dK-series [1] is a family of randomization methods for networks. The dK-series produces randomized networks that preserve up to the individual node’s degree, node’s degree correlation, and node’s clustering coefficient of the given unweighted network, depending on the parameter value d = 0, 1, 2, or 2.5.
We provide code for the dK-series in C++.
Require gcc version 4.2.1 or later.
We have confirmed that our code works in the following environment.
- macOS 12.4
(i) Clone this repository:
git clone [email protected]:kazuibasou/dk-series.git
(ii) Go to dk-series/src/:
cd dk-series/src
(iii) Run the make command:
make
This generates the following structure of the directory.
dk-series/
├ bin/
├ data/
├ src/
└ rand_network/
If you find a file dk_series in the folder dk-series/bin/, the build has been successfully completed.
We need to feed a file, network.txt, where network indicates the name of the network and is arbitrary.
The file should be placed in dk-series/data/.
In the file, each line contains two integers separated by a half-width space, which represents an edge between two nodes.
This follows the standard network data format.
Note that our code successfully works as long as each node's index is an integer. There is no need for the node index to start at 0 or to increment by 1.
Let's consider a network that consists of a set of nodes V = {0, 1, 2, 3, 4} and a set of edges E = {{0, 1}, {0, 4}, {1, 2}, {2, 3}, {3, 4}}. Then, the input file should be as follows:
0 1
0 4
1 2
2 3
3 4
Go to dk-series/bin/ and run the following command:
./dk_series <network> <d> <num_gen>
The three arguments are as follows.
The name of the network.
The value of d, which should be 0, 1, 2, or 2.5.
The number of networks to be generated.
To generate three randomized networks with d = 2.5 for the network named example-network, go to dk-series/bin/ and run the following command:
./dk_series example-network 2.5 3
The n th (n=1, ..., num_gen) randomized network, i.e., network_d_n.txt will be created in the folder dk-series/rand_network/.
-
In general, when d = 0, 1, or 2, the code runs fast. When d = 2.5, it takes longer.
-
Multiple edges and loops are allowed in randomized networks.
[1] Orsini, C., Dankulov, M., Colomer-de-Simón, P. et al. Quantifying randomness in real networks. Nat. Commun. 6, 8627 (2015). [paper]
This source code is released under the MIT License, see LICENSE.txt.
- Kazuki Nakajima (https://kazuibasou.github.io/index_en.html)
- kazuibasou[at]gmail.com
(Last update: 2022/10/14)