Added a function to ensure the zero-devision and devision by zero

README.md

@@ -42,7 +42,6 @@ In addition - most of these methods available here have also been implemented in
 
 The BiasAdjustCXX command-line tool is the subject of a publication by Schwertfeger, Benjamin Thomas, Lohmann, Gerrit, and Lipskoch, Henrik (2023) _"Introduction of the BiasAdjustCXX command-line tool for the application of fast and efficient bias corrections in climatic research"_. It provides an insight into the architecture, possible applications and new scientific questions. This publication referencing [BiasAdjustCXX v1.8.1](https://github.com/btschwertfeger/BiasAdjustCXX/tree/v1.8.1) was published in the journal SoftwareX in March 2023 and is available at [https://doi.org/10.1016/j.softx.2023.101379](https://doi.org/10.1016/j.softx.2023.101379).
 
-
 These tool and data structures are developed with the aim of reducing discrepancies between modeled and observed climate data. Historical data is utilized to calibrate variables from current and future time series to achieve distributional properties that closely resemble the possible actual values.
 
 <figure>
@@ -53,7 +52,7 @@ These tool and data structures are developed with the aim of reducing discrepanc
   <figcaption>Figure 1: Schematic representation of a bias adjustment procedure</figcaption>
 </figure>
 
-For instance, modeled data typically indicate values that are colder than the actual values. To address this issue, an adjustment procedure is employed. The figure below illustrates the observed, modeled, and adjusted values, revealing that the delta adjusted time series ($T^{\*DM}{sim,p}$) are significantly more similar to the observed data ($T{obs,p}$) than the raw modeled data ($T_{sim,p}$).
+For instance, modeled data typically indicate values that are colder than the actual values. To address this issue, an adjustment procedure is employed. The figure below illustrates the observed, modeled, and adjusted values, revealing that the delta adjusted time series ($T^{\*DM}_{sim,p}$) are significantly more similar to the observed data ($T{obs,p}$) than the raw modeled data ($T_{sim,p}$).
 
 <figure>
   <img
@@ -264,7 +263,6 @@ BiasAdjustCXX --help
 
 <a name="notes"></a>
 
-
 ## 📍 6. Notes
 
 - For adjusting data using the linear scaling, variance scaling or delta method and the `--no-group` flag: You have to separate the input files by month and then apply the correction for each month individually. e.g. For 30 years of data to correct, you need to prepare the three input data sets so that they first contain all time series for all Januaries and then apply the adjustment for this data set. After that you have to do the same for the rest of the months (see `/examples/example_all_methods.run.sh`).

include/MathUtils.hxx

@@ -66,6 +66,8 @@ class MathUtils {
     static std::vector<int> get_pdf(std::vector<float>& arr, std::vector<double>& bins);
     static std::vector<int> get_cdf(std::vector<float>& arr, std::vector<double>& bins);
     static double interpolate(std::vector<double>& xData, std::vector<double>& yData, double x, bool extrapolate);
+    static double ensure_devidable(double numerator, double denominator, double max_scaling_factor);
+    static float ensure_devidable(float numerator, float denominator, double max_scaling_factor);
 
    private:
 };

src/CMethods.cxx

@@ -180,9 +180,8 @@ void CMethods::Linear_Scaling(
 
         } else if (settings.kind == "mult" || settings.kind == "*") {
             const double
-                adjusted_scaling_factor = get_adjusted_scaling_factor(
-                    MathUtils::mean(v_reference) / MathUtils::mean(v_control),
-
+                adjusted_scaling_factor = MathUtils::ensure_devidable(
+                    MathUtils::mean(v_reference), MathUtils::mean(v_control),
                     settings.max_scaling_factor
                 );
             for (unsigned ts = 0; ts < v_scenario.size(); ts++)
@@ -214,8 +213,8 @@ void CMethods::Linear_Scaling(
             std::vector<float> adj_scaling_factors;
             for (unsigned day = 0; day < 365; day++)
                 adj_scaling_factors.push_back(
-                    get_adjusted_scaling_factor(
-                        ref_365_means[day] / contr_365_means[day],
+                    MathUtils::ensure_devidable(
+                        ref_365_means[day], contr_365_means[day],
                         settings.max_scaling_factor
                     )
                 );
@@ -285,8 +284,8 @@ void CMethods::Variance_Scaling(
             VS1_scen[ts] = LS_scen[ts] - LS_scen_mean;  // Eq. 4
 
         const double
-            adjusted_scaling_factor = get_adjusted_scaling_factor(
-                MathUtils::sd(v_reference) / MathUtils::sd(VS1_contr),
+            adjusted_scaling_factor = MathUtils::ensure_devidable(
+                MathUtils::sd(v_reference), MathUtils::sd(VS1_contr),
                 settings.max_scaling_factor
             );
 
@@ -327,8 +326,8 @@ void CMethods::Variance_Scaling(
 
         std::vector<double> adj_scaling_factors;
         for (unsigned day = 0; day < 365; day++)
-            adj_scaling_factors.push_back(get_adjusted_scaling_factor(
-                ref_365_standard_deviations[day] / VS1_contr_365_standard_deviations[day],
+            adj_scaling_factors.push_back(MathUtils::ensure_devidable(
+                ref_365_standard_deviations[day], VS1_contr_365_standard_deviations[day],
                 settings.max_scaling_factor
             ));
 
@@ -378,8 +377,8 @@ void CMethods::Delta_Method(
                 v_output[ts] = v_reference[ts] + scaling_factor;  // Eq. 1
 
         } else if (settings.kind == "mult" || settings.kind == "*") {
-            const double adjusted_scaling_factor = get_adjusted_scaling_factor(
-                MathUtils::mean(v_scenario) / MathUtils::mean(v_control),
+            const double adjusted_scaling_factor = MathUtils::ensure_devidable(
+                MathUtils::mean(v_scenario), MathUtils::mean(v_control),
                 settings.max_scaling_factor
             );
             for (unsigned ts = 0; ts < v_scenario.size(); ts++)
@@ -411,8 +410,8 @@ void CMethods::Delta_Method(
             std::vector<float> adj_scaling_factors;
             for (unsigned day = 0; day < 365; day++)
                 adj_scaling_factors.push_back(
-                    get_adjusted_scaling_factor(
-                        scen_365_means[day] / contr_365_means[day],
+                    MathUtils::ensure_devidable(
+                        scen_365_means[day], contr_365_means[day],
                         settings.max_scaling_factor
                     )
                 );
@@ -630,12 +629,12 @@ void CMethods::Quantile_Delta_Mapping(
     } else {
         // clang-format off
         for (unsigned ts = 0; ts < v_scenario.size(); ts++) {
-            double delta_basis = MathUtils::interpolate(contr_cdf, v_xbins, epsilon[ts], false);
-            v_output[ts] = (delta_basis == 0)
-                ? (float)(QDM1[ts] * settings.max_scaling_factor)                                                           // Eq. 2.3f.
-                : (float)(QDM1[ts] * get_adjusted_scaling_factor(
-                    v_scenario[ts] / delta_basis, settings.max_scaling_factor  // Eq. 2.3f.
-                ));
+
+            v_output[ts] = QDM1[ts] * MathUtils::ensure_devidable(
+                (double) v_scenario[ts],
+                MathUtils::interpolate(contr_cdf, v_xbins, epsilon[ts], false),
+                settings.max_scaling_factor
+            ); // Eq. 2.3f.
         }
         // clang-format on
     }

src/MathUtils.cxx

@@ -351,3 +351,17 @@ double MathUtils::interpolate(std::vector<double>& xData, std::vector<double>& y
         dydx = (yR - yL) / (xR - xL);  // gradient
     return yL + dydx * (x - xL);       // linear interpolation
 }
+
+double MathUtils::ensure_devidable(double numerator, double denuminator, double max_scaling_factor) {
+    if (denuminator == (double)0)
+        return numerator * max_scaling_factor;
+
+    return numerator / denuminator;
+}
+
+float MathUtils::ensure_devidable(float numerator, float denuminator, double max_scaling_factor) {
+    if (denuminator == (double)0)
+        return numerator * max_scaling_factor;
+
+    return numerator / denuminator;
+}

tests/src/TestMathUtils.cxx

@@ -213,6 +213,15 @@ TEST_F(TestMathUtils, CheckLinearInterpolation2d) {
         ASSERT_EQ(targets_with_extrapolation[i], ::MathUtils::interpolate(xData, yData, x[i], true));
     }
 }
+
+TEST_F(TestMathUtils, CheckEnsureDevidable) {
+    ASSERT_EQ(::MathUtils::ensure_devidable((double)5, (double)5, 10), 1);
+    ASSERT_EQ(::MathUtils::ensure_devidable((double)0, (double)5, 10), 0);
+    ASSERT_EQ(::MathUtils::ensure_devidable((double)5, (double)0, 10), 50);
+    ASSERT_EQ(::MathUtils::ensure_devidable((float)5, (float)5, 10), 1);
+    ASSERT_EQ(::MathUtils::ensure_devidable((float)0, (float)5, 10), 0);
+    ASSERT_EQ(::MathUtils::ensure_devidable((float)5, (float)0, 10), 50);
+}
 }  // namespace
 }  // namespace MathUtils
 }  // namespace TestBiasAdjustCXX