JAVA中控制double和float的精度

本篇介绍了在JAVA中如何对double或者float的浮点数进行精度计算，在JAVA中提供了多种参数来实现精度的不同控制方式。具体例子如下： 

转自http://stephen830.iteye.com/blog/260776

/*
 * Created on 2005-6-5
 * Author stephen
 * Email zhoujianqiang AT gmail DOT com
 * CopyRight(C)2005-2008 , All rights reserved.
 */
package com.soft4j.utility;

import java.math.BigDecimal;

/**
 * 与小数位精度(四舍五入等)相关的一些常用工具方法.
 *
 * float/double的精度取值方式分为以下几种: <br>
 * java.math.BigDecimal.ROUND_UP <br>
 * java.math.BigDecimal.ROUND_DOWN <br>
 * java.math.BigDecimal.ROUND_CEILING <br>
 * java.math.BigDecimal.ROUND_FLOOR <br>
 * java.math.BigDecimal.ROUND_HALF_UP<br>
 * java.math.BigDecimal.ROUND_HALF_DOWN <br>
 * java.math.BigDecimal.ROUND_HALF_EVEN <br>
 *
 * @author stephen
 * @version 1.0.0
 */
public final class RoundTool {

    /**
     * 对double数据进行取精度.
     * <p>
     * For example: <br>
     * double value = 100.345678; <br>
     * double ret = round(value,4,BigDecimal.ROUND_HALF_UP); <br>
     * ret为100.3457 <br>
     *
     * @param value
     *            double数据.
     * @param scale
     *            精度位数(保留的小数位数).
     * @param roundingMode
     *            精度取值方式.
     * @return 精度计算后的数据.
     */
    public static double round(double value, int scale, int roundingMode) {
        BigDecimal bd = new BigDecimal(value);
        bd = bd.setScale(scale, roundingMode);
        double d = bd.doubleValue();
        bd = null;
        return d;
    }

    /**
     * 测试用的main方法.
     *
     * @param argc
     *            运行参数.
     */
    public static void main(String[] argc) {
        //下面都以保留2位小数为例

        //ROUND_UP
        //只要第2位后面存在大于0的小数，则第2位就+1
        System.out.println(round(12.3401,2,BigDecimal.ROUND_UP));//12.35
        System.out.println(round(-12.3401,2,BigDecimal.ROUND_UP));//-12.35
        //ROUND_DOWN
        //与ROUND_UP相反
        //直接舍弃第2位后面的所有小数
        System.out.println(round(12.349,2,BigDecimal.ROUND_DOWN));//12.34
        System.out.println(round(-12.349,2,BigDecimal.ROUND_DOWN));//-12.34
        //ROUND_CEILING
        //如果数字>0 则和ROUND_UP作用一样
        //如果数字<0 则和ROUND_DOWN作用一样
        System.out.println(round(12.3401,2,BigDecimal.ROUND_CEILING));//12.35
        System.out.println(round(-12.349,2,BigDecimal.ROUND_CEILING));//-12.34
        //ROUND_FLOOR
        //如果数字>0 则和ROUND_DOWN作用一样
        //如果数字<0 则和ROUND_UP作用一样
        System.out.println(round(12.349,2,BigDecimal.ROUND_FLOOR));//12.34
        System.out.println(round(-12.3401,2,BigDecimal.ROUND_FLOOR));//-12.35
        //ROUND_HALF_UP [这种方法最常用]
        //如果第3位数字>=5,则第2位数字+1
        //备注:只看第3位数字的值,不会考虑第3位之后的小数的
        System.out.println(round(12.345,2,BigDecimal.ROUND_HALF_UP));//12.35
        System.out.println(round(12.3449,2,BigDecimal.ROUND_HALF_UP));//12.34
        System.out.println(round(-12.345,2,BigDecimal.ROUND_HALF_UP));//-12.35
        System.out.println(round(-12.3449,2,BigDecimal.ROUND_HALF_UP));//-12.34
        //ROUND_HALF_DOWN
        //如果第3位数字>=5,则做ROUND_UP
        //如果第3位数字<5,则做ROUND_DOWN
        System.out.println(round(12.345,2,BigDecimal.ROUND_HALF_DOWN));//12.35
        System.out.println(round(12.3449,2,BigDecimal.ROUND_HALF_DOWN));//12.34
        System.out.println(round(-12.345,2,BigDecimal.ROUND_HALF_DOWN));//-12.35
        System.out.println(round(-12.3449,2,BigDecimal.ROUND_HALF_DOWN));//-12.34
        //ROUND_HALF_EVEN
        //如果第3位是偶数,则做ROUND_HALF_DOWN
        //如果第3位是奇数,则做ROUND_HALF_UP
        System.out.println(round(12.346,2,BigDecimal.ROUND_HALF_EVEN));//12.35
        System.out.println(round(12.345,2,BigDecimal.ROUND_HALF_EVEN));//12.35
    }
}