ert 16.0.9__py3-none-any.whl → 19.0.0rc2__py3-none-any.whl

ert/gui/tools/plot/plottery/plots/gaussian_kde.py

@@ -6,7 +6,7 @@ import numpy as np
 import pandas as pd
 from scipy.stats import gaussian_kde
 
-from ert.gui.tools.plot.plot_api import EnsembleObject
+from ert.gui.tools.plot.plot_api import EnsembleObject, PlotApiKeyDefinition
 
 from .plot_tools import ConditionalAxisFormatter, PlotTools
 
@@ -21,6 +21,7 @@ if TYPE_CHECKING:
 class GaussianKDEPlot:
     def __init__(self) -> None:
         self.dimensionality = 1
+        self.requires_observations = False
 
     @staticmethod
     def plot(
@@ -29,10 +30,16 @@ class GaussianKDEPlot:
         ensemble_to_data_map: dict[EnsembleObject, pd.DataFrame],
         observation_data: pd.DataFrame,
         std_dev_images: dict[str, npt.NDArray[np.float32]],
+        key_def: PlotApiKeyDefinition | None = None,
     ) -> None:
         plotGaussianKDE(figure, plot_context, ensemble_to_data_map, observation_data)
 
 
+def _array_is_constant(data: pd.Series | pd.DataFrame) -> bool:
+    array = data.to_numpy()
+    return array.shape[0] == 0 or (array[0] == array).all()
+
+
 def plotGaussianKDE(
     figure: Figure,
     plot_context: PlotContext,
@@ -55,11 +62,14 @@ def plotGaussianKDE(
         if data.empty:
             continue
         data = data[0]
-        if data.nunique() > 1:
+        if not _array_is_constant(data):
             _plotGaussianKDE(
                 axes, config, data, f"{ensemble.experiment_name} : {ensemble.name}"
             )
 
+    if plot_context.log_scale:
+        axes.set_xscale("log")
+
     PlotTools.finalizePlot(
         plot_context, figure, axes, default_x_label="Value", default_y_label="Density"
     )

ert/gui/tools/plot/plottery/plots/histogram.py

@@ -8,7 +8,7 @@ import numpy as np
 import pandas as pd
 from matplotlib.patches import Rectangle
 
-from ert.gui.tools.plot.plot_api import EnsembleObject
+from ert.gui.tools.plot.plot_api import EnsembleObject, PlotApiKeyDefinition
 from ert.shared.status.utils import convert_to_numeric
 
 from .plot_tools import ConditionalAxisFormatter, PlotTools
@@ -24,6 +24,7 @@ if TYPE_CHECKING:
 class HistogramPlot:
     def __init__(self) -> None:
         self.dimensionality = 1
+        self.requires_observations = False
 
     @staticmethod
     def plot(
@@ -32,6 +33,7 @@ class HistogramPlot:
         ensemble_to_data_map: dict[EnsembleObject, pd.DataFrame],
         observation_data: pd.DataFrame,
         std_dev_images: dict[str, npt.NDArray[np.float32]],
+        key_def: PlotApiKeyDefinition | None = None,
     ) -> None:
         plotHistogram(figure, plot_context, ensemble_to_data_map)
 

ert/gui/tools/plot/plottery/plots/misfits.py

@@ -0,0 +1,436 @@
+from __future__ import annotations
+
+from typing import TYPE_CHECKING, Literal, cast
+
+import matplotlib.dates as mdates
+import numpy as np
+import pandas as pd
+import polars as pl
+import seaborn as sns
+from matplotlib import pyplot as plt
+from matplotlib.lines import Line2D
+
+if TYPE_CHECKING:
+    import numpy.typing as npt
+    from matplotlib.figure import Figure
+
+    from ert.gui.tools.plot.plot_api import EnsembleObject, PlotApiKeyDefinition
+    from ert.gui.tools.plot.plottery import PlotContext
+
+
+class MisfitsPlot:
+    """
+    Visualize signed chi-squared misfits between simulated responses and observations.
+
+    Layout:
+        - X-axis: index for gen data, time for summary
+        - Y-axis: signed chi-squared misfits
+        - Glyphs: One boxplot per time step / gendata index
+    """
+
+    def __init__(self) -> None:
+        self.dimensionality = 2
+        self.requires_observations = True
+
+    @staticmethod
+    def _fade_axis_ticklabels(ax: plt.Axes) -> None:
+        """Apply the same alpha to all tick labels on an axis."""
+        for label in (*ax.get_xticklabels(), *ax.get_yticklabels()):
+            label.set_alpha(0.4)
+
+        ax.xaxis.label.set_alpha(0.4)
+        ax.yaxis.label.set_alpha(0.4)
+
+    @staticmethod
+    def _draw_legend(
+        figure: Figure,
+        ensemble_colors: dict[tuple[str, str], str],
+        sorted_ensemble_keys: list[tuple[str, str]],
+    ) -> None:
+        legend_handles = [
+            Line2D(
+                [],
+                [],
+                marker="s",
+                linestyle="None",
+                color=ensemble_colors[key],
+                label=key[0],
+            )
+            for key in sorted_ensemble_keys
+        ]
+        figure.legend(
+            handles=legend_handles,
+            loc="upper center",
+            bbox_to_anchor=(0.5, 0.93),
+            ncol=min(len(legend_handles), 4),
+        )
+
+    @staticmethod
+    def _make_ensemble_colors(
+        sorted_ensemble_keys: list[tuple[str, str]],
+    ) -> dict[tuple[str, str], str]:
+        """
+        Build a consistent color map and figure-level legend
+        used by all misfit plots
+        """
+        colors = plt.rcParams["axes.prop_cycle"].by_key()["color"]
+        return {
+            key: colors[i % len(colors)] for i, key in enumerate(sorted_ensemble_keys)
+        }
+
+    @staticmethod
+    def _compute_misfits_padded_minmax(
+        misfits_df: pl.DataFrame, relative_pad_y_axis: float
+    ) -> tuple[float, float]:
+        y_min, y_max = (
+            cast(float, misfits_df["misfit"].min()),
+            cast(float, misfits_df["misfit"].max()),
+        )
+        abs_pad_y = (y_max - y_min) * relative_pad_y_axis
+        y_min -= abs_pad_y
+        y_max += abs_pad_y
+
+        return y_min, y_max
+
+    @staticmethod
+    def _wide_pandas_to_long_polars_with_misfits(
+        ensemble_to_data_map: dict[tuple[str, str], pd.DataFrame],
+        observation_data: pd.DataFrame,
+        response_type: Literal["summary", "gen_data"],
+    ) -> dict[tuple[str, str], pl.DataFrame]:
+        if response_type == "summary":
+            key_index_with_correct_dtype = pl.col("key_index").str.to_datetime(
+                strict=False
+            )
+        elif response_type == "gen_data":
+            key_index_with_correct_dtype = (
+                pl.col("key_index").cast(pl.Float32).cast(pl.UInt16)
+            )
+        else:
+            raise ValueError(f"Unsupported response_type: {response_type}")
+
+        obs_df = (
+            pl.from_pandas(observation_data.T)
+            .rename({"OBS": "observation", "STD": "error"})
+            .with_columns(pl.col("key_index").cast(pl.String))
+            .with_columns(key_index_with_correct_dtype)
+        )
+
+        return {
+            ens_key: (
+                pl.from_pandas(df, include_index=True)
+                .unpivot(
+                    index=df.index.name,
+                    variable_name="key_index",
+                    value_name="response",
+                )
+                .with_columns(key_index_with_correct_dtype)
+                .join(obs_df, on="key_index", how="inner")
+                .with_columns(
+                    (pl.col("response") - pl.col("observation")).alias("residual")
+                )
+                .with_columns(
+                    (
+                        pl.col("residual").sign()
+                        * (pl.col("residual") / pl.col("error")).pow(2)
+                    ).alias("misfit")
+                )
+                .drop("residual")
+            )
+            for ens_key, df in ensemble_to_data_map.items()
+        }
+
+    def plot(
+        self,
+        figure: Figure,
+        plot_context: PlotContext,
+        ensemble_to_data_map: dict[EnsembleObject, pd.DataFrame],
+        observation_data: pd.DataFrame,
+        std_dev_images: dict[str, npt.NDArray[np.float32]],
+        key_def: PlotApiKeyDefinition | None = None,
+    ) -> None:
+        assert key_def is not None
+        response_type = key_def.metadata["data_origin"]
+        data_with_misfits = self._wide_pandas_to_long_polars_with_misfits(
+            {(eo.name, eo.id): df for eo, df in ensemble_to_data_map.items()},
+            observation_data,
+            response_type,
+        )
+
+        if response_type == "summary":
+            self._plot_summary_misfits_boxplots(
+                figure,
+                data_with_misfits,
+                plot_context,
+            )
+
+        elif response_type == "gen_data":
+            self._plot_gendata_misfits(
+                figure,
+                data_with_misfits,
+                plot_context,
+            )
+
+    def _plot_gendata_misfits(
+        self,
+        figure: Figure,
+        data_with_misfits: dict[tuple[str, str], pl.DataFrame],
+        plot_context: PlotContext,
+    ) -> None:
+        # Only plot ensembles with data (i.e., they pertain to an experiment
+        # with observations, and there are responses towards those in the ens)
+        ensemble_to_misfit_df = {
+            k: v for k, v in data_with_misfits.items() if not v.is_empty()
+        }
+        sorted_ensemble_keys = sorted(ensemble_to_misfit_df.keys())
+
+        all_misfits = pl.concat(
+            [
+                df.with_columns(pl.lit(key).alias("ensemble_key"))
+                for key, df in ensemble_to_misfit_df.items()
+            ]
+        ).select(["realization", "key_index", "misfit", "ensemble_key"])
+
+        distinct_gendata_index = all_misfits["key_index"].unique().sort().to_list()
+        num_gendata_index = len(distinct_gendata_index)
+
+        color_map = self._make_ensemble_colors(sorted_ensemble_keys)
+        self._draw_legend(
+            figure=figure,
+            ensemble_colors=color_map,
+            sorted_ensemble_keys=sorted_ensemble_keys,
+        )
+
+        y_min, y_max = self._compute_misfits_padded_minmax(all_misfits, 0.05)
+
+        # Create subplot grid (2 rows, N columns)
+        axes = figure.subplots(
+            nrows=2, ncols=num_gendata_index, sharex="col", sharey=True
+        )
+        axes = (
+            axes.reshape(2, num_gendata_index)
+            if num_gendata_index > 1
+            else np.array([[axes[0]], [axes[1]]])
+        )
+        axes_top, axes_bottom = axes[0, :], axes[1, :]
+
+        x_positions = np.arange(len(sorted_ensemble_keys))
+        box_width_relative = 0.6
+
+        for col_idx, key_index in enumerate(distinct_gendata_index):
+            ax_top, ax_bottom = axes_top[col_idx], axes_bottom[col_idx]
+
+            for ens_idx, ens_key in enumerate(sorted_ensemble_keys):
+                color = color_map.get(ens_key, "C0")
+
+                # Filter for the specific key and ensemble
+                mis_vals = all_misfits.filter(
+                    (pl.col("key_index") == key_index)
+                    & (pl.col("ensemble_key") == ens_key)
+                )["misfit"].to_numpy()
+
+                if mis_vals.size == 0:
+                    continue
+
+                x_center = x_positions[ens_idx]
+
+                # Top: Boxplot
+                ax_top.boxplot(
+                    mis_vals,
+                    positions=[x_center],
+                    widths=box_width_relative,
+                    patch_artist=True,
+                    showfliers=False,
+                    boxprops={"facecolor": color, "alpha": 0.35},
+                    whiskerprops={"color": color, "alpha": 0.8},
+                    capprops={"color": color, "alpha": 0.8},
+                    medianprops={"color": color, "alpha": 0.8},
+                )
+                ax_top.plot(x_center, np.mean(mis_vals), "o", markersize=4, color=color)
+
+                # Bottom: Strip plot with dynamic marker size
+                num_points = len(mis_vals)
+
+                if num_points >= 200:
+                    marker_size = 2
+                elif num_points >= 100:
+                    marker_size = 3
+                else:
+                    marker_size = 4
+
+                # Use stripplot as a robust alternative for dense data
+                sns.stripplot(
+                    x=[x_center] * num_points,  # Plot all points at the same x-center
+                    y=mis_vals,
+                    ax=ax_bottom,
+                    color=color,
+                    size=marker_size,
+                    alpha=0.35,
+                    jitter=True,  # Explicitly add jitter
+                )
+
+        # Axis/spine styling
+        (n_rows, n_cols) = axes.shape
+        for r_idx in range(n_rows):
+            for c_idx in range(n_cols):
+                ax = axes[r_idx, c_idx]
+                is_first_col = c_idx == 0
+                is_bottom_row = r_idx == (n_rows - 1)
+
+                # Common styling
+                ax.set(ylim=(y_min, y_max))
+                ax.axhline(0.0, color="black", linewidth=0.5, alpha=0.5)
+                ax.grid(True, axis="y", linestyle=":", alpha=0.4)
+                self._fade_axis_ticklabels(ax)
+
+                # Spines
+                ax.spines["top"].set_visible(False)
+                ax.spines["right"].set_visible(False)
+                ax.spines["left"].set_visible(is_first_col)
+                ax.spines["bottom"].set_visible(is_bottom_row)
+
+                # Ticks
+                ax.set_xticks(x_positions, labels=[])
+                if not is_bottom_row:
+                    ax.tick_params(axis="x", which="both", bottom=False)
+                if not is_first_col:
+                    ax.tick_params(axis="y", which="both", left=False)
+
+        for ax, key_val in zip(axes_bottom, distinct_gendata_index, strict=True):
+            ax.set_xlabel(f"index={int(key_val)}", rotation=25, ha="right")
+
+        figure.suptitle(
+            f"{plot_context.key()} (Signed Chi-squared misfits per index)",
+            fontsize=14,
+            y=0.98,
+        )
+        figure.tight_layout(rect=(0.02, 0.02, 0.98, 0.88))
+
+    def _plot_summary_misfits_boxplots(
+        self,
+        figure: Figure,
+        data_with_misfits: dict[tuple[str, str], pl.DataFrame],
+        plot_context: PlotContext,
+    ) -> None:
+        # Calculate shared y-axis limits from all misfits
+        all_misfits = pl.concat(
+            [df.select("misfit") for df in data_with_misfits.values()]
+        )
+        y_min, y_max = self._compute_misfits_padded_minmax(all_misfits, 0.05)
+
+        # Prepare ensemble colors and draw the legend
+        sorted_ensemble_keys = sorted(data_with_misfits.keys())
+        color_map = self._make_ensemble_colors(sorted_ensemble_keys)
+        self._draw_legend(
+            figure=figure,
+            ensemble_colors=color_map,
+            sorted_ensemble_keys=sorted_ensemble_keys,
+        )
+
+        # Create all subplots at once with shared axes
+        n_ens = len(sorted_ensemble_keys)
+        axes = figure.subplots(nrows=n_ens, ncols=1, sharex=True, sharey=True)
+        axes = [axes] if n_ens == 1 else axes.tolist()
+        axes[0].set_ylim(y_min, y_max)
+
+        for ax, ensemble_key in zip(axes, sorted_ensemble_keys, strict=True):
+            df = data_with_misfits[ensemble_key]
+            if df.is_empty():
+                continue
+
+            df = df.select(["key_index", "misfit"]).sort("key_index")
+            times_py = df["key_index"].unique(maintain_order=True).to_list()
+            positions = mdates.date2num(times_py)  # type: ignore[no-untyped-call]
+
+            # Calculate dynamic box width based on time spacing
+            min_dt = np.min(np.diff(positions)) if len(positions) > 1 else 1.0
+
+            # multiplier to downsize outlier sizes etc
+            # (without this, outliers, whiskers etc are sized way
+            # out of proportion when there are many tiny boxplots)
+            many_boxes_factor = min(1, len(times_py) / 50)
+            box_width = min_dt * (0.7 - 0.3 * (1 - many_boxes_factor))
+
+            # One boxplot per time step
+            grouped_misfits = df.group_by("key_index", maintain_order=True).agg(
+                pl.col("misfit")
+            )
+            data_for_boxes = [
+                s.to_numpy() if s.len() > 0 else np.array([np.nan])
+                for s in grouped_misfits["misfit"]
+            ]
+
+            color = color_map.get(ensemble_key)
+
+            # Draw the boxplots with inlined styles
+            bp = ax.boxplot(
+                data_for_boxes,
+                positions=positions,
+                widths=box_width,
+                whis=(5, 95),
+                showfliers=True,
+                manage_ticks=False,
+                patch_artist=True,
+                boxprops={
+                    "facecolor": color,
+                    "alpha": 0.18,
+                    "edgecolor": color,
+                    "linewidth": 0.7,
+                },
+                whiskerprops={"color": color, "alpha": 0.6, "linewidth": 0.7},
+                capprops={"color": color, "alpha": 0.6, "linewidth": 0.7},
+                medianprops={"color": color, "linewidth": 0.6, "alpha": 0.9},
+                flierprops={
+                    "marker": "o",
+                    "markersize": min(6, box_width * (0.4 - (0.2 * many_boxes_factor))),
+                    "alpha": 0.7,
+                    "markeredgewidth": 0.3 + (0.4 * (1 - many_boxes_factor)),
+                    "markeredgecolor": color,
+                    "markerfacecolor": "none",
+                },
+            )
+            plt.setp(bp["fliers"], zorder=1.5)  # Put fliers behind other elements
+
+            # Add ensemble name text and a horizontal line at y=0
+            ax.text(
+                0.01,
+                0.85,
+                f"{ensemble_key[0]}",
+                transform=ax.transAxes,
+                ha="left",
+                va="center",
+                fontsize=9,
+                alpha=0.75,
+            )
+            ax.axhline(0.0, color="black", linewidth=0.8, alpha=0.4, zorder=0)
+
+        # Apply common styling to all axes
+        for ax in axes:
+            ax.spines["top"].set_visible(False)
+            ax.spines["right"].set_visible(False)
+            ax.spines["left"].set_visible(True)
+            ax.tick_params(axis="y", labelsize=10, width=0.5, length=3)
+            ax.grid(True, axis="y", linestyle=":", linewidth=0.5, alpha=0.75)
+            self._fade_axis_ticklabels(ax)
+
+        # Hide the x-axis on all but the last plot
+        for ax in axes[:-1]:
+            ax.spines["bottom"].set_visible(False)
+            ax.tick_params(axis="x", which="both", bottom=False, labelbottom=False)
+
+        # Style the x-axis only on the last plot
+        bottom_ax = axes[-1]
+        bottom_ax.spines["bottom"].set_visible(True)
+        bottom_ax.xaxis.set_major_locator(mdates.AutoDateLocator())  # type: ignore[no-untyped-call]
+        bottom_ax.xaxis.set_major_formatter(mdates.DateFormatter("%Y-%m-%d"))  # type: ignore[no-untyped-call]
+
+        bottom_ax.tick_params(axis="x", labelsize=8, width=0.5, length=3)
+        plt.setp(bottom_ax.get_xticklabels(), rotation=25, ha="right")
+
+        figure.suptitle(
+            f"{plot_context.key()} (Signed Chi-squared misfits over time)",
+            fontsize=14,
+            y=0.97,
+            alpha=0.9,
+        )
+        figure.tight_layout(rect=(0.02, 0.02, 0.98, 0.92))

ert/gui/tools/plot/plottery/plots/observations.py

@@ -27,7 +27,10 @@ def plotObservations(
 
 
 def _plotObservations(
-    axes: Axes, plot_config: PlotConfig, data: DataFrame, value_column: str
+    axes: Axes,
+    plot_config: PlotConfig,
+    data: DataFrame,
+    value_column: str,
 ) -> None:
     """
     Observations are always plotted on top. z-order set to 1000
@@ -46,11 +49,22 @@ def _plotObservations(
     # line style set to 'off' toggles errorbar visibility
     if not style.line_style:
         style.width = 0
+    if plot_config.depth_y_axis:
+        errorbar_data = {
+            "x": data.loc["OBS"].to_numpy(),
+            "y": data.loc["key_index"].to_numpy(),
+            "xerr": data.loc["STD"].to_numpy(),
+        }
+        axes.yaxis.set_inverted(True)
+    else:
+        errorbar_data = {
+            "x": data.loc["key_index"].to_numpy(),
+            "y": data.loc["OBS"].to_numpy(),
+            "yerr": data.loc["STD"].to_numpy(),
+        }
 
     axes.errorbar(
-        x=data.loc["key_index"].values,
-        y=data.loc["OBS"].values,
-        yerr=data.loc["STD"].values,
+        **errorbar_data,
         fmt=style.line_style,
         ecolor=style.color,
         color=style.color,