我有一个 Scala TreeMap 可以自动对键进行排序。我想知道的是,是否有比以下示例更高效的在 map 中查找第 N 个键的方法:
treeMap.take(N).lastKey
谢谢,
布鲁斯
编辑:
我使用以下代码创建了一个小测试:
class Test {
var treeMap = new scala.collection.immutable.TreeMap[Double,String]()
val numberOfEntries = 1000
(0 until numberOfEntries) map { i => {treeMap += {i.toDouble -> i.toString}}}
val iterations = 2000
var N = 1
while(N < numberOfEntries) {
// my original version
var i = 0
val start1 = System.nanoTime()
while(i < iterations) {
i += 1
val v = treeMap.take(N).lastKey
}
val end1 = System.nanoTime()
val elapsed1 = end1 - start1
// Daniel's suggestion
i = 0
val start2 = System.nanoTime()
while(i < iterations) {
i += 1
val v = treeMap.keysIterator.drop(N - 1).next
}
val end2 = System.nanoTime()
val elapsed2 = end2 - start2
println("N = %d, elapsed1 = %d, elapsed2 = %d".format(N,elapsed1,elapsed2))
N += 50
}
}
object Test {
def main(args:Array[String]) {
val test = new Test
}
}
看来大牛的建议确实更好
结果
N = 1, elapsed1 = 956492000, elapsed2 = 700300000
N = 51, elapsed1 = 1103271000, elapsed2 = 936045000
N = 101, elapsed1 = 1286896000, elapsed2 = 1041744000
N = 151, elapsed1 = 1368854000, elapsed2 = 1199766000
N = 201, elapsed1 = 1584878000, elapsed2 = 1333284000
N = 251, elapsed1 = 1790965000, elapsed2 = 1468806000
N = 301, elapsed1 = 2052298000, elapsed2 = 1649021000
N = 351, elapsed1 = 2294625000, elapsed2 = 1819525000
N = 401, elapsed1 = 2529855000, elapsed2 = 1961699000
N = 451, elapsed1 = 2762582000, elapsed2 = 2100127000
N = 501, elapsed1 = 2977613000, elapsed2 = 2232108000
N = 551, elapsed1 = 3211812000, elapsed2 = 2384940000
N = 601, elapsed1 = 3437116000, elapsed2 = 2539431000
N = 651, elapsed1 = 3652749000, elapsed2 = 2650910000
N = 701, elapsed1 = 3900431000, elapsed2 = 2807085000
N = 751, elapsed1 = 4123141000, elapsed2 = 2934904000
N = 801, elapsed1 = 4337909000, elapsed2 = 3060158000
N = 851, elapsed1 = 4554490000, elapsed2 = 3188378000
N = 901, elapsed1 = 4768488000, elapsed2 = 3306528000
N = 951, elapsed1 = 4978839000, elapsed2 = 3413813000
最佳答案
不可变的 SortedSet
和 SortedMap
最近进行了大修。 2.10 版将有许多改进,包括对查找第一个和最后一个元素以及 take
、 drop
、 slice
的改进。
现在,您提出的解决方案 take(N).lastKey
并不是特别有效。相反,我会做 iterator.drop(n - 1).next._1
。它也不是特别有效,我会在选择一个之前对这两种解决方案进行基准测试。
关于performance - 在 TreeMap 中最有效地找到第 N 个键,我们在Stack Overflow上找到一个类似的问题:https://stackoverflow.com/questions/9670925/