根据python中的列表重新排列数组元素
原文标题 :rearrange an array element based on list in python
我有一个二维数组a
,大小2, 1403
和一个列表b
,它有 2 个列表。
a.shape = (2, 1403) # a is 2D array, each row has got unique elements.
len(b) = 2 # b is list
len(b[0]), len(b[1]) = 415, 452 # here also both the list inside b has got unique elements
b[0] and b[1]
中存在的所有元素分别存在于a[0] and a[1]
中
现在我想根据b
的元素重新排列a
的元素。我想重新排列,使得b[0]
中的所有元素也出现在a[0]
中,都应该出现在a[0]
的结尾,意思是新的a
应该是这样的a[0][:-len(b[0])] = b[0]
,同样地a[1][:-len(b[1])] = b[1]
。
玩具示例
a
有像[[1,2,3,4,5,6,7,8,9,10,11,12],[1,2,3,4,5,6,7,8,9,10,11,12]
这样的元素
b
有像[[5, 9, 10], [2, 6, 8, 9, 11]]
这样的元素
new_a
变成[[1,2,3,4,6,7,8,11,12,5,9,10], [1,3,4,5,7,10,12,2,6,8,9,11]]
我编写了一个代码,它遍历所有变得非常慢的元素,如下所示
a_temp = []
remove_temp = []
for i, array in enumerate(a):
a_temp_inner = []
remove_temp_inner = []
for element in array:
if element not in b[i]:
a_temp_inner.append(element) # get all elements first which are not present in b
else:
remove_temp_inner.append(element) #if any element present in b, remove it from main array
a_temp.append(a_temp_inner)
remove_temp.append(b_temp_inner)
a_temp = torch.tensor(a_temp)
remove_temp = torch.tensor(remove_temp)
a = torch.cat((a_temp, remove_temp), dim = 1)
谁能帮我做一些比这更好的更快的实现
回复
我来回复-
paul-shuvo 评论
这是我的方法:
index_ = np.array([[False if i in d else True for i in c] for c, d in zip(a,b)]) arr_filtered =[[np.extract(ind, c) for c, d, ind in zip(a,b,index_)], [np.extract(np.logical_not(ind), c) for c, d, ind in zip(a,b, index_)]] arr_final = ar = np.array([np.concatenate((i, j)) for i, j in zip(*arr_filtered)])
2年前 -
假设gais a
np.array
,bis alist
你可以使用np.array([np.concatenate((i[~np.in1d(i, j)], j)) for i, j in zip(a,b)])
输出
array([[ 1, 2, 3, 4, 6, 7, 8, 11, 12, 5, 9, 10], [ 1, 3, 4, 5, 7, 10, 12, 2, 6, 8, 9, 11]])
如果bcontains empty,可以进行微优化
lists
np.array([np.concatenate((i[~np.in1d(i, j)], j)) if j else i for i, j in zip(a,b)])
在我的基准测试中,对于
np.arrays
少于 ~100 个元素的转换.tolist()
比np.concatenate
快np.array([i[~np.in1d(i, j)].tolist() + j for i, j in zip(a,b)])
此解决方案的数据示例和导入
import numpy as np a = np.array([ [1,2,3,4,5,6,7,8,9,10,11,12], [1,2,3,4,5,6,7,8,9,10,11,12] ]) b = [[5, 9, 10], [2, 6, 8, 9, 11]]
2年前