python尾递归

在华迪实训的第二天,还在讲python的基础知识,一些函数例如:map、filter、reduce、lambda……这些会的就不记了,还是来看看尾递归。

尾递归的定义:如果一个函数中所有递归形式的调用都出现在函数的末尾,我们称这个递归函数是尾递归的。 当递归调用是整个函数体中最后执行的语句且它的返回值不属于表达式的一部分时,这个递归调用就是尾递归。尾递归函数的特点是在回归过程中不用做任何操作,这个特性很重要,因为大多数现代的编译器会利用这种特点自动生成优化的代码。(摘自百度百科)

以阶乘为例:
普通递归

def func(n):
    if n == 0:
        return 1
    return func(n - 1) * n

尾递归

def func(n, res=1):
    if n == 0:
        return res
    return func(n - 1, n * res)

普通的递归,每一次递归都需要调用函数,会创建新的栈,这样就容易造成栈溢出,而尾递归则可以很好地解决这个问题,尾递归不会保存中间变量,每一级调用直接返回函数的返回值更新调用栈,而不用创建新的调用栈。

但是!!! python的编译器就是百度百科里说的那些少数不支持尾递归的编译器……

Traceback (most recent call last):
  File "C:/demo/TR.py", line 13, in <module>
    print(func(999))
  File "C:/demo/TR.py", line 10, in func
    return func(n - 1, n * res)
  File "C:/demo/TR.py", line 10, in func
    return func(n - 1, n * res)
  File "C:/demo/TR.py", line 10, in func
    return func(n - 1, n * res)
  [Previous line repeated 995 more times]
  File "C:/demo/TR.py", line 8, in func
    if n == 0:
RecursionError: maximum recursion depth exceeded in comparison

Process finished with exit code 1

可以看到,上述代码发生了溢出。
网上有大佬用装饰器实现了python的尾递归
还有种方法就是用生成器,用yield关键词替换return。

import types


def func(n, res=1):
    if n == 0:
        yield res
    yield func(n - 1, n * res)


def cal(gen, *args, **kwargs):
    g = gen(*args, **kwargs)
    while isinstance(g, types.GeneratorType):
        g = g.__next__()
    return g


print(cal(func, 999))

运行一下

402387260077093773543702433923003985719374864210714632543799910429938512398629020592044208486969404800479988610197196058631666872994808558901323829669944590997424504087073759918823627727188732519779505950995276120874975462497043601418278094646496291056393887437886487337119181045825783647849977012476632889835955735432513185323958463075557409114262417474349347553428646576611667797396668820291207379143853719588249808126867838374559731746136085379534524221586593201928090878297308431392844403281231558611036976801357304216168747609675871348312025478589320767169132448426236131412508780208000261683151027341827977704784635868170164365024153691398281264810213092761244896359928705114964975419909342221566832572080821333186116811553615836546984046708975602900950537616475847728421889679646244945160765353408198901385442487984959953319101723355556602139450399736280750137837615307127761926849034352625200015888535147331611702103968175921510907788019393178114194545257223865541461062892187960223838971476088506276862967146674697562911234082439208160153780889893964518263243671616762179168909779911903754031274622289988005195444414282012187361745992642956581746628302955570299024324153181617210465832036786906117260158783520751516284225540265170483304226143974286933061690897968482590125458327168226458066526769958652682272807075781391858178889652208164348344825993266043367660176999612831860788386150279465955131156552036093988180612138558600301435694527224206344631797460594682573103790084024432438465657245014402821885252470935190620929023136493273497565513958720559654228749774011413346962715422845862377387538230483865688976461927383814900140767310446640259899490222221765904339901886018566526485061799702356193897017860040811889729918311021171229845901641921068884387121855646124960798722908519296819372388642614839657382291123125024186649353143970137428531926649875337218940694281434118520158014123344828015051399694290153483077644569099073152433278288269864602789864321139083506217095002597389863554277196742822248757586765752344220207573630569498825087968928162753848863396909959826280956121450994871701244516461260379029309120889086942028510640182154399457156805941872748998094254742173582401063677404595741785160829230135358081840096996372524230560855903700624271243416909004153690105933983835777939410970027753472000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000

Process finished with exit code 0

完事儿~

你可能感兴趣的:(编程之路)