有没有更快的方法来搜索numpy数组

I have a numpy array of roughly 3125000 entries the data is structured using the following dtype

我有一个大约3125000个条目的numpy数组，使用以下dtype构造数据

dt = np.dtype([('startPoint', '<u8' ), ('endPoint', '<u8')])

The data is from a file that has been previously sorted by endPoint before it is read into the array.

数据来自先前已被endPoint排序的文件，然后将其读入阵列。

I now need to search the array and check if it contains a particular endpoint and I'm doing this using a binary search using the following code

我现在需要搜索数组并检查它是否包含特定端点，并且我使用以下代码使用二进制搜索来执行此操作

def binarySearch(array, index):
lowPoint = 0
highpoint = len(array) - 1


while (lowPoint <= highpoint):
    midPoint = int((lowPoint + highpoint) / 2)

    if(index == array[midPoint]['endPoint']):
        return midPoint

    elif(index < array[midPoint]['endPoint']):
        highpoint = midPoint - 1

    else:
        lowPoint = midPoint + 1

return -1

My question is is there a faster way to search for an entry in this array. As in is there a built in Numpy search that may be faster than my binary search.

我的问题是有更快的方法来搜索此数组中的条目。因为有一个内置的Numpy搜索可能比我的二进制搜索更快。

1 个解决方案

#1

Try numpy.searchsorted, also you can use memory mapping if the array is too large. searchsorted is implemented as binary search.

尝试numpy.searchsorted，如果数组太大，你也可以使用内存映射。 searchsorted实现为二进制搜索。