Python中的Dask数组

Python Dask数组 (Python Dask Array)

Dask is parallel computing python library and it is mainly used to run across multiple systems. Dask is used to process the data efficiently on a different cluster of machines. Dask can completely use all the cores available in the machine.

Dask是并行计算的python库，主要用于跨多个系统运行。 Dask用于在其他计算机群集上有效地处理数据。 Dask可以完全使用机器中可用的所有内核。

Dask stores the complete data on the disk and uses chunks of data from the disk for processing. Dask analyzes the large data sets with the help of Pandas data frame and "numpy arrays".

Dask将完整的数据存储在磁盘上，并使用磁盘中的数据块进行处理。 Dask借助Pandas数据框和“ numpy数组” 来分析大型数据集。

Basically, dask arrays are distributed "numpy arrays". A large "numpy array" is divided into smaller arrays and they are grouped together to form dask array.

基本上， dask数组是分布式的“ numpy数组”。大的“ numpy数组”分为较小的数组，它们组合在一起形成dask数组 。

Install using this command:

使用以下命令进行安装：

    pip install dask

Dask array.asarray is used to convert the given input into dask array. It converts lists, tuples, numpy array to dask array.

Dask array.asarray用于将给定的输入转换为dask array 。它将列表，元组，numpy数组转换为dask数组 。

Program to create a dask array:

程序创建一个dask数组：

Example #1:

范例1：

import dask.array as p 
rk = [1,2,3,4,5] 	#converts the list into dask array 
d=p.asarray(rk) 
print(d.compute()) 	#print type of d 
print(type(d)) 
r = (1,2,3) 		#converts the tuple into dask array 
k=p.asarray(r) 
print(k.compute()) 	#print type of k 
print(type(k))

Output

输出量

[1 2 3 4 5]
<class 'dask.array.core.Array'>
[1 2 3]
<class 'dask.array.core.Array'>

Example #2:

范例2：

import dask.array as p 
import numpy as np 
#create a numpy array 
r=np.arange(5) 
print(r) 		#print type of numpy array 
print(type(r)) 	#converting numpy array to dask array 
d=p.asarray(r) 
print(d.compute()) 
print(type(d)) 
t=np.array([1,2,3]) 
print(t) 		#print type of numpy array 
print(type(t)) 	#converting numpy array to dask array 
f=p.asarray(t) 
print(f.compute()) #print type of dask array 
print(type(f))

Output

输出量

[0 1 2 3 4]
<class 'numpy.ndarray'>
[0 1 2 3 4]
<class 'dask.array.core.Array'>
[1 2 3]
<class 'numpy.ndarray'>
[1 2 3]
<class 'dask.array.core.Array'>