DoubleSummaryStatistics.java

/*
 * Copyright (c) 2012, 2017, Oracle and/or its affiliates. All rights reserved.
 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
 *
 * This code is free software; you can redistribute it and/or modify it
 * under the terms of the GNU General Public License version 2 only, as
 * published by the Free Software Foundation.  Oracle designates this
 * particular file as subject to the "Classpath" exception as provided
 * by Oracle in the LICENSE file that accompanied this code.
 *
 * This code is distributed in the hope that it will be useful, but WITHOUT
 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
 * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
 * version 2 for more details (a copy is included in the LICENSE file that
 * accompanied this code).
 *
 * You should have received a copy of the GNU General Public License version
 * 2 along with this work; if not, write to the Free Software Foundation,
 * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
 *
 * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
 * or visit www.oracle.com if you need additional information or have any
 * questions.
 */
package java.util;

import java.util.function.DoubleConsumer;
import java.util.stream.Collector;
import java.util.stream.DoubleStream;

/**
 * A state object for collecting statistics such as count, min, max, sum, and
 * average.
 *
 * <p>This class is designed to work with (though does not require)
 * {@linkplain java.util.stream streams}. For example, you can compute
 * summary statistics on a stream of doubles with:
 * <pre> {@code
 * DoubleSummaryStatistics stats = doubleStream.collect(DoubleSummaryStatistics::new,
 *                                                      DoubleSummaryStatistics::accept,
 *                                                      DoubleSummaryStatistics::combine);
 * }</pre>
 *
 * <p>{@code DoubleSummaryStatistics} can be used as a
 * {@linkplain java.util.stream.Stream#collect(Collector) reduction}
 * target for a {@linkplain java.util.stream.Stream stream}. For example:
 *
 * <pre> {@code
 * DoubleSummaryStatistics stats = people.stream()
 *     .collect(Collectors.summarizingDouble(Person::getWeight));
 * }</pre>
 *
 * This computes, in a single pass, the count of people, as well as the minimum,
 * maximum, sum, and average of their weights.
 *
 * @implNote This implementation is not thread safe. However, it is safe to use
 * {@link java.util.stream.Collectors#summarizingDouble(java.util.function.ToDoubleFunction)
 * Collectors.summarizingDouble()} on a parallel stream, because the parallel
 * implementation of {@link java.util.stream.Stream#collect Stream.collect()}
 * provides the necessary partitioning, isolation, and merging of results for
 * safe and efficient parallel execution.
 * @since 1.8
 */

// 对double类型的元素统计相关信息：计数、求和、均值、最小值、最大值
public class DoubleSummaryStatistics implements DoubleConsumer {
    private double min = Double.POSITIVE_INFINITY;  // 最小值
    private double max = Double.NEGATIVE_INFINITY;  // 最大值
    
    private long count; // 计数
    private double sum; // 求和，考虑了误差
    
    // Low order bits of sum
    private double sumCompensation; // 上一次计算产生的误差
    // Used to compute right sum for non-finite inputs
    private double simpleSum;       // 简单求和，不考虑误差，只是累加
    
    /**
     * Constructs an empty instance with zero count, zero sum,
     * {@code Double.POSITIVE_INFINITY} min, {@code Double.NEGATIVE_INFINITY}
     * max and zero average.
     */
    public DoubleSummaryStatistics() {
    }
    
    /**
     * Constructs a non-empty instance with the specified {@code count},
     * {@code min}, {@code max}, and {@code sum}.
     *
     * <p>If {@code count} is zero then the remaining arguments are ignored and
     * an empty instance is constructed.
     *
     * <p>If the arguments are inconsistent then an {@code IllegalArgumentException}
     * is thrown.  The necessary consistent argument conditions are:
     * <ul>
     * <li>{@code count >= 0}</li>
     * <li>{@code (min <= max && !isNaN(sum)) || (isNaN(min) && isNaN(max) && isNaN(sum))}</li>
     * </ul>
     *
     * @param count the count of values
     * @param min   the minimum value
     * @param max   the maximum value
     * @param sum   the sum of all values
     *
     * @throws IllegalArgumentException if the arguments are inconsistent
     * @apiNote The enforcement of argument correctness means that the retrieved set of
     * recorded values obtained from a {@code DoubleSummaryStatistics} source
     * instance may not be a legal set of arguments for this constructor due to
     * arithmetic overflow of the source's recorded count of values.
     * The consistent argument conditions are not sufficient to prevent the
     * creation of an internally inconsistent instance.  An example of such a
     * state would be an instance with: {@code count} = 2, {@code min} = 1,
     * {@code max} = 2, and {@code sum} = 0.
     * @since 10
     */
    public DoubleSummaryStatistics(long count, double min, double max, double sum) throws IllegalArgumentException {
        if(count<0L) {
            throw new IllegalArgumentException("Negative count value");
        } else if(count>0L) {
            if(min>max)
                throw new IllegalArgumentException("Minimum greater than maximum");
            
            // All NaN or non NaN
            var ncount = DoubleStream.of(min, max, sum).filter(Double::isNaN).count();
            if(ncount>0 && ncount<3)
                throw new IllegalArgumentException("Some, not all, of the minimum, maximum, or sum is NaN");
            
            this.count = count;
            this.sum = sum;
            this.simpleSum = sum;
            this.sumCompensation = 0.0d;
            this.min = min;
            this.max = max;
        }
        // Use default field values if count == 0
    }
    
    /**
     * Records another value into the summary information.
     *
     * @param value the input value
     */
    // 每遇到一个新的value，需要更新计数、计数、最小值、最大值的数据
    @Override
    public void accept(double value) {
        ++count;
        simpleSum += value;
        sumWithCompensation(value);
        min = Math.min(min, value);
        max = Math.max(max, value);
    }
    
    /**
     * Combines the state of another {@code DoubleSummaryStatistics} into this
     * one.
     *
     * @param other another {@code DoubleSummaryStatistics}
     *
     * @throws NullPointerException if {@code other} is null
     */
    // 合并两个数据源的统计信息（计数与求和相加，最小值取最小的，最大值取最大的）
    public void combine(DoubleSummaryStatistics other) {
        count += other.count;
        simpleSum += other.simpleSum;
        sumWithCompensation(other.sum);
        sumWithCompensation(other.sumCompensation);
        min = Math.min(min, other.min);
        max = Math.max(max, other.max);
    }
    
    /**
     * Return the count of values recorded.
     *
     * @return the count of values
     */
    // 计数
    public final long getCount() {
        return count;
    }
    
    /**
     * Returns the sum of values recorded, or zero if no values have been
     * recorded.
     *
     * <p> The value of a floating-point sum is a function both of the
     * input values as well as the order of addition operations. The
     * order of addition operations of this method is intentionally
     * not defined to allow for implementation flexibility to improve
     * the speed and accuracy of the computed result.
     *
     * In particular, this method may be implemented using compensated
     * summation or other technique to reduce the error bound in the
     * numerical sum compared to a simple summation of {@code double}
     * values.
     *
     * Because of the unspecified order of operations and the
     * possibility of using differing summation schemes, the output of
     * this method may vary on the same input values.
     *
     * <p>Various conditions can result in a non-finite sum being
     * computed. This can occur even if the all the recorded values
     * being summed are finite. If any recorded value is non-finite,
     * the sum will be non-finite:
     *
     * <ul>
     *
     * <li>If any recorded value is a NaN, then the final sum will be
     * NaN.
     *
     * <li>If the recorded values contain one or more infinities, the
     * sum will be infinite or NaN.
     *
     * <ul>
     *
     * <li>If the recorded values contain infinities of opposite sign,
     * the sum will be NaN.
     *
     * <li>If the recorded values contain infinities of one sign and
     * an intermediate sum overflows to an infinity of the opposite
     * sign, the sum may be NaN.
     *
     * </ul>
     *
     * </ul>
     *
     * It is possible for intermediate sums of finite values to
     * overflow into opposite-signed infinities; if that occurs, the
     * final sum will be NaN even if the recorded values are all
     * finite.
     *
     * If all the recorded values are zero, the sign of zero is
     * <em>not</em> guaranteed to be preserved in the final sum.
     *
     * @return the sum of values, or zero if none
     *
     * @apiNote Values sorted by increasing absolute magnitude tend to yield
     * more accurate results.
     */
    // 求和
    public final double getSum() {
        // Better error bounds to add both terms as the final sum
        double tmp = sum + sumCompensation; // 比较精确的值
        
        if(Double.isNaN(tmp) && Double.isInfinite(simpleSum)) {
            /*
             * If the compensated sum is spuriously NaN from accumulating one or more same-signed infinite values,
             * return the correctly-signed infinity stored in simpleSum.
             */
            return simpleSum;
        } else {
            return tmp;
        }
    }
    
    /**
     * Returns the minimum recorded value, {@code Double.NaN} if any recorded
     * value was NaN or {@code Double.POSITIVE_INFINITY} if no values were
     * recorded. Unlike the numerical comparison operators, this method
     * considers negative zero to be strictly smaller than positive zero.
     *
     * @return the minimum recorded value, {@code Double.NaN} if any recorded
     * value was NaN or {@code Double.POSITIVE_INFINITY} if no values were
     * recorded
     */
    // 最小值
    public final double getMin() {
        return min;
    }
    
    /**
     * Returns the maximum recorded value, {@code Double.NaN} if any recorded
     * value was NaN or {@code Double.NEGATIVE_INFINITY} if no values were
     * recorded. Unlike the numerical comparison operators, this method
     * considers negative zero to be strictly smaller than positive zero.
     *
     * @return the maximum recorded value, {@code Double.NaN} if any recorded
     * value was NaN or {@code Double.NEGATIVE_INFINITY} if no values were
     * recorded
     */
    // 最大值
    public final double getMax() {
        return max;
    }
    
    /**
     * Returns the arithmetic mean of values recorded, or zero if no
     * values have been recorded.
     *
     * <p> The computed average can vary numerically and have the
     * special case behavior as computing the sum; see {@link #getSum}
     * for details.
     *
     * @return the arithmetic mean of values, or zero if none
     *
     * @apiNote Values sorted by increasing absolute magnitude tend to yield
     * more accurate results.
     */
    // 最后计算平均值
    public final double getAverage() {
        return getCount()>0 ? getSum() / getCount() : 0.0d;
    }
    
    /**
     * Returns a non-empty string representation of this object suitable for
     * debugging. The exact presentation format is unspecified and may vary
     * between implementations and versions.
     */
    @Override
    public String toString() {
        return String.format("%s{count=%d, sum=%f, min=%f, average=%f, max=%f}", this.getClass().getSimpleName(), getCount(), getSum(), getMin(), getAverage(), getMax());
    }
    
    /**
     * Incorporate a new double value using Kahan summation /
     * compensated summation.
     */
    // Kahan求和精度补偿算法，弥补浮点运算中的精度损失
    private void sumWithCompensation(double value) {
        double tmp = value - sumCompensation;   // 当前的值补上上一次的误差
        // Little wolf of rounding error
        double velvel = sum + tmp;  // 本次的求和结果，已补上上次的误差
        sumCompensation = (velvel - sum) - tmp; // 本次运算产生的新误差
        sum = velvel;   // 记下本次的求和结果
    }
}