如何从文本文件创建快速图表？ [英] How to create a fast graph from a text file ?

问题描述

我想从包含36692个节点的文本文件中创建一个图形，文本的每一行都包含图形边缘的源节点和目标节点。我使用igraph库来创建图形。我写了下面的代码，但它太慢了。如何将整个文本文件读入内存缓冲区以使算法更快？

I want to make a graph from a text file containing 36692 nodes and each line of the text contains the source and the target node of an edge of the graph. I used the igraph library to create the graph. I wrote the following code but it is too slow. how can I read the whole text file into memory buffer to make the algorithm faster?

igraph_empty(&graph, vertexNum, 0);
ifstream inputFile("Email-Enron.txt");
int v1, v2;
while( ( inputFile >>  v1 >> v2 ))
{
	igraph_add_edge(&graph, v1, v2);
}

我的尝试：



此代码适用于小数据集，但大数据集需要很长时间。

What I have tried:

this code works for small data set, but it takes a long time for big dataset.

推荐答案

读取一个数据集，然后将其添加到图表一遍又一遍地使代码变慢，因为每次都是磁盘访问而不是图形添加。

Reading one data set and than adding it to the graph over and over again makes the code slow, because everytime is a disk access and than the graph addition.

int countLines = vertextNum;// compute size
int *v = new int[2*countLines];//double the amount of data
int i = 0;
//read all data
for( int i =  0; i < countLines; i += 2)
{
	inputFile >>  v[i] >> v[i+1];
}
//make graph
for( int i =  0; i < countLines; i += 2)
{
	igraph_add_edge(&graph,  v[i], v[i+1] );
}

如果可以在加载过程中停用重新计算或绘制图形。



如果发布代码是comimg some编译器中的优化标志和与展开相关的链接器。



提示：真正的快速解决方案是int字节在二进制文件中，你可以读取它们直接。这将使这一单行读取：

If it is possible deactivate recalculating or drawing of the graph in the load process.

For release code are comimg some optimization flags in the compiler and linkers in relevance like unrolling.

Tip: the real fast solution would be that the int bytes are in a binary file and you may read them direct. This would make this single one reading line:

inputFile.read( (char*)v, sizeof(unsigned int) * 2 * countLines);

数据以这种方式编写

Data was written in that manner

outputFile.write( (char*)v, sizeof(unsigned int) * 2 * countLines);

您可能最好使用 igraph_add_edges 函数，一次添加多个边，如 igraph参考手册 [ ^ ]。

You probably better use the igraph_add_edges function, for adding multiple edges at once, as shown in the igraph Reference Manual[^].

这篇关于如何从文本文件创建快速图表？的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！