我正在为fibonacci算法做一个小项目。
我使用下面的方法来计算算法。注意,elapsedTime()返回一个double。
public static void fibonacciSequence(long n1, long n2) {
t0 = stopwatch.elapsedTime();
System.out.print("index: " + index + " -> " + n1 + "\t");
t1 = stopwatch.elapsedTime();
lapTime = (1000 * t1 - 1000 * t0) / 1000;
StdOut.println(" (" + lapTime + "\t " + t1 + ")");
if (index == stoppingPoint) {
return;
}
index++;
fibonacciSequence(n2, n1 + n2);
}现在,不要过多地关注算法本身--它是固定的。我只是不明白lapTime的公式。为什么不能
lapTime = t1-t0; 发布于 2014-09-09 18:03:25
这属于来自Java普林斯顿Stdlib的Java普林斯顿Stdlib类。这个类是用来度量算法执行时间的。实现如下(删除注释和此类数据):
public class Stopwatch {
private final long start;
public Stopwatch() {
start = System.currentTimeMillis();
}
public double elapsedTime() {
long now = System.currentTimeMillis();
return (now - start) / 1000.0;
}
}由于它使用的是System.currentTimeMillis(),所以它的工作时间为毫秒。但是elapsedTime方法已经将其转换为秒,但作为一个double。所以你现在的公式是:
lapTime = (1000 * t1 - 1000 * t0) / 1000;确保将数据转换为纯double操作。因此,公式可以重写为:
lapTime = t1 - t0;没有问题。
注意,仍然是,这不是度量代码执行时间的正确方法。您应该使用System.nanoTime()代替。
更多信息:
为了更深入地理解这些操作之间是否没有区别,让我们创建一个基本测试:
public class FormulaTest {
static double formula1(double t0, double t1) {
return (1000 * t1 - 1000 * t0) / 1000;
}
static double formula2(double t0, double t1) {
return t1 - t0;
}
static void printResults(double t0, double t1) {
System.out.println("t0: " + t0);
System.out.println("t1: " + t1);
System.out.println("Formula1: " + formula1(t0, t1));
System.out.println("Formula2: " + formula1(t0, t1));
System.out.println("---------------------------------------------------");
}
public static void main(String[] args) throws java.lang.Exception {
// your code goes here
printResults(0, 10);
printResults(System.currentTimeMillis(), System.currentTimeMillis());
printResults(System.currentTimeMillis(), System.currentTimeMillis() + 142);
printResults(1.7976931348623157e+300 - 5000, 1.7976931348623157e+300);
printResults(
109999999999999999999999999999999999999999999999999999999999999999999999.0,
119999999999999999999999999999999999999999999999999999999999999999999999.0);
printResults(
10.9999999999999999999999999999999999999999999999999999999999999999999999,
11.9999999999999999999999999999999999999999999999999999999999999999999999);
printResults(
823145321462149234.651985149616914621346234923149621346921394613293423951932415934159213226314,
844329146321496321.532159341563149513495139159341593415793415431951349513891585443951391593151);
printResults(
82314532.1462149234651985149616914621346234923149621346921394613293423951932415934159213226314,
84432914.6321496321532159341563149513495139159341593415793415431951349513891585443951391593151);
printResults(1.7976931348623157e+307, 4.9e-323);
printResults(Double.MAX_VALUE, Double.MAX_VALUE);
printResults(Double.MIN_VALUE - 10, Double.MIN_VALUE);
}
}输出:
t0: 0.0
t1: 10.0
Formula1: 10.0
Formula2: 10.0
---------------------------------------------------
t0: 1.410290897577E12
t1: 1.410290897577E12
Formula1: 0.0
Formula2: 0.0
---------------------------------------------------
t0: 1.410290897577E12
t1: 1.410290897719E12
Formula1: 142.0
Formula2: 142.0
---------------------------------------------------
t0: 1.7976931348623156E300
t1: 1.7976931348623156E300
Formula1: 0.0
Formula2: 0.0
---------------------------------------------------
t0: 1.1E71
t1: 1.2E71
Formula1: 9.999999999999985E69
Formula2: 9.999999999999985E69
---------------------------------------------------
t0: 11.0
t1: 12.0
Formula1: 1.0
Formula2: 1.0
---------------------------------------------------
t0: 8.2314532146214925E17
t1: 8.4432914632149632E17
Formula1: 2.1183824859347024E16
Formula2: 2.1183824859347024E16
---------------------------------------------------
t0: 8.231453214621492E7
t1: 8.443291463214964E7
Formula1: 2118382.4859347227
Formula2: 2118382.4859347227
---------------------------------------------------
t0: 1.7976931348623158E307
t1: 4.9E-323
Formula1: -Infinity
Formula2: -Infinity
---------------------------------------------------
t0: 1.7976931348623157E308
t1: 1.7976931348623157E308
Formula1: NaN
Formula2: NaN
---------------------------------------------------
t0: -10.0
t1: 4.9E-324
Formula1: 10.0
Formula2: 10.0
---------------------------------------------------即使在边缘情况下,Double.MAX_VALUE和Double.MIN_VALUE公式的结果也是不同的。即使结果是-Infinity或NaN (不是一个数字)。
简而言之:使用最后一个公式:
lapTime = t1 - t0;发布于 2014-09-09 18:31:32
发布于 2014-09-09 17:58:17
应该是的。
(a * x - a * y) / a == (x - y) * a / a == x - y就你而言:
(1000 * t1 - 1000 * t0) / 1000 == (t1 - t0) * 1000 / 1000 == t1 - t0因此,编写lapTime = t1-t0是个好主意,因为它使代码更容易理解。
此外,您还避免了long溢出的风险,不再乘以1000。
https://stackoverflow.com/questions/25750945
复制相似问题