feat(nmNum): 增加误差直方图窗口

- 主界面显示误差柱状图与累计概率曲线，支持调整分箱数。
- 调整计数轴范围和刻度，概率刻度间隔为 10%。
- 支持关闭重开，新模型生成后清除旧图。

Include/nmNum/nmSubWnd/nmSubWndMain.h

@@ -168,6 +168,8 @@ class NM_SUB_WND_EXPORT nmSubWndMain : public iSubWndBaseFit {
     void onSensitivityTaskDestroyed();
     /** @brief 在框架已有的敏感性子窗口叠加显示本分析的双对数结果。 */
     void onShowSensitiveResults();
+    /** @brief 刷新主界面的误差直方图，修改分箱数时复用当前批次结果。 */
+    void updateSensitiveHistogramPlot();
     /** @brief 当前分析的建网提交结束后，恢复待启动或待捕获的求解流程。 */
     void onPebiGridGenerationFinished(quint64 nGridInputRevision,
                                       bool bSucceeded);

Src/nmNum/nmSubWnd/nmSubWndMain.cpp

@@ -13,6 +13,7 @@
 
 #include "ZxPlot.h"
 #include "ZxObjCurve.h"
+#include "ZxObjCurveBar.h"
 #include "ZxSubTitle.h"
 #include "tCurvePlotScene.h"
 #include "tCurvePlotView.h"
@@ -74,6 +75,9 @@
 #include <QSet>
 #include <QHash>
 #include <QPushButton>
+#include <QSpinBox>
+#include <QLabel>
+#include <cmath>
 #include <QMdiSubWindow>
 #include "ZxMdiArea.h"
 
@@ -108,6 +112,25 @@ namespace
 QPointer<nmCalculationDllPebiSolverTask> s_pRunningSolverTask;
 QPointer<nmSubWndMain> s_pPendingSolverWindow;
 
+class nmSensitivityHistogramBars : public ZxObjCurveBar
+{
+public:
+    explicit nmSensitivityHistogramBars(double binWidth) : m_binWidth(binWidth) {}
+
+protected:
+    void drawSubBars(QPainter* painter, QVector<QPointF>& points) override
+    {
+        // 框架柱宽使用绘图坐标，按分箱宽度换算，缩放和窗口调整后仍对应同一区间。
+        const ZxSubAxisX* axis = getAxisX();
+        const double left = axis->getRangeMin();
+        m_dBarWidth = 0.95 * qAbs(axis->getPosForValue(left + m_binWidth) - axis->getPosForValue(left));
+        ZxObjCurveBar::drawSubBars(painter, points);
+    }
+
+private:
+    double m_binWidth;
+};
+
 // 历史窗口仍是单流量曲线结构，因此显示与双对数、半对数后处理一致的制度参考相。
 const QVector<double>* resultReferenceRatesOf(
         const nmPebiResultWellSnapshot* pWell)
@@ -1875,6 +1898,7 @@ void nmSubWndMain::onShowSensitiveResults()
     iSubWndFitting* fitting = m_pResultRefreshFitting.data();
     if(!manager || !fitting || !manager->getSensitiveData()) return;
     updateSensitiveHistoryPlot();
+    updateSensitiveHistogramPlot();
     QSharedPointer<nmDataSensitive::BatchResult> result = manager->getSensitiveData()->getResults();
     iGuiPlot* plot = fitting->getPlotByTag(s_Tag_ChartSensitivity);
     if(result.isNull()) {
@@ -2089,6 +2113,152 @@ void nmSubWndMain::updateSensitiveHistoryPlot()
     if(created || restore) area->relayoutSubWnds();
 }
 
+void nmSubWndMain::updateSensitiveHistogramPlot()
+{
+    iSubWndFitting* fitting = m_pResultRefreshFitting.data();
+    nmDataAnalyzeManager* manager = m_pResultRefreshDataManager.data();
+    if(!fitting || !fitting->m_pMdiArea || !manager || !manager->getSensitiveData()) return;
+    ZxMdiArea* area = fitting->m_pMdiArea;
+    QMdiSubWindow* window = area->findChild<QMdiSubWindow*>("nmSensitivityHistogram");
+    const QSharedPointer<nmDataSensitive::BatchResult> batch = manager->getSensitiveData()->getResults();
+    if(batch.isNull() || batch->m_calcType != nmDataSensitive::CALC_MONTE_CARLO) {
+        // 新模型或确定性批次不保留上一批 Monte-Carlo 的统计图。
+        if(window) {
+            area->removeSubWindow(window);
+            delete window;
+        }
+        return;
+    }
+    const bool created = window == nullptr;
+    if(created) {
+        QWidget* page = new QWidget;
+        QVBoxLayout* layout = new QVBoxLayout(page);
+        layout->setContentsMargins(0, 0, 0, 0);
+        QHBoxLayout* options = new QHBoxLayout;
+        options->addWidget(new QLabel(tr("Bins:"), page));
+        QSpinBox* bins = new QSpinBox(page);
+        bins->setObjectName("nmHistogramBins");
+        bins->setRange(1, 1000);
+        bins->setValue(20);
+        options->addWidget(bins);
+        QLabel* count = new QLabel(page);
+        count->setObjectName("nmHistogramCount");
+        options->addWidget(count);
+        options->addStretch();
+        layout->addLayout(options);
+        iGuiPlot* plot = new iGuiPlot(true, page);
+        plot->initUI(tr("Sensitivity histogram plot"));
+        plot->setDropEnabled(false);
+        plot->setAxisX(tr("Goodness of fit"), "MPa");
+        plot->setAxisY(tr("Count"));
+        plot->setAxisY(tr("Cumulative probability"), "%", true);
+        plot->m_pPlot->setBKColor(Qt::white);
+        for(int i = 0; i < 3; ++i) {
+            plot->setAxisXYLog(false, i);
+            plot->getAxisOf(i)->setVisible(true);
+        }
+        layout->addWidget(plot);
+        window = area->addSubWindow(page);
+        window->setObjectName("nmSensitivityHistogram");
+        window->setWindowTitle(tr("Sensitivity histogram plot"));
+        window->setAttribute(Qt::WA_DeleteOnClose, false);
+        connect(bins, SIGNAL(valueChanged(int)), this, SLOT(updateSensitiveHistogramPlot()));
+    }
+    iGuiPlot* plot = window->findChild<iGuiPlot*>();
+    const int binCount = window->findChild<QSpinBox*>("nmHistogramBins")->value();
+    QVector<double> errors;
+    for(size_t i = 0; i < batch->m_runs.size(); ++i) {
+        const nmDataSensitive::RunResult& run = batch->m_runs[i];
+        // 显示勾选不改变统计总体，失败、取消或无有效误差的运行不参与统计。
+        if(run.m_bSucceeded && !run.m_bCancelled && qIsFinite(run.m_dPressureRmse) && run.m_dPressureRmse >= 0.0)
+            errors.append(run.m_dPressureRmse);
+    }
+    window->findChild<QLabel*>("nmHistogramCount")->setText(errors.isEmpty()
+                ? tr("No valid goodness-of-fit results") : tr("Valid runs: %1").arg(errors.size()));
+    double minimum = 0.0, maximum = 1.0;
+    if(!errors.isEmpty()) {
+        minimum = maximum = errors.first();
+        for(int i = 1; i < errors.size(); ++i) {
+            minimum = qMin(minimum, errors[i]);
+            maximum = qMax(maximum, errors[i]);
+        }
+        // 单个结果或全部相同的误差也需要非零横轴范围。
+        if(minimum == maximum) {
+            const double padding = qMax(1e-6, maximum * 0.05);
+            minimum = qMax(0.0, minimum - padding);
+            maximum += padding;
+        }
+    }
+    const double width = (maximum - minimum) / binCount;
+    QVector<int> counts(binCount, 0);
+    for(int i = 0; i < errors.size(); ++i) {
+        // 等宽分箱，左闭右开，最大误差归入最后一箱。
+        const int index = qMin(binCount - 1, int((errors[i] - minimum) / width));
+        ++counts[index];
+    }
+    for(int i = plot->m_pPlot->getObjCount() - 1; i >= 0; --i)
+        if(qobject_cast<ZxObjCurve*>(plot->m_pPlot->getObjByIndex(i))) plot->m_pPlot->removeObjByIndex(i);
+    plot->clearUndoStack();
+    QVector<iCurveDesc> curves;
+    int largestCount = 0;
+    if(!errors.isEmpty()) {
+        QVector<QPointF> bars;
+        iCurveDesc cumulative;
+        cumulative.m_sCurveName = tr("Cumulative probability");
+        cumulative.m_oCurveType = POT_CurveDiscrete;
+        cumulative.m_bUseY2 = true;
+        cumulative.m_bLineVisible = true;
+        cumulative.m_bPointVisible = true;
+        cumulative.m_oPen.setColor(Qt::red);
+        cumulative.m_oDot = ZxDot(DTS_Circle, Qt::red, 0.6, true);
+        int total = 0;
+        for(int i = 0; i < binCount; ++i) {
+            const double center = minimum + (i + 0.5) * width;
+            bars.append(QPointF(center, counts[i]));
+            largestCount = qMax(largestCount, counts[i]);
+            total += counts[i];
+            cumulative.m_vecX.append(center);
+            cumulative.m_vecY.append(100.0 * total / errors.size());
+        }
+        // 普通曲线入口不支持柱状图，直接使用框架柱状图元，累计概率仍走已有曲线入口。
+        nmSensitivityHistogramBars* histogram = new nmSensitivityHistogramBars(width);
+        histogram->init(tr("Count"), plot->m_pPlot->getMainAxisX(), plot->m_pPlot->getMainAxisY());
+        plot->m_pPlot->addOneObj(histogram);
+        histogram->setAllValues(bars);
+        histogram->setCurvePty(QPen(Qt::black), ZxDot(), true, false);
+        histogram->setBarFill(true);
+        histogram->setBarFillClr(QColor(85, 107, 47));
+        histogram->setShow3D(false);
+        histogram->setPinValue(false);
+        curves.append(cumulative);
+    }
+    plot->setCurveData(curves);
+    plot->runUpdate();
+    plot->setReadOnly(true);
+    VecDouble ticksX, ticksY, ticksY2;
+    for(int i = 0; i <= 5; ++i) {
+        ticksX.append(minimum + i * (maximum - minimum) / 5.0);
+    }
+    // 累计概率轴每隔 10% 标注，100% 以上保留顶部留白。
+    for(int i = 0; i <= 10; ++i) ticksY2.append(i * 10.0);
+    // 小样本时放大计数轴范围，使柱子保持较低；样本较多时自动扩展并留出顶部空间。
+    const double countRange = qMax(16.0, largestCount * 1.2);
+    const double step = std::ceil(countRange / 8.0);
+    const double top = std::ceil(countRange / step) * step;
+    // 保持原轴范围，将主刻度加密一倍，计数刻度仍使用整数。
+    const double tickStep = qMax(1.0, std::ceil(step / 2.0));
+    for(double tick = 0.0; tick < top; tick += tickStep) ticksY.append(tick);
+    ticksY.append(top);
+    plot->setAxisScaleX(minimum, maximum, ticksX);
+    plot->setAxisScaleY(0.0, top, ticksY);
+    plot->setAxisScaleY2(0.0, 110.0, ticksY2);
+    const bool restore = window->isHidden();
+    plot->show();
+    window->widget()->show();
+    window->show();
+    if(created || restore) area->relayoutSubWnds();
+}
+
 void nmSubWndMain::startSolverTask()
 {
     if(!m_bSolverStartPending) {