Skip to main content
Log in

Fibonacci helps to evacuate from a convex region in a grid network

  • Published:
Journal of Combinatorial Optimization Aims and scope Submit manuscript

Abstract

This study considers an evacuation problem where the evacuees try to escape to the boundary of an affected area, which is convex, and a grid network is embedded in the area. The boundary is unknown to the evacuees and we propose an online evacuation strategy based on the Fibonacci sequence. This strategy is proved to have a competitive ratio of 19.5, which is better than the best previously reported result of 21.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5

Similar content being viewed by others

References

  • Batta R, Chiu SS (1988) Optimal obnoxious paths on a network: transportation of hazardous materials. Oper Res 36:84–92

    Article  Google Scholar 

  • Borodin A, El-Yaniv R (1998) Online computation and competitive analysis. Cambridge University Press, Cambridge

    MATH  Google Scholar 

  • Burgard W, Moors M, Fox D, Simmons R, Thrun S (2000) Collaborative multi-robot exploration. In: Proceedings 2000 ICRA Millennium Conference IEEE International Conference on Robotics and Automation, vol. 1, pp. 476–481

  • Chen X, Zhan FB (2008) Agent-based simulation of evacuation strategies under different road network structures. J Oper Res Soc 59:25–33

    Article  Google Scholar 

  • Cohn JHE (1964) Square Fibonacci numbers, etc. Fibonacci Quart 2:109–113

    MathSciNet  MATH  Google Scholar 

  • Deng XT, Kameda T, Papadimitriou C (1998) How to learn an unknown environment I: the rectilinear case. J ACM 45(2):215–245

    Article  MathSciNet  MATH  Google Scholar 

  • Dunlap RA, Dunlap RA (1997) The golden ratio and Fibonacci numbers. World Scientific, Hackensack

    Book  MATH  Google Scholar 

  • Fleischer R, Kamphans T, Klein R, Langetepe E, Trippen G (2008) Competitive online approximation of the optimal search ratio. SIAM J Comput 38(3):881–898

    Article  MathSciNet  MATH  Google Scholar 

  • Fredman ML, Tarjan RE (1987) Fibonacci heaps and their uses in improved network optimization algorithms. J ACM (JACM) 34(3):596–615

    Article  MathSciNet  Google Scholar 

  • Hamacher HW, Tjandra SA (2002) Mathematical modeling of evacuation problems: a state of the art. Pedestrian and evacuation dynamics. Springer, Berlin, pp 227–266

    Google Scholar 

  • Jiang S, Ford D, Mullen J (2015) Massive blasts rock Chinese city of Tianjin; 44 dead, hundreds injured. The News in CNN, http://edition.cnn.com/2015/08/12/asia/china-port-explosion/

  • Lim GJ, Rungta M, Baharnemati MR (2015) Reliability analysis of evacuation routes under capacity uncertainty of road links. IIE Trans 47(1):50–63

    Article  Google Scholar 

  • Lu QS, George B, Shekhar S (2005) Capacity constrained routing algorithms for evacuation planning: a summary of results. LNCS 3633:291–307

    Google Scholar 

  • Panaite P, Pelc A (1999) Exploring unknown undirected graphs. J Algorithms 33(2):281–295

    Article  MathSciNet  MATH  Google Scholar 

  • Papadimitriou CH, Yannakakis M (1991) Shortest path without a map. Theor Comput Sci 84(1):127–150

    Article  MathSciNet  MATH  Google Scholar 

  • Shekhar S, Yang K, Gunturi VMV, Manikonda L et al (2012) Experiences with evacuation route planning algorithms. Int J Geogr Inf Sci 26(12):2253–2265

    Article  Google Scholar 

  • Wei Q, Tan XH, Jiang B, Wang LJ (2014) On-line strategies for evacuating from a convex region in the plane. LNCS in the conference COCOA2014, 74–85

  • Wei Q, Wang LJ, Jiang B (2013) Tactics for evacuating from an affected area. Int J Mach Learn Comput 3(5):435–439

    Article  Google Scholar 

  • Xu YF, Qin L (2013) Strategies of groups evacuation from a convex region in the plane. FAW-AAIM 2013. LNCS 7924:250–260

    MATH  Google Scholar 

  • Zhang HL, Xu YF, Qin L (2013) The k-Canadian Travelers Problem with communication. J Comb Optim 26(2):251–265

    Article  MathSciNet  MATH  Google Scholar 

Download references

Acknowledgments

The authors sincerely thank Ke Cao, the editor, and the anonymous reviewers for their comments and suggestions, which have improved this paper considerably. This study was partially supported by the National Natural Science Foundation of China (Grant 61221063) and the Program for Changjiang Scholars and Innovative Research Team in University (IRT1173).

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Lan Qin.

Appendix

Appendix

1.1 The proof of Theorem 3.1

Situation 2: The point P is on the sides of the square \(g_{4n+1}\).

Similar to situation 1, draw the points \(A_1\), \(B_1\), \(C_1\), \(D_1\), \(E_1\), \(F_1\), \(G_1\), \(H_1\), \(P_1'\), \(E_1'\) in the corresponding positions. The case when P is at the first point \(A_1\) on \(g_{4n+1}\) was considered in case 3 for situation 1. In addition, the case when P is on line segment \(F_1G_1\) and \(G_1B_1\) may be similar to the case where P is at the end point \(B_1\) on \(g_{4n+1}\). Therefore, we consider two cases in the following.

Case 1 P is on the line segment \(A_1F_1\).

Similar to case 2 in situation 1, we consider the ratio of the increment \({{T({A_1}{F_1}')} \over {D({H_1}{E_1}')}}\). Thus, it follows that

$$\begin{aligned} {{{h_{4n}} + {h_{4n - 2}}} \over {{h_{4n - 2}}}}= & {} 1 + {{{1 \over {\sqrt{5} }} \cdot {{{{\alpha }^4} \cdot \left( 1 - {{\alpha }^{4n}}\right) } \over {1 - {{\alpha }^4}}} - {1 \over {\sqrt{5} }} \cdot {{{{\beta }^4} \cdot \left( 1 - {{\beta }^{4n}}\right) } \over {1 - {{\beta }^4}}} + 1} \over {{1 \over {\sqrt{5} }} \cdot {{{{\alpha }^2} \cdot \left( 1 - {{\alpha }^{4n}}\right) } \over {1 - {{\alpha }^4}}} - {1 \over {\sqrt{5} }} \cdot {{{{\beta }^2} \cdot \left( 1 - {{\beta }^{4n}}\right) } \over {1 - {{\beta }^4}}}}}\\< & {} 1 + {{{1 \over {\sqrt{5} }} \cdot {{{{\alpha }^4} \cdot \left( 1 - {{\alpha }^{4n}}\right) } \over {1 - {{\alpha }^4}}} + 1} \over {{1 \over {\sqrt{5} }} \cdot {{{{\alpha }^2} \cdot \left( 1 - {{\alpha }^{4n}}\right) } \over {1 - {{\alpha }^4}}} - {1 \over {\sqrt{5} }} \cdot {{{{\beta }^2}} \over {1 - {{\beta }^4}}}}} \\= & {} 1 + {\alpha ^2} + {{{{{{\alpha }^2} \cdot {{\beta }^2}} \over {1 - {{\beta }^4}}} + \sqrt{5} } \over {{{{{\alpha }^2} \cdot \left( {{\alpha }^{4n}} - 1\right) } \over {{{\alpha }^4} - 1}} - {{{{\beta }^2}} \over {1 - {{\beta }^4}}}}}\\\le & {} 1 + {\alpha ^2} + {{{{{{\alpha }^2} \cdot {{\beta }^2}} \over {1 - {{\beta }^4}}} + \sqrt{5} } \over {{{{{\alpha }^2} \cdot \left( {{\alpha }^4} - 1\right) } \over {{{\alpha }^4} - 1}} - {{{{\beta }^2}} \over {1 - {{\beta }^4}}}}}\\< & {} 19.5. \end{aligned}$$

Therefore, \(R(FSS)<19.5\) in this case by Lemma 3.

Case 2 P is at the end point \(B_1\) on \(g_{4n+1}\).

Similarly, in this case, the competitive ratio for the FSS is

$$\begin{aligned} R(FSS) = {{T(O{B_1})} \over {D(O{C_1})}} = {{2{S_{4n + 1}} - 1} \over {{h_{4n - 2}}}}. \end{aligned}$$

Note that R(FSS) is equal to \({{175} \over 9}\) and less than 19.5 when \(n=2\). Since

$$\begin{aligned} - {{2\left( 2 - \sqrt{5} \right) } \over {\sqrt{5} }} \cdot {\beta ^{4n}} - 3 \le - {{2\left( 2 - \sqrt{5} \right) } \over {\sqrt{5} }}{\beta ^{12}} - 3< 0, \end{aligned}$$

for \(n \ge 3\), then we have

$$\begin{aligned} R(FSS)= & {} {{{{2\left( 2 + \sqrt{5} \right) } \over {\sqrt{5} }} \cdot {{\alpha }^{4n}} - {{2\left( 2 - \sqrt{5} \right) } \over {\sqrt{5} }} \cdot {{\beta }^{4n}} - 3} \over {{1 \over {\sqrt{5} }} \cdot {{{{\alpha }^2} \cdot \left( 1 - {{\alpha }^{4n}}\right) } \over {1 - {{\alpha }^4}}} - {1 \over {\sqrt{5} }} \cdot {{{{\beta }^2} \cdot \left( 1 - {{\beta }^{4n}}\right) } \over {1 - {{\beta }^4}}}}}\\\le & {} {{2\left( 2 + \sqrt{5} \right) \cdot {{\alpha }^{4n}}} \over {{{{{\alpha }^2} \cdot \left( 1 - {{\alpha }^{4n}}\right) } \over {1 - {{\alpha }^4}}} - {{{{\beta }^2} \cdot \left( 1 - {{\beta }^{4n}}\right) } \over {1 - {{\beta }^4}}}}}\\< & {} {{2\left( 2 + \sqrt{5} \right) \cdot {{\alpha }^{4n}}} \over {{{{{\alpha }^2} \cdot \left( 1 - {{\alpha }^{4n}}\right) } \over {1 - {{\alpha }^4}}} - {{{{\beta }^2}} \over {1 - {{\beta }^4}}}}}\\= & {} {{2\left( 2 + \sqrt{5} \right) } \over {{{{{\alpha }^2}} \over {1 - {{\alpha }^4}}} \cdot \left( {1 \over {{{\alpha }^{4n}}}} - 1\right) - {{{{\beta }^2}} \over {1 - {{\beta }^4}}} \cdot \left( {1 \over {{{\alpha }^{4n}}}} - 1\right) }}\\\le & {} {{2\left( 2 + \sqrt{5} \right) } \over {{{{{\alpha }^2}} \over {1 - {{\alpha }^4}}} \cdot \left( {1 \over {{{\alpha }^{12}}}} - 1\right) - {{{{\beta }^2}} \over {1 - {{\beta }^4}}} \cdot \left( {1 \over {{{\alpha }^{12}}}} - 1\right) }} \\< & {} 19.5. \end{aligned}$$

In conclusion, R(FSS) is less than 19.5 when \(n \ge 2\).

Situation 3 The point P is on the sides of the square \(g_{4n+2}\).

Similarly, we draw the points \(A_2\), \(B_2\), \(C_2\), \(D_2\), \(E_2\), \(F_2\), \(G_2\), \(H_2\), \(P_2'\), \(E_2'\) and we consider two cases for this situation in the following.

Case 1 P is on the line segment \(A_2F_2\).

We consider the ratio of increment \({{T({A_2}{F_2}')} \over {D({H_2}{E_2}')}}\), which is equal to

$$\begin{aligned} {{{h_{4n + 1}} + {h_{4n - 1}}} \over {{h_{4n - 1}}}}= & {} 1 + {{{1 \over {\sqrt{5} }} \cdot {{\alpha \cdot (1 - {{\alpha }^{4n + 1}})} \over {1 - {{\alpha }^4}}} - {1 \over {\sqrt{5} }} \cdot {{\beta \cdot \left( 1 - {{\beta }^{4n + 1}}\right) } \over {1 - {{\beta }^4}}}} \over {{1 \over {\sqrt{5} }} \cdot {{{{\alpha }^3} \cdot \left( 1 - {{\alpha }^{4n}}\right) } \over {1 - {{\alpha }^4}}} - {1 \over {\sqrt{5} }} \cdot {{{{\beta }^3} \cdot \left( 1 - {{\beta }^{4n}}\right) } \over {1 - {{\beta }^4}}}}}\\< & {} 1 + {{\left( {{\alpha }^{4n + 1}} - 1\right) - {{\beta \cdot \left( {{\alpha }^4} - 1\right) } \over {\alpha \cdot \left[ 1 - {{\beta }^4}\right] }}} \over {{{\alpha }^2} \cdot \left( {{\alpha }^{4n}} - 1\right) }}\\= & {} 1 + {\alpha ^{ - 1}} + {{\alpha - 1 - {{\beta \cdot \left( {{\alpha }^4} - 1\right) } \over {\alpha \cdot \left( 1 - {{\beta }^4}\right) }}} \over {{{\alpha }^2} \cdot \left( {{\alpha }^{4n}} - 1\right) }}\\\le & {} 1 + {\alpha ^{ - 1}} + {{(\alpha - 1) - {{\beta \cdot \left( {{\alpha }^4} - 1\right) } \over {\alpha \cdot \left( 1 - {{\beta }^4}\right) }}} \over {{{\alpha }^2} \cdot \left( {{\alpha }^4} - 1\right) }}\\< & {} 19.5. \end{aligned}$$

Thus, R(FSS) is less than 19.5 by Lemma 3.

Case 2 P is at the end point \(B_2\) on \(g_{4n+2}\).

Note that

$$\begin{aligned} R(FSS) = {{39} \over 2} = 19.5, \end{aligned}$$

when \(n=1\). For \(n \ge 2\), we obtain

$$\begin{aligned} R(FSS)= & {} {{2{S_{4n + 2}} - 1} \over {{h_{4n - 1}}}}\\= & {} {{{{2\left( 2 + \sqrt{5} \right) } \over {\sqrt{5} }} \cdot {{\alpha }^{4n + 1}} - {{2(2 - \sqrt{5} )} \over {\sqrt{5} }} \cdot {{\beta }^{4n + 1}} - 3} \over {{1 \over {\sqrt{5} }} \cdot {{{{\alpha }^3} \cdot \left( 1 - {{\alpha }^{4n}}\right) } \over {1 - {{\alpha }^4}}} - {1 \over {\sqrt{5} }} \cdot {{{{\beta }^3} \cdot \left( 1 - {{\beta }^{4n}}\right) } \over {1 - {{\beta }^4}}}}}\\< & {} {{{{2\left( 2 + \sqrt{5} \right) } \over {\sqrt{5} }} \cdot {{\alpha }^{4n + 1}}} \over {{1 \over {\sqrt{5} }} \cdot {{{{\alpha }^3} \cdot \left( 1 - {{\alpha }^{4n}}\right) } \over {1 - {{\alpha }^4}}}}}\\= & {} 2\left( 2 + \sqrt{5} \right) \cdot {\alpha ^{ - 2}} \cdot \left( {\alpha ^4} - 1\right) \cdot \left( 1 + {1 \over {{{\alpha }^{4n}} - 1}}\right) \\\le & {} 2\left( 2 + \sqrt{5} \right) \cdot {\alpha ^{ - 2}} \cdot \left( {\alpha ^4} - 1\right) \cdot \left( 1 + {1 \over {{{\alpha }^8} - 1}}\right) \\< & {} 19.5. \end{aligned}$$

Therefore, R(FSS) is no more than 19.5 when \(n \ge 1\).

Situation 4 The point P is on the sides of the square \(g_{4n+3}\).

Again, we draw the points \(A_3\), \(B_3\), \(C_3\), \(D_3\), \(E_3\), \(F_3\), \(G_3\), \(H_3\), \(P_3'\), \(E_3'\) and we consider two cases in the following.

Case 1 P is on the line segment \(A_3F_3\).

Similar to case 2 in situation 1, we consider the ratio of increment

$$\begin{aligned} {{T({A_3}{F_3}')} \over {D({H_3}{E_3}')}}= & {} {{{h_{4n}} + {h_{4n - 2}}} \over {{h_{4n - 2}}}}\\= & {} 1 + {{{1 \over {\sqrt{5} }} \cdot {{{{\alpha }^2} \cdot \left( 1 - {{\alpha }^{4n + 2}}\right) } \over {1 - {{\alpha }^4}}} - {1 \over {\sqrt{5} }} \cdot {{{{\beta }^2} \cdot \left( 1 - {{\beta }^{4n + 2}}\right) } \over {1 - {{\beta }^4}}}} \over {{1 \over {\sqrt{5} }} \cdot {{{{\alpha }^4} \cdot \left( 1 - {{\alpha }^{4n}}\right) } \over {1 - {{\alpha }^4}}} - {1 \over {\sqrt{5} }} \cdot {{{{\beta }^4} \cdot \left( 1 - {{\beta }^{4n}}\right) } \over {1 - {{\beta }^4}}} + 1}}. \end{aligned}$$

Note that \({ - {1 \over {\sqrt{5} }} \cdot {{{{\beta }^4} \cdot (1 - {{\beta }^{4n}})} \over {1 - {{\beta }^4}}} + 1}\) is a positive value, as proved above, and the formula is

$$\begin{aligned} R(increment)< & {} 1 + {{{1 \over {\sqrt{5} }} \cdot {{{{\alpha }^2} \cdot \left( 1 - {{\alpha }^{4n + 2}}\right) } \over {1 - {{\alpha }^4}}}} \over {{1 \over {\sqrt{5} }} \cdot {{{{\alpha }^4} \cdot \left( 1 - {{\alpha }^{4n}}\right) } \over {1 - {{\alpha }^4}}}}}\\= & {} 2 + {{{{\alpha }^2} - 1} \over {{{\alpha }^2} \cdot \left( {{\alpha }^{4n}} - 1\right) }} \le 2 + {{{{\alpha }^2} - 1} \over {{{\alpha }^2} \cdot \left( {{\alpha }^4} - 1\right) }}\\< & {} 19.5. \end{aligned}$$

Therefore, we find that R(FSS) is less than 19.5 by Lemma 3.

Case 2 P is at the end point \(B_3\) on \(g_{4n+3}\).

In this case, R(FFS) is equal to 16.25 when \(n=1\). When \(n \ge 2\), the ratio is

$$\begin{aligned} R(FSS)= & {} {{2{S_{4n + 3}} - 1} \over {{h_{4n}}}} \\= & {} {{{{2\left( 2 + \sqrt{5} \right) } \over {\sqrt{5} }} \cdot {{\alpha }^{4n + 2}} - {{2\left( 2 - \sqrt{5} \right) } \over {\sqrt{5} }} \cdot {{\beta }^{4n + 2}} - 3} \over {{1 \over {\sqrt{5} }} \cdot {{{{\alpha }^4} \cdot \left( 1 - {{\alpha }^{4n}}\right) } \over {1 - {{\alpha }^4}}} - {1 \over {\sqrt{5} }} \cdot {{{{\beta }^4} \cdot \left( 1 - {{\beta }^{4n}}\right) } \over {1 - {{\beta }^4}}} + 1}}. \end{aligned}$$

Note that for \(n \ge 2\),

$$\begin{aligned} - {{2\left( 2 - \sqrt{5} \right) } \over {\sqrt{5} }} \cdot {\beta ^{4n + 2}} - 3 \le - {{2\left( 2 - \sqrt{5} \right) } \over {\sqrt{5} }} \cdot {\beta ^{10}} - 3 < 0, \end{aligned}$$

and

$$\begin{aligned} - {1 \over {\sqrt{5} }} \cdot {{{{\beta }^4} \cdot \left( 1 - {{\beta }^{4n}}\right) } \over {1 - {{\beta }^4}}} + 1 > - {1 \over {\sqrt{5} }} \cdot {{{{\beta }^4}} \over {1 - {{\beta }^4}}} + 1 = 0.9236 > 0. \end{aligned}$$

Thus, the solution to the formula above is less than

$$\begin{aligned} R(FSS)= & {} {{{{2\left( 2 + \sqrt{5} \right) } \over {\sqrt{5} }} \cdot {{\alpha }^{4n + 2}}} \over {{1 \over {\sqrt{5} }} \cdot {{{{\alpha }^4} \cdot \left( 1 - {{\alpha }^{4n}}\right) } \over {1 - {{\alpha }^4}}}}}\\= & {} 2\left( 2 + \sqrt{5} \right) \cdot {\alpha ^{ - 2}} \cdot \left( {\alpha ^4} - 1\right) \cdot \left( 1 + {1 \over {{{\alpha }^{4n}} - 1}}\right) \\< & {} 19.5 \end{aligned}$$

Therefore, R(FSS) is less than 19.5 when \(n \ge 1\).

In summary, the competitive ratio for FSS is no more than 19.5 in this situation.

The Theorem follows.

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Qin, L., Xu, Y. Fibonacci helps to evacuate from a convex region in a grid network. J Comb Optim 34, 398–413 (2017). https://doi.org/10.1007/s10878-016-9998-7

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s10878-016-9998-7

Keywords

Navigation