Merge pull request #421 from anpicci/revert-420-modify-plotting-functions-for-custom-bins

Revert "Rebin stacked ratio plots with custom bin edges"

Author: anpicci

analysis/topeft_run2/make_cr_and_sr_plots.py

@@ -1142,7 +1142,6 @@ def _draw_stacked_panel(
     h_mc_sumw2,
     mc_scaled,
     mc_norm_factor,
-    bins,
     *,
     log_scale=False,
     style=None,
@@ -1211,7 +1210,8 @@ def _get_grouped_vals(hist_obj, grouping_map):
 
             mc_sumw2_vals[proc_name] = grouped_vals + fallback_vals
 
-    bins = np.array(bins, dtype=float, copy=True)
+    bins = h_data[{"process": sum}].as_hist({}).axes[var].edges
+    bins = np.append(bins, [bins[-1] + (bins[-1] - bins[-2]) * 0.3])
 
     log_scale_requested = bool(log_scale)
     log_y_baseline = None
@@ -3308,206 +3308,6 @@ def make_region_stacked_ratio_fig(
     if getattr(h_data, "empty", False) and h_data.empty():
         return None
 
-    def _clone_histogram(obj):
-        if obj is None:
-            return None
-        if hasattr(obj, "copy"):
-            return obj.copy()
-        return copy.deepcopy(obj)
-
-    def _extract_nominal_histogram_edges(histogram, axis_name):
-        try:
-            hist_view = histogram[{"process": sum}].as_hist({})
-            edges = np.array(hist_view.axes[axis_name].edges, dtype=float, copy=True)
-            values = np.asarray(hist_view.values(flow=True), dtype=float)
-            if values.size >= 2:
-                values = values[1:-1]
-            else:
-                values = np.asarray(hist_view.values(), dtype=float)
-            return edges, values
-        except Exception:
-            return None, None
-
-    def _locate_edge_index(edges, value):
-        matches = np.nonzero(np.isclose(edges, value, rtol=0, atol=1e-9))[0]
-        if matches.size:
-            return int(matches[0])
-        raise ValueError
-
-    def _rebin_syst_arrays(
-        err_up,
-        err_down,
-        ratio_up,
-        ratio_down,
-        nominal_values,
-        old_edges,
-        new_edges,
-    ):
-        if (
-            nominal_values is None
-            or old_edges is None
-            or new_edges is None
-            or nominal_values.size != max(len(old_edges) - 1, 0)
-        ):
-            return err_up, err_down, ratio_up, ratio_down
-
-        old_edge_count = len(old_edges)
-        new_edge_count = len(new_edges)
-        if old_edge_count < 2 or new_edge_count < 2:
-            return err_up, err_down, ratio_up, ratio_down
-
-        try:
-            start_indices = [
-                _locate_edge_index(old_edges, new_edges[idx])
-                for idx in range(new_edge_count - 1)
-            ]
-            end_indices = [
-                _locate_edge_index(old_edges, new_edges[idx + 1])
-                for idx in range(new_edge_count - 1)
-            ]
-        except ValueError:
-            logger.warning(
-                "Unable to rebin systematic arrays: custom bin edges do not align with the original histogram binning."
-            )
-            return err_up, err_down, ratio_up, ratio_down
-
-        nominal_values = np.asarray(nominal_values, dtype=float)
-        rebinned_nominal = np.zeros(new_edge_count - 1, dtype=float)
-
-        if err_up is not None:
-            err_up = np.asarray(err_up, dtype=float)
-            syst_up_deltas = np.clip(err_up - nominal_values, a_min=0, a_max=None)
-            rebinned_err_up = np.zeros_like(rebinned_nominal)
-        else:
-            syst_up_deltas = None
-            rebinned_err_up = None
-
-        if err_down is not None:
-            err_down = np.asarray(err_down, dtype=float)
-            syst_down_deltas = np.clip(nominal_values - err_down, a_min=0, a_max=None)
-            rebinned_err_down = np.zeros_like(rebinned_nominal)
-        else:
-            syst_down_deltas = None
-            rebinned_err_down = None
-
-        for idx, (start, end) in enumerate(zip(start_indices, end_indices)):
-            if end < start:
-                start, end = end, start
-            rebinned_nominal[idx] = nominal_values[start:end].sum()
-            if rebinned_err_up is not None:
-                delta = syst_up_deltas[start:end]
-                rebinned_err_up[idx] = np.clip(
-                    rebinned_nominal[idx] + np.sqrt(np.sum(delta**2)),
-                    a_min=0,
-                    a_max=None,
-                )
-            if rebinned_err_down is not None:
-                delta = syst_down_deltas[start:end]
-                rebinned_err_down[idx] = np.clip(
-                    rebinned_nominal[idx] - np.sqrt(np.sum(delta**2)),
-                    a_min=0,
-                    a_max=None,
-                )
-
-        if rebinned_err_up is not None:
-            rebinned_ratio_up = _safe_divide(
-                rebinned_err_up,
-                rebinned_nominal,
-                default=1.0,
-                zero_over_zero=1.0,
-            )
-        else:
-            rebinned_ratio_up = ratio_up
-
-        if rebinned_err_down is not None:
-            rebinned_ratio_down = _safe_divide(
-                rebinned_err_down,
-                rebinned_nominal,
-                default=1.0,
-                zero_over_zero=1.0,
-            )
-        else:
-            rebinned_ratio_down = ratio_down
-
-        return (
-            rebinned_err_up if rebinned_err_up is not None else err_up,
-            rebinned_err_down if rebinned_err_down is not None else err_down,
-            rebinned_ratio_up,
-            rebinned_ratio_down,
-        )
-
-    orig_edges, orig_nominal_vals = _extract_nominal_histogram_edges(h_mc, var)
-    bins_array = None
-    if bins:
-        rebin_edges = np.array(bins, dtype=float, copy=True)
-        axis_name = var
-        try:
-            axis_name = getattr(h_mc.axes[var], "name", var)
-        except Exception:
-            axis_name = var
-
-        new_axis = hist.axis.Variable(rebin_edges, name=axis_name)
-
-        h_mc = _clone_histogram(h_mc)
-        h_data = _clone_histogram(h_data)
-        h_mc = h_mc.rebin(axis_name, new_axis)
-        h_data = h_data.rebin(axis_name, new_axis)
-
-        if h_mc_sumw2 is not None:
-            h_mc_sumw2 = _clone_histogram(h_mc_sumw2)
-            try:
-                h_mc_sumw2 = h_mc_sumw2.rebin(axis_name, new_axis)
-            except Exception:
-                logger.warning(
-                    (
-                        "Failed to rebin MC sumw2 histogram for variable '%s'; "
-                        "disabling MC statistical uncertainty for rebinned plots."
-                    ),
-                    var,
-                )
-                h_mc_sumw2 = None
-
-        if any(
-            arr is not None
-            for arr in (err_p_syst, err_m_syst, err_ratio_p_syst, err_ratio_m_syst)
-        ):
-            (
-                err_p_syst,
-                err_m_syst,
-                err_ratio_p_syst,
-                err_ratio_m_syst,
-            ) = _rebin_syst_arrays(
-                err_p_syst,
-                err_m_syst,
-                err_ratio_p_syst,
-                err_ratio_m_syst,
-                orig_nominal_vals,
-                orig_edges,
-                rebin_edges,
-            )
-
-        if rebin_edges.size:
-            if rebin_edges.size >= 2:
-                last_width = rebin_edges[-1] - rebin_edges[-2]
-            else:
-                last_width = rebin_edges[-1]
-            bins_array = np.append(rebin_edges, [rebin_edges[-1] + last_width * 0.3])
-        else:
-            bins_array = rebin_edges
-    else:
-        default_edges = h_data[{"process": sum}].as_hist({}).axes[var].edges
-        default_edges = np.array(default_edges, dtype=float, copy=True)
-        if default_edges.size:
-            if default_edges.size >= 2:
-                last_width = default_edges[-1] - default_edges[-2]
-            else:
-                last_width = default_edges[-1]
-            bins_array = np.append(
-                default_edges, [default_edges[-1] + last_width * 0.3]
-            )
-        else:
-            bins_array = default_edges
-
     default_colors = [
         "tab:blue",
         "darkgreen",
@@ -3601,7 +3401,6 @@ def _rebin_syst_arrays(
         h_mc_sumw2,
         mc_scaled,
         mc_norm_factor,
-        bins_array,
         log_scale=log_scale,
         style=style,
     )