1.递归
#1.斐波那契(1,1,2,3,5,8,...前两项之和是后一项),递归问题(调用自身,结束条件) #一 。时间复杂度2**n,即n+1,时间复杂度翻一倍 def fibnacci(n): if n==0 or n==1: return 1 #结束条件 else: return fibnacci(n-1)+fibnacci(n-2)#前两项和,调用自身 @cal_time#递归函数不能直接用装饰器 def fib1(n): return fibnacci(n) #二。时间复杂度n,空间复杂度n, @cal_time def fib2(n): li=[1,1] for i in range(2,n+1): li.append(li[-1]+li[-2]) return li[n] #三。时间复杂度n def fib3(n): a=1 b=1 c=1 for i in range(2,n+1): c=a+b a=b b=c return c print(fib1(33)) #时间装饰器 import time def cal_time(func): def wrapper(*args,**kwargs): t1=time.time() result=func(*args,**kwargs) t2=time.time() print("%s running time:%s secs."%(func.__name__,t2-t1)) return result return wrapper
#递归,汉诺塔 def hanoi(n,A,B,C): if n>0: hanoi(n-1,A,C,B) print('%s->%s'%(A,C)) hanoi(n-1,B,A,C) hanoi(3,'A','B','C')
2.列表查找(2种)
#列表查找(2种):顺序查找O(n),二分查找O(logn) li=[1,2,5,67,2,6,7] 6 in li#True li.index(6)#6的索引是5 #线性查找,O(n) def line_search(num,li): count=0 for i in li: if i==num: return count else: count+=1 return -1 print(line_search(67,li)) #二分查找O(logn),有序列表才可以二分查找 def bin_search(li,val): low=0 high=len(li)-1 while low<=high: mid=(low+high)//2 if li[mid]==val: return mid elif li[mid]<val: low=mid+1 else: high=mid-1 return -1 print(bin_search([1,2,3,4,5,6,7,8,9],4)) #二分查找的递归调用 def bin_search_rec(data_set,value,low,high): if low<=high: mid=(low_high)//2 if data_set[mid]==value: return mid elif data_set[mid]>value: bin_search_rec[data_set,value,low,mid-1] else: bin_search_rec[data_set,value,low+1,high] else: return -1
动态规划
#动态规划 #递归 arr=[1,2,4,1,7,8,3] def rec_opt(arr,i): if i==0: return arr[0] elif i==1: return max(arr[0],arr[1]) else: A=rec_opt(arr,i-2)+arr[i] B=rec_opt(arr,i-1) return max(A,B) #列表保存 def dp_opt(arr): opt=np.zeros(len(arr)) opt[0]=arr[0] opt[1]=max(arr[0],arr[1]) for i in rang(2,len(arr)): A=opt[i-2]+arr[i] B=opt[i-1] opt[i]=max(A,B) return opt(len(arr)-1) arr=[3,34,4,12,5,2] def rec_subset(arr,i,s): if s==0: return True elif i==0: return arr[0]==s elif arr[i]>s: return rec_subset(arr,i-1,s) else: A=rec_subset(arr,i-1,s-arr[i]) B=rec_subset(arr,i-1,s) return A or B def dp_subset(arr,S): subset=np.zeros((len(arr),S+1),dtype=bool) subset[:,0]=True subset[0,:]=False subset[0,arr[0]]=True for i in rang(1,len(arr)): for s in range(1,S+1): if arr[i]>s: subset[i,s]=subset[i-1,s] else: A=subset[i-1,s-arr[i]] B=subset[i-1,s] subset[i,s]=A or B r,c=subset.shape return subset[r-1,c-1]