Pigeon-inspired optimization (PIO) is a swarm intelligence optimizer inspired by the homing behavior of pigeons. PIO consists of two optimization stages which employ the map and compass operator, and the landmark operator, respectively. In canonical PIO, these two operators treat every bird equally, which deviates from the fact that birds usually act heterogenous roles in nature. In this paper, we propose a new variant of PIO algorithm considering bird heterogeneity — HPIO. Both of the two operators are improved through dividing the birds into hub and non-hub roles. By dividing the birds into two groups, these two groups of birds are respectively assigned with different functions of “exploitation” and “exploration”, so that they can closely interact with each other to locate the best promising solution. Extensive experimental studies illustrate that the bird heterogeneity produced by our algorithm can benefit the information exchange between birds so that the proposed PIO variant significantly outperforms the canonical PIO.
This work was supported by National Key Research and Development Program of China (Grant No. 2016YFB1200100), National Natural Science Foundation of China (Grant Nos. 61425014, 61521091, 91538204, 61671031, 61722102).
[1] Goldberg D E. Genetic Algorithms in Search, Optimization, and Machine Learning. Massachusetts: Addison-Wesley Professional, 1989. Google Scholar
[2] Kennedy J, Eberhart R. Particle swarm optimization. In: Proceedings of the International Conference on Neural Networks, Perth, 1995. 1942--1948. Google Scholar
[3] Liang J J, Qin A K, Suganthan P N. Comprehensive learning particle swarm optimizer for global optimization of multimodal functions. IEEE Trans Evol Computat, 2006, 10: 281-295 CrossRef Google Scholar
[4] Duan H, Qiao P. Pigeon-inspired optimization: a new swarm intelligence optimizer for air robot path planning. Int Jnl Intel Comp Cyber, 2014, 7: 24-37 CrossRef Google Scholar
[5] Duan H, Luo Q. New progresses in swarm intelligence-based computation. IJBIC, 2015, 7: 26-35 CrossRef Google Scholar
[6] Duan H, Wang X. Echo State Networks With Orthogonal Pigeon-Inspired Optimization for Image Restoration.. IEEE Trans Neural Netw Learning Syst, 2016, 27: 2413-2425 CrossRef PubMed Google Scholar
[7] Dou R, Duan H. Lévy flight based pigeon-inspired optimization for control parameters optimization in automatic carrier landing system. Aerospace Sci Tech, 2017, 61: 11-20 CrossRef Google Scholar
[8] Li Z, Liu J, Wu K. A Multiobjective Evolutionary Algorithm Based on Structural and Attribute Similarities for Community Detection in Attributed Networks.. IEEE Trans Cybern, 2018, 48: 1963-1976 CrossRef PubMed Google Scholar
[9] Wang S, Liu J. A Multi-Objective Evolutionary Algorithm for Promoting the Emergence of Cooperation and Controllable Robustness on Directed Networks. IEEE Trans Netw Sci Eng, 2018, 5: 92-100 CrossRef Google Scholar
[10] Xin L, Xian N. Biological object recognition approach using space variant resolution and pigeon-inspired optimization for UAV. Sci China Technol Sci, 2017, 60: 1577-1584 CrossRef Google Scholar
[11] West D B. Introduction to Graph Theory (2nd ed). New Jersey: Prentice Hall, 2001. Google Scholar
[12] Watts D J, Strogatz S H. Collective dynamics of `small-world' networks. Nature, 1998, 393: 440-442 CrossRef PubMed ADS Google Scholar
[13] Nagy M, ákos Z, Biro D. Hierarchical group dynamics in pigeon flocks. Nature, 2010, 464: 890-893 CrossRef PubMed ADS arXiv Google Scholar
[14] Couzin I D, Krause J, Franks N R. Effective leadership and decision-making in animal groups on the move. Nature, 2005, 433: 513-516 CrossRef PubMed ADS Google Scholar
[15] Cavagna A, Cimarelli A, Giardina I. Scale-free correlations in starling flocks. Proc Natl Acad Sci USA, 2010, 107: 11865-11870 CrossRef PubMed ADS arXiv Google Scholar
[16] Couzin I D, Krause J, Franks N R. Particle Swarm Optimization with Scale-Free Interactions. PLoS ONE, 2014, 9: e97822 CrossRef PubMed ADS Google Scholar
[17] Mendes R, Kennedy J, Neves J. The Fully Informed Particle Swarm: Simpler, Maybe Better. IEEE Trans Evol Computat, 2004, 8: 204-210 CrossRef Google Scholar
[18] Albert R, Barabási A L. Statistical mechanics of complex networks. Rev Mod Phys, 2002, 74: 47-97 CrossRef ADS Google Scholar
[19] Gao Y, Du W, Yan G. Selectively-informed particle swarm optimization. Sci Rep, 2015, 5: 9295 CrossRef PubMed ADS Google Scholar
[20] Xin Yao , Yong Liu , Guangming Lin . Evolutionary programming made faster. IEEE Trans Evol Computat, 1999, 3: 82-102 CrossRef Google Scholar
[21] Liang J J, Suganthan P N, Deb K, et al. Novel composition test functions for numerical global optimization. In: Proceedings of 2005 IEEE Swarm Intelligence Symposium Swarm Intelligence Symposium, Pasadena, 2005. 68--75. Google Scholar
Figure 1
(Color online) Networks associating with the bird topology. (a) A fully-connected network, where each two nodes are linked by an edge; (b) a scale-free network, where the node degree follows the power-law distribution, which means most nodes (non-hub nodes) have low degrees, while a few nodes (hub nodes) have high degrees; (c) a random network
Figure 2
(Color online) Variation of the average fitness of birds as the optimization proceeds.
Figure 3
(Color online) Variation of hub/non-hub birds' characteristics as optimization proceeds. (a) Average fitness of the hub/non-hub birds; (b) average times of hub/non-hub birds being learned from other birds.
Canonical PIO | HPIO |
Input | |
$N_p$: number of individuals in pigeon swarm; $D$: dimension of the search space; $R$: the map and compass factor; | |
$t_1$: the number of iterations that the map and compass operator is executed; | |
$t_2$: the maximum number of iterations that the landmark operator is executed; | |
Search range: the borders of the search space. | |
$k_i$: degree of $i$th bird; | |
$k_c$: minimum of hub birds' degrees (viz. the threshold which splits the birds into hub and non-hub groups). | |
Initialization | |
Set initial values for $N_p,~D,~R,~t_1,~t_2$ and the search range. | |
Set initial position $\boldsymbol{X}_i^0\in\mathbb{R}^D$ and velocity $\boldsymbol{V}_i^0\in\mathbb{R}^D$ in the search range for each pigeon individual $f\left(\boldsymbol{X}_i^0\right)~(1\le~i\le~N_p)$. | |
Calculate fitness values of different pigeon individuals $f\left(\boldsymbol{X}_i^0\right)~(1\le~i\le~N_p)$. | |
Set $\boldsymbol{X}_{\rm~gbest}$ is the position of the bird having the best fitness among all the birds. | |
Set $\boldsymbol{X}_{{\rm~pbest}_i}=\boldsymbol{X}_i^0,~1\le~i\le~N_p$. | |
Set $\boldsymbol{X}_{{\rm~cbest}_i}=\boldsymbol{X}_j^0,~j\in\left\{i,\mathcal{N}\left(i\right)\right\}$ and $j$'s fitness is the best, $\mathcal{N}\left(i\right)$ is the set of the neighbors of $i$. | |
Randomly generate a scale-free network using Barabási-Albert model. This network has $N_p$ nodes, each representing an individual bird. Regard bird $j~(j\neq~i)$ as bird $i$'s neighbor if $j$ is linked to $i$ in this network. | |
Map and compass operator execution | |
For $N_c=1$ to $t_1$ do | For $N_c=1$ to $t_1$ do |
For $i=1$ to $N_p$ do | For $i=1$ to $N_p$ do |
$\boldsymbol{V}_i^{N_c}=\boldsymbol{V}_i^{N_c-1}{\rm~exp}\left(-R\cdot~N_c\right)+{\rm~rand}\left(0,1\right)$ | $\boldsymbol{V}_i^{N_c}=\boldsymbol{V}_i^{N_c}\cdot~{\rm~exp}{\left(-R\cdot~t\right)}$ |
$\cdot(\boldsymbol{X}_{\rm~gbest}^{N_c-1}-\boldsymbol{X}_i^{N_c-1}),$ | |
(${\rm~rand}\left(0,1\right)$ is a random number among 0,1)) | $+\left\{ ~~\begin{aligned} ~~\frac{1}{k_i}\sum_{j\in\mathcal{N}\left(i\right)}{{{\rm~rand}\left(0,1\right)}\cdot\left(\boldsymbol{X}_{{\rm~pbest}_j}-x_i\right)},&~~k_i>k_c,\\ ~~{\rm~rand}\left(0,1\right)\cdot(\boldsymbol{X}_{{\rm~cbest}_i}-x_i),&~~k_i\leq~k_c, ~~\end{aligned} ~~\right.$ ($\boldsymbol{X}_{{\rm~pbest}_j}$ is defined as $j$'s best position through all the iterations before $N_c$ (not included), $\boldsymbol{X}_{{\rm~cbest}_i}$ is defined as the best position of $i$'s neighbors and $i$ through all iterations before $N_c$ (not included); (${\rm~rand}\left(0,1\right)$ is a random number among 0,1)) |
$\boldsymbol{X}_i^{N_c}=\boldsymbol{X}_i^{N_c-1}+\boldsymbol{V}_i^{N_c}$, | $\boldsymbol{X}_i^{N_c}=\boldsymbol{X}_i^{N_c-1}+\boldsymbol{V}_i^{N_c}$, |
Update $\boldsymbol{X}_{\rm~gbest}$. | Update $\boldsymbol{X}_{{\rm~pbest}_j},~\boldsymbol{X}_{{\rm~cbest}_i}$. |
End for | End for |
End for | End for |
Landmark operator execution | |
For $N_c=t_1+1$ to $t_1+t_2$ do | For $N_c=t_1+1$ to $t_1+t_2$ do |
Rank all the available birds individuals according to their | Rank all the available birds individuals according to |
fitness values $f(\boldsymbol{X}_i^{N_c})~(1\le~i\le~N_p)$. | their degrees in the scale-free network. |
Abandon half of the birds having relatively worse fitness | Abandon the birds whose degrees are the lowest in |
in the swarm ($N_p=N_p/2$). | the swarmp (update $N_p$). |
$F(\boldsymbol{X}_i^{N_c})=\left\{ ~~\begin{aligned} ~~\frac{1}{f(\boldsymbol{X}_i^{N_c})+\varepsilon},~&{\text{for}}~~f~~~{\text{minimization}},\\ ~~f(\boldsymbol{X}_i^{N_c}),~~~~~~&{\text{for}}~~f~~{\text{maximization}},\\ ~~\end{aligned} ~~\right.$ | $F(\boldsymbol{X}_i^{N_c})=\left\{ ~~\begin{aligned} ~~\frac{1}{f(\boldsymbol{X}_i^{N_c})+\varepsilon},~&{\text{for}}~~f~~{\text{minimization}},\\ ~~f(\boldsymbol{X}_i^{N_c}),~~~~~~&{\text{for}}~~f~~{\text{maximization}},\\ ~~\end{aligned} ~~\right.$ |
$\boldsymbol{X}_{\rm~center}^{N_c}=\frac{\sum_{i=1}^{N_p}{\boldsymbol{X}_i^{N_c}F(\boldsymbol{X}_i^{N_c})}}{N_p\sum_{i=1}^{N_p}{F(\boldsymbol{X}_i^{N_c})}}$, | $\boldsymbol{X}_{\rm~center}^{N_c}=\frac{\sum_{i=1}^{N_p}{\boldsymbol{X}_i^{N_c}F(\boldsymbol{X}_i^{N_c})}}{N_p\sum_{i=1}^{N_p}{F(\boldsymbol{X}_i^{N_c})}}$, |
$\boldsymbol{X}_i^{N_c}=\boldsymbol{X}_i^{N_c-1}+{\rm~rand}\cdot(\boldsymbol{X}_{\rm~center}^{N_c}-\boldsymbol{X}_i^{N_c-1})$, | $\boldsymbol{X}_i^{N_c}=\boldsymbol{X}_i^{N_c-1}+{\rm~rand}\cdot(\boldsymbol{X}_{\rm~center}^{N_c}-\boldsymbol{X}_i^{N_c-1})$, |
Update $\boldsymbol{X}_{\rm~gbest}$. | Update $\boldsymbol{X}_{\rm~gbest}$. |
End for | End for |
Note: if $N_p=1$ after a number of iterations executing the landmark operator, the iterative process will be forced to | |
terminated, namely, the actual number of the landmark operator execution iteration may less than $t_2$. | |
Output | |
$\boldsymbol{X}_{\rm~gbest}$ is output as the best final solution according to the fitness function $f$. |
Method | Description |
PIO | The canonical PIO |
ERPIO | The PIO using the bird interaction topology of ErdH os-Rényi random network (refer to Figure |
EHPIO | The HPIO whose bird interaction topology network is replaced as ErdH os-Rényi random network |
SWPIO | The PIO using the bird interaction topology of small-world network (refer to Figure |
SHPIO | The HPIO whose bird interaction topology network is replaced as small-world network |
SFPIO | The PIO using the bird interaction topology of scale-free network (please refer to Figure |
SIPIO | The SFPIO whose birds use a selective-informed learning strategy aiming at bringing in bird heterogeneity (The map and compass operator of SIPIO is the same with that of HPIO, while the landmark of SIPIO remains the same with that of canonical PIO) |
Category | Number | Function | Expression |
Unimodal | 1 | Sphere | $f_1\left(x\right)=\sum_{i=1}^{D}x_i^2$ |
2 | Rosenbrock | $f_2\left(x\right)=\sum_{i=1}^{D}{{(100(x}_i^2-x_{i+1}})^2+{(x_i-1)}^2)$ | |
3 | Schwefel P2.22 | $f_3\left(x\right)=\sum_{i=1}^{D}\left|x_i\right|+\prod_{i=1}^{D}\left|x_i\right|$ | |
4 | Quartic Noise | $f_4\left(x\right)=\sum_{i=1}^{D}{{ix}_i^2+{\rm~random}\left[0,\left.~1\right)\right.}$ | |
5 | Schwefel P1.22 | $f_5\left(x\right)={(\sum_{i=1}^{D}x_i)}^2$ | |
Mutimodal | 6 | Ackley | $f_6\left(x\right)=-20\exp{\left(-0.2\sqrt{\frac{1}{D}\sum_{i=1}^{D}x_i^2}\right)}$ |
$-\exp{\left(\frac{1}{D}\sum_{i=1}^{D}\cos{\left(2\pi~x_i\right)}\right)}+20+e$ | |||
7 | Rastrigin | $f_7\left(x\right)=\sum_{i=1}^{D}{(x_i^2-10\cos{\left(2\pi~x_i\right)}+10)}$ | |
8 | Rastrigin (discrete) | $f_8\left(x\right)=\sum_{i=1}^{D}{(y_i^2-10\cos{\left(2\pi~y_i\right)}+10)},$ | |
$y_i=\left\{ ~~\begin{aligned} ~~x_i,&~~|x_i|<\frac{1}{2},\\ ~~{\rm~round}(2x_i)/2,&~~|x_i|\ge\frac{1}{2}\\ ~~\end{aligned} \right.$ | |||
9 | Weierstrass | $f_9\left(x\right)=\sum_{i=1}^{D}\left(\sum_{k=0}^{k_{\rm~max}}\left[a^k\cos{\left(2\pi~b^k\left(x_i+0.5\right)\right)}\right]\right)$ | |
$-D\sum_{k=0}^{k_{\rm~max}}\left[a^k\cos{\left(\pi~b^k\right)}\right]$ | |||
10 | Griewank | $f_{10}\left(x\right)=\frac{1}{4000}\sum_{i=1}^{D}x_i^2-\prod_{i=1}^{D}{\cos{\left(\frac{x_i}{\sqrt~i}\right)}+1}$ | |
Rotated multimal$^{\rm~c)}$ | 11 | Ackley (rotated) | $f_{11}\left(x\right)=-20\exp{\left(-0.2\sqrt{\frac{1}{D}\sum_{i=1}^{D}y_i^2}\right)}$ |
$-\exp{\left(\frac{1}{D}\sum_{i=1}^{D}\cos{\left(2\pi~y_i\right)}\right)}+20+e$ | |||
12 | Rastrigin (rotated) | $f_{12}\left(x\right)=\sum_{i=1}^{D}{(y_i^2-10\cos{\left(2\pi~y_i\right)}+10)}$ | |
13 | Rastrigin (discrete and rotated) | $f_{13}\left(x\right)=\sum_{i=1}^{D}{(z_i^2-10\cos{\left(2\pi~z_i\right)}+10)},$ | |
$z_i=\left\{ ~~\begin{aligned} ~~y_i,&~~|y_i|<\frac{1}{2},\\ ~~{\rm~round}(2y_i)/2,&~~|y_i|\ge\frac{1}{2}\\ ~~\end{aligned} \right.$ | |||
14 | Weierstrass (rotated) | $f_{14}\left(x\right)=\sum_{i=1}^{D}\left(\sum_{k=0}^{k_{\rm~max}}\left[a^k\cos{\left(2\pi~b^k\left(y_i+0.5\right)\right)}\right]\right)$ | |
$-D\sum_{k=0}^{k_{\rm~max}}\left[a^k\cos{\left(\pi~b^k\right)}\right]$ | |||
15 | Griewank (rotated) | $f_{15}\left(x\right)=\frac{1}{4000}\sum_{i=1}^{D}y_i^2-\prod_{i=1}^{D}{\cos{\left(\frac{y_i}{\sqrt~i}\right)}+1}$ | |
Composite | 16 | Composite 1 | $f_{16}$ is composed by 10 Sphere functions |
17 | Composite 2 | $f_{17}$ is composed by 10 functions |
a
Category | Number | Function | Search range of $x_i$ ($1\leqslant~i\leqslant~D$) | Goal |
1 | Sphere | $\left[-100,100\right]$ | 0.01 | |
2 | Rosenbrock | $\left[-2.048,2.048\right]$ | 100 | |
Unimodal | 3 | Schwefel P2.22 | $\left[-10,10\right]$ | 0.01 |
4 | Quartic noise | $\left[-1.28,1.28\right]$ | 0.05 | |
5 | Schwefel P1.22 | $\left[-10,10\right]$ | 100 | |
6 | Ackley | $\left[-32,32\right]$ | 0.01 | |
7 | Rastrigin | $\left[-5.12,5.12\right]$ | 100 | |
Mutimodal | 8 | Rastrigin (discrete) | $\left[-0.5,0.5\right]$ | 0.01 |
9 | Weierstrass | $\left[-0.5,0.5\right]$ | 0.01 | |
10 | Griewank | $\left[-600,600\right]$ | 0.05 | |
11 | Ackley (rotated) | $\left[-32,32\right]$ | 0.01 | |
12 | Rastrigin (rotated) | $\left[-5.12,5.12\right]$ | 100 | |
Rotated multimal | 13 | Rastrigin (discrete and rotated) | $\left[-5.12,5.12\right]$ | 100 |
14 | Weierstrass (rotated) | $\left[-0.5,0.5\right]$ | 1 | |
15 | Griewank (rotated) | $\left[-600,600\right]$ | 0.05 | |
Composite | 16 | Composite 1 | $\left[-5,5\right]$ | 0.01 |
17 | Composite 2 | $\left[-5,5\right]$ | 10 |
Benchmark | PIO | ERPIO | EHPIO | SWPIO | SHPIO | SFPIO | SIPIO | HPIO | HPIO |
function | ($k_c=9$) | ($k_c=9$) | ($k_c=9$) | ($k_c=9$) | (best $k_c$) | ||||
$f_1$ | 1.92${\rm~E}-$3 | 5.74${\rm~E}-$7 | 1.35${\rm~E}-$14 | 5.73${\rm~E}-$6 | 3.25${\rm~E}-$14 | 2.22${\rm~E}-$4 | 4.66${\rm~E}-$5 | 1.05${\rm~E}-$14 $(k_c=15)$ | |
$f_2$ | 3.11${\rm~E}$1 | 2.82${\rm~E}$1 | 2.87E1 | 2.90${\rm~E}$1 | 2.91${\rm~E}$1 | 2.87${\rm~E}$1 | 2.88${\rm~E}$1 | 2.87${\rm~E}$1 $(k_c=17)$ | |
$f_3$ | – | 3.42${\rm~E}-$6 | 2.25${\rm~E}-$9 | 5.20${\rm~E}-$4 | 5.25${\rm~E}-$9 | 3.35${\rm~E}-$3 | 9.08${\rm~E}-$4 | 9.62${\rm~E}-$10 $(k_c=17)$ | |
$f_4$ | 6.99${\rm~E}-$4 | 4.99${\rm~E}-$5 | 5.28${\rm~E}-$5 | 5.13${\rm~E}-$5 | 9.87${\rm~E}-$5 | 5.48${\rm~E}-$5 | 5.45${\rm~E}-$5 | 4.81${\rm~E}-$5 $(k_c=5)$ | |
$f_5$ | 1.77 | 6.96${\rm~E}-$2 | 7.35${\rm~E}-$11 | 1.35${\rm~E}-$1 | 6.13${\rm~E}-$11 | 2.55${\rm~E}-$1 | 1.46${\rm~E}-$2 | 5.36${\rm~E}-$11 $(k_c=9)$ | |
$f_6$ | – | 2.82${\rm~E}-$5 | 4.31${\rm~E}-$9 | 4.11${\rm~E}-$4 | 5.09${\rm~E}-$9 | 4.33${\rm~E}-$3 | 2.18${\rm~E}-$3 | 4.27${\rm~E}-$9 $(k_c=9)$ | |
$f_7$ | 7.42${\rm~E}-$1 | 3.66${\rm~E}-$1 | 5.28${\rm~E}-$12 | 3.87${\rm~E}-$1 | 6.02${\rm~E}-$12 | 3.92${\rm~E}-$1 | 5.27${\rm~E}-$2 | 1.25${\rm~E}-$12 $(k_c=5)$ | |
$f_8$ | 9.12${\rm~E}-$1 | 4.58${\rm~E}-$1 | 9.88${\rm~E}-$11 | 4.89${\rm~E}-$1 | 9.98${\rm~E}-$11 | 5.47${\rm~E}-$1 | 9.87${\rm~E}-$2 | 2.84${\rm~E}-$11 $(k_c=1)$ | |
$f_9$ | 6.41${\rm~E}-$3 | 3.59${\rm~E}-$3 | 3.29${\rm~E}-$9 | 4.36${\rm~E}-$3 | 9.17${\rm~E}-$9 | 4.30${\rm~E}-$3 | 6.55${\rm~E}-$3 | 2.03${\rm~E}-$9 $(k_c=1)$ | |
$f_{10}$ | 3.43${\rm~E}-$3 | 3.60${\rm~E}-$6 | 4.68${\rm~E}-$14 | 1.11${\rm~E}-$4 | 4.17${\rm~E}-$14 | 7.26${\rm~E}-$4 | 5.34${\rm~E}-$4 | 4.98${\rm~E}-$15 $(k_c=15)$ | |
$f_{11}$ | – | 5.01${\rm~E}-$5 | 6.25${\rm~E}-$8 | 7.68${\rm~E}-$4 | 8.30${\rm~E}-$8 | 4.43${\rm~E}-$3 | 4.31${\rm~E}-$3 | 4.21${\rm~E}-$8 $(k_c=9)$ | |
$f_{12}$ | 1.12 | 5.22${\rm~E}-$1 | 5.00${\rm~E}-$12 | 5.82${\rm~E}-$1 | 5.82${\rm~E}-$12 | 6.45${\rm~E}-$1 | 8.64${\rm~E}-$2 | 4.02${\rm~E}-$12 $(k_c=5)$ | |
$f_{13}$ | 1.34 | 6.08${\rm~E}-$1 | 6.49${\rm~E}-$11 | 6.78${\rm~E}-$1 | 7.68${\rm~E}-$11 | 7.21${\rm~E}-$1 | 1.42${\rm~E}-$1 | 2.25${\rm~E}-$11 $(k_c=1)$ | |
$f_{14}$ | 1.06${\rm~E}-$2 | 6.66${\rm~E}-$3 | 5.93${\rm~E}-$9 | 7.21${\rm~E}-$3 | 3.84${\rm~E}-$9 | 7.25${\rm~E}-$3 | 3.02${\rm~E}-$3 | 8.65${\rm~E}-$10 $(k_c=1)$ | |
$f_{15}$ | 4.94${\rm~E}-$3 | 2.14${\rm~E}-$5 | 8.63${\rm~E}-$13 | 1.14${\rm~E}-$4 | 9.51${\rm~E}-$13 | 6.32${\rm~E}-$4 | 5.04${\rm~E}-$4 | 1.06${\rm~E}-$13 $(k_c=9)$ | |
$f_{16}$ | 1.34 | 8.99${\rm~E}-$1 | 5.26${\rm~E}-$1 | 9.00${\rm~E}-$1 | 6.00${\rm~E}-$1 | 9.21${\rm~E}-$1 | 6.55${\rm~E}-$1 | 1.68${\rm~E}-$1 $(k_c=17)$ | |
$f_{17}$ | 2.66 | 1.32 | 8.12${\rm~E}-$1 | 9.57${\rm~E}-$1 | 8.62${\rm~E}-$1 | 9.68${\rm~E}-$1 | 9.59${\rm~E}-$1 | 1.01${\rm~E}-$1 $(k_c=17)$ |
a
Copyright 2019 Science China Press Co., Ltd. 《中国科学》杂志社有限责任公司 版权所有
京ICP备18024590号-1