Skip to main content
Log in

A lower bound for online rectangle packing

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

Abstract

We slightly improve the known lower bound on the asymptotic competitive ratio for online bin packing of rectangles. We present a complete proof for the new lower bound, whose value is above 1.91.

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

Similar content being viewed by others

References

  • Balogh J, Békési J, Galambos G (2012) New lower bounds for certain classes of bin packing algorithms. Theor Comput Sci 440:1–13

    Article  MathSciNet  MATH  Google Scholar 

  • Balogh J, Békési J, Dósa Gy, Galambos G, Tan Z (2016) Lower bound for 3-batched bin packing. Discrete Optim 21:14–24

    Article  MathSciNet  MATH  Google Scholar 

  • Balogh J, Békési J, Dósa Gy, Epstein L, Levin A (2017) Online bin packing with cardinality constraints resolved. In: Proceedings of the 25th European symposium on algorithms (ESA2017), pp 10:1–10:14

  • Balogh J, Békési J, Dósa Gy, Epstein L, Levin A (2018a) A new and improved algorithm for online bin packing. In: Proceedings of the 26th European symposium on algorithms (ESA2018), pp 5:1–5:14

  • Balogh J, Békési J, Dósa Gy, Epstein L, Levin A (2018b) A new lower bound for classic online bin packing. CoRR, arXiv:1807.05554

  • Balogh J, Békési J, Dósa Gy, Epstein L, Levin A (2019) Lower bounds for several online variants of bin packing. Theory Comput Syst. https://doi.org/10.1007/s00224-019-09915-1

  • Békési J, Dósa Gy, Epstein L (2016) Bounds for online bin packing with cardinality constraints. Inf Comput 249:190–204

    Article  MathSciNet  MATH  Google Scholar 

  • Blitz D (1996) Lower bounds on the asymptotic worst-case ratios of on-line bin packing algorithms. Master’s thesis, University of Rotterdam, Rotterdam, The Netherlands

  • Blitz D, van Vliet A, Woeginger GJ (1996) Lower bounds on the asymptotic worst-case ratio of online bin packing algorithms. Unpublished manuscript

  • Blitz D, Heydrich S, van Stee R, van Vliet A, Woeginger GJ (2016) Improved lower bounds for online hypercube and rectangle packing. CoRR, arXiv:1607.01229

  • Coppersmith D, Raghavan P (1989) Multidimensional on-line bin packing: algorithms and worst-case analysis. Oper Res Lett 8(1):17–20

    Article  MathSciNet  MATH  Google Scholar 

  • Csirik J, van Vliet A (1993) An on-line algorithm for multidimensional bin packing. Oper Res Lett 13(3):149–158

    Article  MathSciNet  MATH  Google Scholar 

  • Csirik J, Frenk JBG, Labbe M (1993) Two-dimensional rectangle packing: on-line methods and results. Discrete Appl Math 45(3):197–204

    Article  MathSciNet  MATH  Google Scholar 

  • Epstein L (2010) Two-dimensional online bin packing with rotation. Theor Comput Sci 411(31–33):2899–2911

    Article  MathSciNet  MATH  Google Scholar 

  • Epstein L, van Stee R (2005a) Online square and cube packing. Acta Inform 41(9):595–606

    Article  MathSciNet  MATH  Google Scholar 

  • Epstein L, van Stee R (2005b) Optimal online algorithms for multidimensional packing problems. SIAM J Comput 35(2):431–448

    Article  MathSciNet  MATH  Google Scholar 

  • Fujita S, Hada T (2002) Two-dimensional on-line bin packing problem with rotatable items. Theor Comput Sci 289(2):939–952

    Article  MathSciNet  MATH  Google Scholar 

  • Fujiwara H, Kobayashi KM (2015) Improved lower bounds for the online bin packing problem with cardinality constraints. J Comb Optim 29(1):67–87

    Article  MathSciNet  MATH  Google Scholar 

  • Galambos G (1991) A 1.6 lower-bound for the two-dimensional on-line rectangle bin-packing. Acta Cybern 10(1–2):21–24

    MathSciNet  MATH  Google Scholar 

  • Galambos G, van Vliet A (1994) Lower bounds for 1-, 2- and 3-dimensional on-line bin packing algorithms. Computing 52(3):281–297

    Article  MathSciNet  MATH  Google Scholar 

  • Han X, Ye D, Zhou Y (2010) A note on online hypercube packing. Cent Eur J Oper Res 18(2):221–239

    Article  MathSciNet  MATH  Google Scholar 

  • Han X, Chin FY, Ting H-F, Zhang G, Zhang Y (2011) A new upper bound 2.5545 on 2D online bin packing. ACM Trans Algorithms 7(4): Article 50

  • Johnson DS, Demers A, Ullman JD, Garey MR, Graham RL (1974) Worst-case performance bounds for simple one-dimensional packing algorithms. SIAM J Comput 3:256–278

    Article  MathSciNet  MATH  Google Scholar 

  • Seiden SS, van Stee R (2003) New bounds for multidimensional packing. Algorithmica 36(3):261–293

    Article  MathSciNet  MATH  Google Scholar 

  • van Vliet A (1995) Lower and upper bounds for online bin packing and scheduling heuristics. Ph.D. thesis, Erasmus University, Rotterdam, The Netherlands

  • van Vliet A (1992) An improved lower bound for online bin packing algorithms. Inf Process Lett 43(5):277–284

    Article  MATH  Google Scholar 

  • Yang J, Leung JY-T (2003) The ordered open-end bin-packing problem. Oper Res 51(5):759–770

    Article  MathSciNet  MATH  Google Scholar 

Download references

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Leah Epstein.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Leah Epstein: Partially supported by a grant from GIF—the German-Israeli Foundation for Scientific Research and Development (Grant Number I-1366-407.6/2016).

Appendix A: Proofs of technical lemmas

Appendix A: Proofs of technical lemmas

1.1 Appendix A.1: Proof of Lemma 4.1

The first part holds since \(\varepsilon < \frac{1}{7224}\), and the second part holds by definition.

Consider the third part. For \(1 \le t \le k-2\), we have

$$\begin{aligned} \sum _{i=1}^{t} w_{1i}= & {} (1+\delta )\sum _{i=1}^t \frac{1}{5^{k-i-1}}= (1+\delta )\frac{1}{5^{k-t-1}}\sum _{i=1}^t \frac{1}{5^{t-i}}= (1+\delta )\frac{1}{5^{k-t-1}}\sum _{j=0}^{t-1} \frac{1}{5^{j}}\\= & {} (1+\delta )\frac{1}{5^{k-t-1}}\frac{1-\frac{1}{5^t}}{1-\frac{1}{5}}=(1+\delta )\frac{1}{4 \cdot 5^{k-t-2}}\left( 1-\frac{1}{5^t}\right) \\= & {} (1+\delta )\left( \frac{1}{4 \cdot 5^{k-t-2}}-\frac{1}{4 \cdot 5^{k-2}}\right) <\frac{1-2^{42}\delta }{4 \cdot 5^{k-t-2}} , \end{aligned}$$

by \(2^{42}+1<2^{43}\) and since

$$\begin{aligned} \frac{2^{43}\delta }{4 \cdot 5^{k-t-2}} < \frac{1+\delta }{4 \cdot 5^{k-2}} \end{aligned}$$

holds by \(t \le k-2\) and

$$\begin{aligned} \frac{4 \cdot 5^{k-2}}{4 \cdot 5^{k-t-2}} = 5^{t} \le 5^{k-2} \text{ while } \frac{1+\delta }{2^{43}\delta }> \frac{1}{2^{43}\delta }> 2^{3k+7} > 8^k . \end{aligned}$$

In particular for \(t=k-2\), the total width is below \(\frac{1}{4}\). Thus, we also have

$$\begin{aligned} \sum _{i=1}^{k-1} w_{1i}< \frac{1-2^{42}\delta }{4} + \frac{1+2^{40}\delta }{4}< \frac{1}{2} \text{ and } \sum _{i=1}^{k} w_{1i}< \frac{1-2^{42}\delta }{4} + 3 \cdot \frac{1+2^{40}\delta }{4} < 1 . \end{aligned}$$

The fourth and fifth parts hold by definition. The sixth part holds by definition and since \(3 \cdot w_{20} + w_{1(k-1)} = 3\cdot (\frac{1}{4}-2^{32}\delta )+\frac{1+2^{40}\delta }{4} > 1 - 2^{34}\delta +2^{38}\delta \). The seventh part holds by definition and since \( 3 w_{(j+1)0}+ w_{j1} = 3(\frac{1}{4}-2^{52-10j}\delta )+(\frac{1}{4}+2^{60-10j}\delta ) >1\). \(\square \)

1.2 Appendix A.2: Proof of Lemma 4.4

  1. 1.

    For every \(j \in \{2,3,4\}\), the type \(\ell _{j0}\) dominates \(\ell _{j1}\) since the width of the former type is smaller, and their heights and weights are equal. Type \(\ell _{j1}\) dominates type \(\ell _{j2}\) since the width of the former is twice as small, their heights are equal, and the weight ratio satisfies \(v_{j2}/v_{j1}=2\).

  2. 2.

    For \(1 \le i \le k-3\), type \(\ell _{1i}\) dominates type \(\ell _{1(i+1)}\) as their heights are equal, and \(\frac{w_{1(i+1)}}{w_{1i}}=\frac{v_{1(i+1)}}{v_{1i}}=5\). Type \(\ell _{1(k-2)}\) dominated type \(\ell _{1(k-1)}\) since their heights are equal, \(w_{1(k-2)}<w_{1(k-1)}\) and \(v_{1(k-2)}=v_{1(k-1)}\). Type \(\ell _{1(k-1)}\) dominated type \(\ell _{1k}\) since their heights are equal, \(2 \cdot w_{1(k-1)} = w_{1(k-1)}\) and \(2 \cdot v_{1(k-1)}=v_{1k}\).

  3. 3.

    The height of item type \(\ell _{1(k-2)}\) is \(\frac{1}{43}+\varepsilon \), and the height of item type \(\ell _{20}\) is \(\frac{1}{7}+\varepsilon \). We have \(6(\frac{1}{43}+\varepsilon )<\frac{1}{7}+\varepsilon \) as \(\varepsilon <0.0001\). The width of item type \(\ell _{1(k-2)}\) is \(\frac{1+\delta }{5}\), and the width of item type \(\ell _{20}\) is \(\frac{1}{4}-2^{32}\delta \). We have \(\frac{1+\delta }{5}<\frac{1}{4}-2^{32}\delta \) as \(2^{64}\delta <1\). As the weight of six items of type \(\ell _{1(k-2)}\) is 6, while the weight of one item of type \(\ell _{20}\) is 4, the domination holds.

  4. 4.

    Item type \(\ell _{20}\) has height \(\frac{1}{7}+\varepsilon \) while item type \(\ell _{30}\) has height \(\frac{1}{3}+\varepsilon \), and we have \(2(\frac{1}{7}+\varepsilon )<\frac{1}{3}+\varepsilon \), as \(\varepsilon < 0.0001\). Item type \(\ell _{20}\) has smaller width than item type \(\ell _{30}\). The weight of two items of type \(\ell _{20}\) is 8 while the weight of one type \(\ell _{30}\) item is 6. Thus, the domination holds.

  5. 5.

    Item type \(\ell _{30}\) has both smaller height and smaller width than an item of type \(\ell _{40}\) and they have the same weights. Thus, the domination holds.

\(\square \)

1.3 Appendix A.3: Proof of Lemma 5.3

We start with the numerator. We use

$$\begin{aligned} \sum _{i=1}^{k-2} v_{1i}= \sum _{i=1}^{k-2}\frac{1}{5^{k-i-2}} =\sum _{j=0}^{k-3} \frac{1}{5^{j}}=\frac{1-\frac{1}{5^{k-2}}}{4/5}=1.25-\frac{1}{4\cdot 5^{k-1}} , \end{aligned}$$

and \(\sum _{i=k-1}^k v_{1i}+\sum _{j=2}^4 \sum _{i=0}^2 v_{ji}=67\), to get

$$\begin{aligned} \sum _{i=1}^k v_{1i}+\sum _{j=2}^4 \sum _{i=0}^2 v_{ji}=68.25-\frac{1}{4\cdot 5^{k-1}} . \end{aligned}$$

Next, we consider the denominator. We have

$$\begin{aligned} Q\le & {} \frac{1}{168} \cdot \left( 42\cdot \left( 5-\frac{1}{5^{k-3}}\right) /5^{k-3}+\sum _{i=2}^{k-2}42\cdot \left( 5-\frac{1}{5^{k-i-2}}\right) \right. \cdot \left( \frac{1}{5^{k-i-2}}-\frac{1}{5^{k-(i-1)-2}}\right) \\&+\,1\cdot 126+2\cdot 112+6\cdot 96\\&\left. \,+6\cdot 72+12\cdot 68+14\cdot 48+14\cdot 42+28\cdot 36+21\cdot 24+21\cdot 18+42\cdot 12\right) . \end{aligned}$$

Since

$$\begin{aligned}&\sum _{i=2}^{k-2}\left( 5-\frac{1}{5^{k-i-2}}\right) \cdot \left( \frac{1}{5^{k-i-2}}-\frac{1}{5^{k-(i-1)-2}}\right) =\sum _{i=2}^{k-2}\left( 5-\frac{1}{5^{k-i-2}}\right) \cdot \frac{4}{5^{k-i-1}}\\&\quad =4\cdot \sum _{i=2}^{k-2}\frac{1}{5^{k-i-2}}-0.8 \cdot \sum _{i=2}^{k-2}\frac{1}{25^{k-i-2}}=4\cdot \sum _{u=0}^{k-4}\frac{1}{5^{u}}-0.8\cdot \sum _{u=0}^{k-4}\frac{1}{25^{u}}\\&\quad =4\cdot \left( \frac{1-(1/5)^{k-3}}{0.8}\right) -0.8\cdot \left( \frac{1-(1/25)^{k-3}}{0.96}\right) =\frac{25}{6}-\frac{1}{5^{k-4}}+ \frac{1}{6 \cdot 5^{2k-7}} \ \end{aligned}$$

we have,

$$\begin{aligned} Q\le & {} \frac{1}{168} \cdot \left( 42/5^{k-4}-42/5^{2k-6}+42\left( \frac{25}{6}-\frac{1}{5^{k-4}}+ \frac{1}{6 \cdot 5^{2k-7}}\right) +5828\right) \\= & {} \frac{6003-\frac{7}{5^{2k-6}}}{168} . \end{aligned}$$

\(\square \)

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Epstein, L. A lower bound for online rectangle packing. J Comb Optim 38, 846–866 (2019). https://doi.org/10.1007/s10878-019-00423-z

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s10878-019-00423-z

Keywords

Navigation