}
-
-    #     # log transform
-    #     if log:
-    #         ydata = np.log(Y)
-    #     else:
-    #         ydata = Y
-
-    #     def ffit(x, p):
-    #         for
-    #         self.eval(x)
-
-    #     solution = curve_fit(ffit,
-    #                          xdata=T,
-    #                          ydata=ydata,
-    #                          p0=p0,
-    #                          sigma=sigmaY,
-    #                          absolute_sigma=True,
-    #                          full_output=True)
-    #     if solution[4]:
-    #         popt = solution[0]
-    #         cov = solution[1]
-    #         print("Fit successful")
-    #         print(popt)
-    #         print(cov)
-    #     else:
-    #         print("Fit error: ", solution[3])
-    #     pass
+    def fit(self,
+            T: FloatVector,
+            Y: FloatVector,
+            sigmaY: Optional[FloatVector] = None,
+            fitonly: list[str] = [],
+            logY: bool = False,
+            plot: bool = True,
+            ) -> dict:
+
+        # Current parameter values
+        pnames = self._pnames[0] + self._pnames[1]
+        pdict = {pname: pvalue for pname, pvalue in zip(pnames, self.pvalues)}
+
+        # Select parameters to be fitted
+        pnames_fit = self._pnames[0]
+        if fitonly:
+            pnames_fit = set(fitonly) & set(pnames_fit)
+        p0 = [pdict[pname] for pname in pnames_fit]
+
+        # Fit function
+        def ffit(x, *p):
+            for pname, pvalue in zip(pnames_fit, p):
+                pdict[pname] = pvalue
+            Yfit = self.equation(T=x, **pdict)
+            if logY:
+                Yfit = np.log(Yfit)
+            return Yfit
+
+        solution = curve_fit(ffit,
+                             xdata=T,
+                             ydata=np.log(Y) if logY else Y,
+                             p0=p0,
+                             sigma=sigmaY,
+                             absolute_sigma=True,
+                             full_output=True)
+        result = {}
+        result['success'] = bool(solution[4])
+        if solution[4]:
+            popt = solution[0]
+            pcov = solution[1]
+            print("Fit successful.")
+            for pname, pvalue in zip(pnames_fit, popt):
+                print(f"{pname}: {pvalue}")
+            print("Covariance:")
+            print(pcov)
+            result['covariance'] = pcov
+
+            # Update attributes
+            self.Trange = (min(T), max(T))
+            for pname, pvalue in zip(pnames_fit, popt):
+                pdict[pname] = pvalue
+            self.pvalues = tuple(pdict.values())
+            result['parameters'] = pdict
+
+            # plot
+            if plot:
+                kind = 'semilogy' if logY else 'linear'
+                fig, ax = self.plot(kind=kind, return_objects=True)
+                ax.plot(T, Y, 'o', mfc='none')
+                result['plot'] = (fig, ax)
+        else:
+            print("Fit error: ", solution[3])
+            result['message'] = solution[3]
+
+        return result
 
 
 # %% Functions