Download Full Version ebook - Visit ebookmeta.

com

Algorithmic Aspects in Information and Management:

16th International Conference, AAIM 2022,
Guangzhou, China, August 13–14, 2022, Proceedings
(Lecture Notes in Computer Science, 13513) Qiufen
Ni
https://ebookmeta.com/product/algorithmic-aspects-in-
information-and-management-16th-international-conference-
aaim-2022-guangzhou-china-august-13-14-2022-proceedings-
lecture-notes-in-computer-science-13513-qiufen/

OR CLICK HERE

DOWLOAD NOW

Discover More Ebook - Explore Now at ebookmeta.com

 Instant digital products (PDF, ePub, MOBI) ready for you
Download now and discover formats that fit your needs...

Start reading on any device today!

Cambridge IGCSE and O Level History Workbook 2C - Depth

Study: the United States, 1919-41 2nd Edition Benjamin
Harrison
https://ebookmeta.com/product/cambridge-igcse-and-o-level-history-
workbook-2c-depth-study-the-united-states-1919-41-2nd-edition-
benjamin-harrison/
ebookmeta.com

Computer Information Systems and Industrial Management:

21st International Conference, CISIM 2022, Barranquilla,
Colombia, July 15–17, 2022, Proceedings (Lecture Notes in
Computer Science, 13293) Khalid Saeed (Editor)
https://ebookmeta.com/product/computer-information-systems-and-
industrial-management-21st-international-conference-
cisim-2022-barranquilla-colombia-july-15-17-2022-proceedings-lecture-
notes-in-computer-science-132/
ebookmeta.com

Parallel Computational Technologies 16th International

Conference PCT 2022 Dubna Russia March 29 31 2022 Revised
Selected Papers Communications in Computer and Information
Science 1618 Leonid Sokolinsky
https://ebookmeta.com/product/parallel-computational-
technologies-16th-international-conference-pct-2022-dubna-russia-
march-29-31-2022-revised-selected-papers-communications-in-computer-
and-information-science-1618-leonid-sokolin/
ebookmeta.com

Double Agent Celery MI5 s Crooked Hero Carolinda Witt

Foreword By Nigel West

https://ebookmeta.com/product/double-agent-celery-mi5-s-crooked-hero-
carolinda-witt-foreword-by-nigel-west/

ebookmeta.com
The Dark King 1st Edition Gina L. Maxwell

https://ebookmeta.com/product/the-dark-king-1st-edition-gina-l-
maxwell/

ebookmeta.com

SafeHaven: The Giseria Chronicles Book 2 1st Edition

Xander Jade

https://ebookmeta.com/product/safehaven-the-giseria-chronicles-
book-2-1st-edition-xander-jade/

ebookmeta.com

Give Me Everything A High School Christmas Holidays Sweet

and Steamy Romance Dirty Elite Academy 1st Edition Kate
J. Blake
https://ebookmeta.com/product/give-me-everything-a-high-school-
christmas-holidays-sweet-and-steamy-romance-dirty-elite-academy-1st-
edition-kate-j-blake/
ebookmeta.com

Bernstein s Construction of Movements The Original Text

and Commentaries 1st Edition Nikola■ Aleksandrovich
Bernshte■n
https://ebookmeta.com/product/bernstein-s-construction-of-movements-
the-original-text-and-commentaries-1st-edition-nikolai-aleksandrovich-
bernshtein/
ebookmeta.com

Improving Speech Intelligibility in Adults Clinical

Application of Evidence Based Strategies 1st Edition
Connie K Porcaro
https://ebookmeta.com/product/improving-speech-intelligibility-in-
adults-clinical-application-of-evidence-based-strategies-1st-edition-
connie-k-porcaro-2/
ebookmeta.com
Experiencing the Landscape in Antiquity 2 Armando
Cristilli Editor Fabio De Luca Editor Gioconda Di Luca
Editor
https://ebookmeta.com/product/experiencing-the-landscape-in-
antiquity-2-armando-cristilli-editor-fabio-de-luca-editor-gioconda-di-
luca-editor/
ebookmeta.com
 Qiufen Ni
Weili Wu (Eds.)

Algorithmic Aspects
LNCS 13513

in Information
and Management
16th International Conference, AAIM 2022
Guangzhou, China, August 13–14, 2022
Proceedings
Lecture Notes in Computer Science 13513

Founding Editors
Gerhard Goos
Karlsruhe Institute of Technology, Karlsruhe, Germany
Juris Hartmanis
Cornell University, Ithaca, NY, USA

Editorial Board Members

Elisa Bertino
Purdue University, West Lafayette, IN, USA
Wen Gao
Peking University, Beijing, China
Bernhard Steffen
TU Dortmund University, Dortmund, Germany
Moti Yung
Columbia University, New York, NY, USA
More information about this series at https://link.springer.com/bookseries/558
Qiufen Ni · Weili Wu (Eds.)

Algorithmic Aspects
in Information
and Management
16th International Conference, AAIM 2022
Guangzhou, China, August 13–14, 2022
Proceedings
Editors
Qiufen Ni Weili Wu
Guangdong University of Technology University of Texas at Dallas
Guangzhou, China Richardson, TX, USA

ISSN 0302-9743 ISSN 1611-3349 (electronic)

Lecture Notes in Computer Science
ISBN 978-3-031-16080-6 ISBN 978-3-031-16081-3 (eBook)
https://doi.org/10.1007/978-3-031-16081-3

© The Editor(s) (if applicable) and The Author(s), under exclusive license
to Springer Nature Switzerland AG 2022
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the
material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now
known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this book are
believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors
give a warranty, expressed or implied, with respect to the material contained herein or for any errors or
omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
 Preface

The 16th International Conference on Algorithmic Aspects in Information and

Management (AAIM 2022), took place at Guangzhou, China, August 13–14, 2022.
The conference was held virtually due to the COVID-19 pandemic.
The AAIM conference series, which started in 2005 in Xi’an, China, aims to stimu-
late various fields for which algorithmics has become a crucial enabler, and to strengthen
the ties of various research communities of algorithmics and applications. AAIM 2022
seeks to address emerging and important algorithmic problems by focusing on the funda-
mental background, theoretical technological development, and real-world applications
associated with information and management analysis, modeling and data mining. Spe-
cial considerations are given to algorithmic research that was motivated by real-world
applications. We received 59 submissions, out of which 41 papers were accepted for
publication. Each submission was reviewed by at least three reviewers.
We would like to thank the four keynote speakers, Hui Xiong (IEEE fellow,
a Distinguished Professor at Rutgers University, USA and a Distinguished Guest
Professor at the University of Science and Technology of China), Kui Ren (ACM
Fellow, IEEE Fellow, a Professor at Zhejiang University), Cong Tian (a Professor
at Xidian University), and Xiaoming Sun (Professor at the Institute of Computing
Technology, Chinese Academy of Sciences) for their contributions to the conference.
We would like to express our appreciation to all members of the Program Committee
and the external referees whose efforts enabled us to achieve a high scientific standard
for the proceedings. We would also like to thank all members of the Organizing
Committee for their assistance and contribution which attributed to the success of
the conference. Particularly, we would like to thank Anna Kramer and her colleagues
at Springer for meticulously supporting us in the timely production of this volume.
Last but not least, our special thanks go to all the authors and participants for their
contributions to the success of this event.

July 2022 Qiufen Ni

Weili Wu
 Organization

Program Committee Chairs

Qiufen Ni Guangdong University of Technology, China
Weili Wu University of Texas at Dallas, USA

Program Committee Members

Wolfgang Bein University of Nevada, USA
Gruia Calinescu Illinois Institute of Technology, USA
Xujin Chen Academy of Mathematics and Systems Science,
Chinese Academy of Sciences, China
Zhizhong Chen Tokyo Denki University, Japan
Bhaskar DasGupta University of Illinois at Chicago, USA
Xingjian Ding Beijing University of Technology, Faculty of
Information Technology, China
Hongwei Du Harbin Institute of Technology (Shenzhen), China
Rudolf Fleischer Heinrich Heine University Dusseldorf, Germany
Shuyang Gu Texas A&M University - Central Texas, USA
Jianxiong Guo Beijing Normal University, Advanced Institute of
Natural Sciences, China
Sun-Yuan Hsieh National Cheng Kung University, China
Liying Kang Shanghai University, China
Michael Khachay Krasovsky Institute of Mathematics and
Mechanics, Russia
Chia-Wei Lee National Taitung University, Computer Science
and Information Engineering, Taitung, China
Xianyue Li Lanzhou University, Lanzhou, China
Xiao Li University of Texas at Dallas, USA
Shengxin Liu Harbin Institute of Technology (Shenzhen), China
Chuanwen Luo Beijing Forestry University, Beijing, China
Viet Hung Nguyen University of Clermont-Auvergne, France
Ghosh Smita Santa Clara University, USA
Zhiyi Tan Zhejiang University, China
Weitian Tong Georgia Southern University, USA
Weili Wu University of Texas at Dallas, USA
Yicheng Xu Shenzhen Institute of Advanced Technology,
Chinese Academy of Sciences, China
Boting Yang University of Regina, Canada
viii Organization

Ruiqi Yang Beijing University of Technology, Beijing

Institute for Scientific and Engineering
Computing, China
Nan Zhang Xidian University, China
Yapu Zhang Beijing University of Technology, China
Yong Zhang Shenzhen Institute of Advanced Technology,
Chinese Academy of Sciences, China
Fay Zhong California State University, China
Yuqing Zhu California State University, Los Angeles, USA

Reviewers
Wolfgang Bein University of Nevada, USA
Gruia Calinescu Illinois Institute of Technology, USA
Xujin Chen Academy of Mathematics and Systems Science,
Chinese Academy of Sciences, China
Sijia Dai Shenzhen Institute of Advanced Technology,
Chinese Academy of Sciences, China
Bhaskar DasGupta University of Illinois at Chicago, USA
Xingjian Ding Beijing University of Technology, Faculty of
Information Technology, China
Dingzhu Du University of Texas at Dallas, USA
Hongwei Du Harbin Institute of Technology (Shenzhen), China
Liman Du University of Chinese Academy of Sciences,
China
Guichen Gao Shenzhen Institute of Advanced Technology,
Chinese Academy of Sciences, China
Yuping Gao Lanzhou University, China
Shuyang Gu Texas A&M University - Central Texas, USA
Jianxiong Guo Beijing Normal University, Advanced Institute of
Natural Sciences, China
Xinxin Han Shenzhen Institute of Advanced Technology,
Chinese Academy of Sciences, China
Yi Hong Beijing Forestry University, China
Sun-Yuan Hsieh National Cheng Kung University, China
Liying Kang Shanghai University, China
Michael Khachay Krasovsky Institute of Mathematics and
Mechanics, Russia
Chia-Wei Lee National Taitung University, Computer Science
and Information Engineering, China
Xianyue Li Lanzhou University, China
Xiao Li University of Texas at Dallas, USA
Shengxin Liu Harbin Institute of Technology (Shenzhen), China
Chuanwen Luo Beijing Forestry University, China
 Organization ix

Viet Hung Nguyen University of Clermont-Auvergne, France

Qiufen Ni Guangdong University of Technology, China
Ghosh Smita Santa Clara University, USA
Shaojie Tang University of Texas at Dallas, USA
Weitian Tong Georgia Southern University, USA
Yongcai Wang Renmin University of China, China
Weili Wu University of Texas at Dallas, USA
Yicheng Xu Shenzhen Institute of Advanced Technology,
Chinese Academy of Sciences, China
Boting Yang University of Regina, Canada
Ruiqi Yang Beijing University of Technology, Beijing
Institute for Scientific and Engineering
Computing, China
Wenguo Yang University of Chinese Academy of Sciences,
China
Nan Zhang Xidian University, China
Yapu Zhang Beijing University of Technology, China
Yong Zhang Shenzhen Institute of Advanced Technology,
Chinese Academy of Sciences, Shenzhen,
China
Fay Zhong California State University, USA
Yuqing Zhu California State University, Los Angeles, USA
 Contents

Approximation Algorithms

A Binary Search Double Greedy Algorithm for Non-monotone

DR-submodular Maximization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Shuyang Gu, Chuangen Gao, and Weili Wu

An Approximation Algorithm for the Clustered Path Travelling Salesman

Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
Jiaxuan Zhang, Suogang Gao, Bo Hou, and Wen Liu

Improved Approximation Algorithm for the Asymmetric Prize-Collecting

TSP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
Bo Hou, Zhenzhen Pang, Suogang Gao, and Wen Liu

Scheduling Problem and Game Theory

Approximation Scheme for Single-Machine Rescheduling with Job Delay

and Rejection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Ruiqing Sun and Xiaofei Liu

Online Early Work Maximization Problem on Two Hierarchical Machines

with Buffer or Rearrangements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
Man Xiao, Xihua Bai, and Weidong Li

On-line Single Machine Scheduling with Release Dates and Submodular

Rejection Penalties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Xiaofei Liu, Yaoyu Zhu, Weidong Li, and Lei Ma

The Optimal Dynamic Rationing Policy in the Stock-Rationing Queue . . . . . . . . 66

Quan-Lin Li, Yi-Meng Li, Jing-Yu Ma, and Heng-Li Liu

The Constrained Parallel-Machine Scheduling Problem with Divisible

Processing Times and Penalties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Jianping Li, Runtao Xie, Junran Lichen, Guojun Hu, Pengxiang Pan,
and Ping Yang

Obnoxious Facility Location Games with Candidate Locations . . . . . . . . . . . . . . . 96

Ling Gai, Mengpei Liang, and Chenhao Wang
xii Contents

Nonlinear Combinatorial Optimization

Streaming Adaptive Submodular Maximization . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109

Shaojie Tang and Jing Yuan

Constrained Stochastic Submodular Maximization with State-Dependent

Costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
Shaojie Tang

Bicriteria Algorithms for Maximizing the Difference Between Submodular

Function and Linear Function Under Noise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
Mengxue Geng, Shufang Gong, Bin Liu, and Weili Wu

Monotone k-Submodular Knapsack Maximization: An Analysis

of the Greedy+Singleton Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
Jingwen Chen, Zhongzheng Tang, and Chenhao Wang

Guarantees for Maximization of k-Submodular Functions with a Knapsack

and a Matroid Constraint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
Kemin Yu, Min Li, Yang Zhou, and Qian Liu

Network Problems

Defense of Scapegoating Attack in Network Tomography . . . . . . . . . . . . . . . . . . . 171

Xiaojia Xu, Yongcai Wang, Yu Zhang, and Deying Li

Adaptive Competition-Based Diversified-Profit Maximization with Online

Seed Allocation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
Liman Du, Wenguo Yang, and Suixiang Gao

Collaborative Service Caching in Mobile Edge Nodes . . . . . . . . . . . . . . . . . . . . . . 195

Zichen Wang and Hongwei Du

A Decentralized Auction Framework with Privacy Protection in Mobile

Crowdsourcing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
Jianxiong Guo, Qiufen Ni, and Xingjian Ding

Profit Maximization for Multiple Products in Community-Based Social

Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
Qiufen Ni and Jianxiong Guo

MCM: A Robust Map Matching Method by Tracking Multiple Road

Candidates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
Wanting Li, Yongcai Wang, Deying Li, and Xiaojia Xu
 Contents xiii

Pilot Pattern Design with Branch and Bound in PSA-OFDM System . . . . . . . . . . 244
Shuchen Wang, Suixiang Gao, and Wenguo Yang

AoI Minimizing of Wireless Rechargeable Sensor Network Based

on Trajectory Optimization of Laser-Charged UAV . . . . . . . . . . . . . . . . . . . . . . . . . 255
Chuanwen Luo, Yunan Hou, Yi Hong, Zhibo Chen, Ning Liu,
and Deying Li

Energy-Constrained Geometric Coverage Problem . . . . . . . . . . . . . . . . . . . . . . . . . 268

Huan Lan

Incremental SDN Deployment to Achieve Load Balance in ISP Networks . . . . . 278

Yunlong Cheng, Hao Zhou, Xiaofeng Gao, Jiaqi Zheng, and Guihai Chen

Graph Theory

Polynomial Time Algorithm for k-vertex-edge Dominating Problem

in Interval Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293
Peng Li and Aifa Wang

Cyclically Orderable Generalized Petersen Graphs . . . . . . . . . . . . . . . . . . . . . . . . . 303

Xiaofeng Gu and William Zhang

The r-Dynamic Chromatic Number of Planar Graphs Without Special

Short Cycles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316
Yuehua Bu, Ruiying Yang, and Hongguo Zhu

Distance Magic Labeling of the Halved Folded n-Cube . . . . . . . . . . . . . . . . . . . . . 327

Yi Tian, Na Kang, Weili Wu, Ding-Zhu Du, and Suogang Gao

Balanced Graph Partitioning Based on Mixed 0-1 Linear Programming

and Iteration Vertex Relocation Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339
Zhengxi Yang, Zhipeng Jiang, Wenguo Yang, and Suixiang Gao

Partial Inverse Min-Max Spanning Tree Problem Under the Weighted

Bottleneck Hamming Distance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
Qingzhen Dong, Xianyue Li, and Yu Yang

Mixed Metric Dimension of Some Plane Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . 363

Na Kang, Zhiquan Li, Lihang Hou, and Jing Qu

On the Transversal Number of k-Uniform Connected Hypergraphs . . . . . . . . . . . 376

Zian Chen, Bin Chen, Zhongzheng Tang, and Zhuo Diao
xiv Contents

An Improvement of the Bound on the Odd Chromatic Number of 1-Planar

Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 388
Bei Niu and Xin Zhang

Fast Searching on k-Combinable Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 394

Yuan Xue, Boting Yang, and Sandra Zilles

Class Ramsey Numbers Involving Induced Graphs . . . . . . . . . . . . . . . . . . . . . . . . . 406

Yan Li and Ye Wang

Injective Edge Coloring of Power Graphs and Necklaces . . . . . . . . . . . . . . . . . . . . 413

Yuehua Bu, Wenwen Chen, and Junlei Zhu

Total Coloring of Planar Graphs Without Some Adjacent Cycles . . . . . . . . . . . . . 421

Liting Wang and Huijuan Wang

Logic and Machine Learning

Security on Ethereum: Ponzi Scheme Detection in Smart Contract . . . . . . . . . . . . 435

Hongliang Zhang, Jiguo Yu, Biwei Yan, Ming Jing, and Jianli Zhao

Learning Signed Network Embedding via Muti-attention Mechanism . . . . . . . . . 444

Zekun Lu, Qiancheng Yu, Xiaofeng Wang, and Xiaoning Li

Three Algorithms for Converting Control Flow Statements from Python

to XD-M . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 456
Jiarui Wang, Nan Zhang, and Zhenhua Duan

Hyperspectral Image Reconstruction for SD-CASSI Systems Based

on Residual Attention Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 466
Haobin Luo, Guowei Su, Yi Wang, Jiajia Zhang, and Luobing Dong

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477

Approximation Algorithms
 A Binary Search Double Greedy
Algorithm for Non-monotone
DR-submodular Maximization

Shuyang Gu1(B) , Chuangen Gao2 , and Weili Wu3

1
Department of Computer Information Systems,
Texas A&M University - Central Texas, Killeen, TX 76549, USA
[email protected]
2
School of Computer Science and Technology,
Qilu Technology University, Jinan, China
3
Department of Computer Science,
The University of Texas at Dallas, Dallas, TX, USA
[email protected]

Abstract. In this paper, we study the non-monotone DR-submodular

function maximization over integer lattice. Functions over integer lattice
have been defined submodular property that is similar to submodularity
of set functions. DR-submodular is a further extended submodular con-
cept for functions over the integer lattice, which captures the diminishing
return property. Such functions finds many applications in machine learn-
ing, social networks, wireless networks, etc. The techniques for submodu-
lar set function maximization can be applied to DR-submodular function
maximization, e.g., the double greedy algorithm has a 1/2-approximation
ratio, whose running time is O(nB), where n is the size of the ground
set, B is the integer bound of a coordinate. In our study, we design a
1/2-approximate binary search double greedy algorithm, and we prove
that its time complexity is O(n log B), which significantly improves the
running time.

Keywords: Non-monotone · DR-submodular · Binary search double

greedy · Approximation algorithm

1 Introduction

A lot of real-world problems have objective functions with a so-called submodular

property, which reflects the diminish return nature for the problems. Since such
property exists in a vast amount of applications, submodular optimization has
caught a lot of attention during the past two decades. A set function f : 2E → R
is submodular if f (X) + f (Y ) ≥ f (X ∪ Y ) − f (X ∩ Y ) holds for any two sets
X, Y ⊆ E where E is a ground set. Submodular function has an equivalent
definition in terms of the diminishing return property: f (X ∪ {e}) − f (X) ≥
f (Y ∪ {e}) − f (Y ) X ⊆ Y , the element e ∈ E \ Y .

c The Author(s), under exclusive license to Springer Nature Switzerland AG 2022

Q. Ni and W. Wu (Eds.): AAIM 2022, LNCS 13513, pp. 3–14, 2022.
https://doi.org/10.1007/978-3-031-16081-3_1
4 S. Gu et al.

Submodular set function optimization includes maximizing or minimizing a

submodular function with or without some constraints. One of the directions
is non-monotone submodular maximization without constraint. Given a non-
negative submodular function f , the goal is to find a subset S that maximizes
f (S). Since this problem captures many applications in machine learning, viral
marketing, etc., it has been studied extensively. A deterministic local search gives
a 1/2 -approximation and a randomized smoothed local search algorithm gives
2/5-approximation [3]. Buchbinder et al. further improve that result, they show
that a deterministic double greedy algorithm provides 1/3-approximation, and
the randomized version of it gives a 1/2-approximation, both in linear time [5].
Recently, submodular optimization has been extended to functions over inte-
ger lattice, which considers the situation that each element in the ground set can
be selected as multiple copies. The functions over integer lattice may also have
submodular property, which is defined similarly to set functions’ submodular. A
function defined over the integer lattice ZE + is lattice submodular if the following
inequality holds:
f (x) + f (y) ≥ f (x ∨ y) + f (x ∧ y), x, y ∈ ZE
+.

The techniques for submodular set function optimization can be applied to

lattice submodular optimization. Based on the double greedy algorithm in [5], an
algorithm for submodular functions over the bounded integer lattice is designed
with 1/3 approximation ratio [2].
Although the definition of lattice submodular is similar to set function sub-
modular, the lattice submodular does not imply diminish return property. Soma
et al. [4] give a stronger generalization of submodularity on integer lattice, which
is called diminishing return submodular (DR-submodular) functions, such func-
tions capture various applications with diminishing return property. DR sub-
modular function on a bounded integer lattice satisfies f (x + χe ) − f (x) ≥
f (y + χe ) − f (y) for any x ≤ y and e ∈ E, where χe denotes a unit vector, i.e.
χie ∈ ZE is the vector with χe (e) = 1 and χe (a) = 0 for every a = e. In this
paper, we specifically study the profit maximization problem in social networks
as an application of non-monoton DR-submodular maximization.
The contributions of this paper are summarized as follows.
– To solve the non-monotone DR-submodular maximization problem, we pro-
pose the binary search double greedy algorithm.
– We prove the algorithm has a 1/2- approximate ratio and the time complex-
ity is polynomial(n log B). To the best of our knowledge, this is the fastest
algorithm with the least queries to the objective function.

2 Related Work
Non-monotone DR-submodular Maximization is closely related to non-monotone
submodular set function maximization because the algorithm for the latter prob-
lem can be applied to the former problem directly. The non-monotone submodu-
lar maximization is also called Unconstrained Submodular Maximization (USM).
 A Binary Search Double Greedy Algorithm 5

USM has various applications, such as marketing strategies over social networks
[18], Max-Cut [19], and maximum facility location [20]. USM problem has been
studied extensively [21–23]. Buchbinder gives a tight linear time randomized
(1/2)-approximation for the problem [5].
The topic of functions over integer lattice optimization has attracted much
attention recently. Monotone submodular functions over integer lattice with car-
dinality constraint are addressed in [10,12,14]. Sahin et al. study lattice sub-
modular functions subject to a discrete (integer) polymatroid constraint [13].
Zhang et al. study the problem of maximizing the sum of a monotone non-
negative DR-submodular function and a supermodular function on the integer
lattice subject to a cardinality constraint [11]. The non-submodular functions on
the integer lattice are addressed in [16]. Nong et al. focus on maximizing a non-
monotone weak-submodular function on a bounded integer lattice [17]. For the
problem addressed in this paper, non-monotone DR-submodular function maxi-
1
mization, Soma et al. design a 2+ -approximation algorithm with a running time
n 2
of O(  log B) [6].
In the meantime, the discrete domains of submodular functions over inte-
ger lattice are further extended to continuous domains, Hassani et al. study
stochastic projected gradient methods for maximizing continuous submodular
functions with convex constraints [15]. In [7,8], the authors consider maximizing
a continuous and nonnegative submodular function over a hypercube.

3 Preliminaries
We say that a set function g : f : 2E → R+ is submodular if it satisfies a natural
“diminishing returns” property: the marginal gain from adding an element to a
set X is at least as high as the marginal gain from adding the same element to
a superset of X. Formally, for every set X, Y such that X ⊆ Y ⊆ E and every
e ∈ E \ Y , it follows that

g(X ∪ {e}) − g(X) ≥ g(Y ∪ {e}) − g(Y )

An equivalent definition of the submodularity is

g(X) + g(Y ) ≥ g(X ∪ Y ) + f (X ∩ Y ), ∀X, Y ⊆ E

A set function is monotone if g(X) ≤ g(Y ) for all X ⊆ Y .

Functions over integer lattice has similar property. A function h : ZE+ → R+
that is defined over the integer lattice is submodular if the following holds [2]:

h(x) + h(y) ≥ h(x ∨ y) + h(x ∧ y), x, y ∈ ZE

+.

where (x ∨ y)(i) = max{x(i), y(i)} and (x ∧ y)(i) = min{x(i), y(i)}. Hence

x∨y represents coordinate-wise maximum, and x∧y denote the coordinate-wise
minimum. We can see this form of submodularity is a more generalized definition
of submodularity that covers set functions submodular, because vectors with all
6 S. Gu et al.

entries equal to either 0 or 1 can be seen as a subset including the elements that
are equal to 1 while excluding the elements that are equal to 0, in that case,
x ∧ y and x ∧ y transform to set intersection and set union of the subsets that
x and y represent respectively.
The submodular function over integer lattice does not have the diminishing
return property. To capture such property in real-world problems, a stronger
version of submodularity has been introduced, which is called DR-submodular
[4]. DR submodular function on a bounded integer lattice satisfies the following
diminish return property:

h(x + χe ) − h(x) ≥ h(y + χe ) − h(y), ∀x ≤ y, ∀e ∈ E

where χe denotes a unit vector, i.e. χe ∈ ZE is the vector with χe (e) = 1 and
χe (a) = 0 for every a = e.
The problem we consider is maximizing (non-monotone) DR-submodular
functions. Formally, we study the optimization problem

max f (x)
(1)
subject to 0 ≤ x ≤ B,

where f : ZE+ → R+ is a non-negative DR-submodular function and not neces-

sarily monotone. 0 is the all zero vector, and B ∈ ZE
+ is a vector representing the
maximum value for each coordinate. When B is the all-ones vector, the problem
is equivalent to the original unconstrained submodular set function maximiza-
tion. We assume that f is given as an evaluation oracle; when we specify x ∈ ZE +,
the oracle returns the value of f (x). We define f (x|y) = f (y + x) − f (y).

4 Algorithm
In this section, we present the algorithm for non-monotone DR-submodular func-
tion maximization. The main idea is inspired by the double greedy algorithm for
the unconstrained submodular maximization (USM) [5] on set functions. The
algorithm can be extended to accommodate DR-submodular function over inte-
ger lattice [6], because DR-submodular function can be treated as submodular set
function on each coordinate. We investigate some interesting properties for DR-
submodular functions to further speed up the algorithm. In the rest of the paper
some notations will be used. We define two new functions φ(b) := f (χe |x + bχe ),
ψ(b) := f (−χe |y − bχe ), where b ∈ Z+ . Both functions are non-increasing func-
tions of b because the function f is DR-submodular.
The algorithm starts with two vectors, x = 0 and y = c. For each coordinate
e ∈ E it iteratively either increase x(e) or decrease y(e) by σ, which depends on
the marginal gain by adding σ to x(e) and the marginal gain by removing σ to
y(e). This procedure continues until x(e) = y(e). Then it moves on to work on
 A Binary Search Double Greedy Algorithm 7

Algorithm 1. Binary Search Greedy Algorithm

Input: f : ZE + E
+ → R , c ∈ Z+
Assumption: f is DR-submodular

1: x ← 0, y ← c;
2: for e ∈ E do
3: Find arg minb φ(b) such that φ(b) < 0 by binary search.
4: u ← x(e) + arg minb φ(b) − 1.
5: Find arg minb ψ(b) such that ψ(b) < 0 by binary search.
6: v ← y(e) − arg minb ψ(b) + 1.
7: while x(e) < y(e) do
8: σ ← max( y (e)−x
2
(e)
, 1)
9: α ← f (σχe |x) and β ← f (−σχe |y)
10: if β ≤ 0 then
11: x(e) ← x(e) + σ
12: else if α ≤ 0 then
13: y(e) ← y(e) − σ
14: else
15: Randomly update x(e) ← x(e) + σ or y(e) ← y(e) − σ; the former case
α β
occurs with probability α+β , the later case with the probability α+β .
16: end if
17: end while
18: if x(e) ≥ u then
19: x(e) ← u, y(e) ← u
20: end if
21: if y(e) ≤ v then
22: y(e) ← v, x(e) ← v
23: end if
24: end for
25: return x

the next coordinate, after x and y agrees on all coordinates e ∈ E, x = y, and

the vector is the output of the algorithm. Different from applying the double
greedy algorithm directly, which tightens the gap one unit per step, Algorithm 1
tightens it by half in each iteration. The binary search nature of this algorithm
guarantees that the number of iterations needed is a logarithm of B. Next let
us firstly give a few results based on the diminish return property, which will be
used in proving the theoretical guarantee of the algorithm later.

Lemma 1. For ∀x ≤ y, k ≥ 1, k ∈ Z+ , and the value of k does not violate the

integer bound. We have

f (x + kχe ) − f (x) ≥ f (y + kχe ) − f (y)

8 S. Gu et al.

Proof.

f (x + χe ) − f (x) ≥ f (y + χe ) − f (y)
f (x + 2χe ) − f (x + χe ) ≥ f (y + 2χe ) − f (y + χe )
..
.
f (x + kχe ) − f (x + (k − 1)χe ) ≥ f (y + kχe ) − f (y + (k − 1)χe )

Sum up the above inequalities, the lemma holds. 

Note that from Lemma 1, we have

α + β = f (x + σχe ) − f (x) − (f (y + σχe ) − f (y)) ≥ 0 (2)

Lemma 2. For ∀x ≤ y, k ≥ 1, k ∈ Z+ , k ≤ y(e), we have

f (x − kχe ) − f (x) ≤ f (y − kχe ) − f (y)

Lemma 2 can be easily obtained from Lemma 1.

Lemma 3. Given p ≤ q, p, q ∈ Z + , if f (χe |x + (q − 1)χe ) ≥ 0, then

0 ≤ f (pχe |x) ≤ f (qχe |x)

Proof.

f (pχe |x) = f (χe |x) + f (χe |x + χe ) + · · · + f (χe |x + (p − 1)χe )

f (qχe |x) = f (χe |x) + f (χe |x + χe ) + · · · + f (χe |x + (q − 1)χe )

Since the values of the terms in the above equations are non-increasing from
left to right, f (χe |x + (q − 1)χe ) ≥ 0, so all terms are greater than or equal to
0. And f (qχe |x) has more terms, so the lemma holds. 

We can obtain a similar property in terms of the vector y.

Lemma 4. Given p ≤ q, p, q ∈ Z + , if f (−χe |y − (q − 1)χe ) ≥ 0, then

0 ≤ f (−pχe |y) ≤ f (−qχe |y)

The rest of this section is devoted to proving that Algorithm 1 provides an

approximation ratio of 1/2 for DR-submodular maximization. Let us begin the
analysis of Algorithm 1 with the introduction of some notation. Let xei and y ei
be random variables denoting the vectors generated by the algorithm at the end
of the i-th iteration for coordinate e, let the number of iterations for coordinate
e is θe , note that 1 ≤ i ≤ θe ≤ log B. Denote by opt the optimal solution.
Let us define the following random variable: optei  (opt ∨ xei ) ∧ y ei . Note that
xe0 (e) = 0, y e0 (e) = B, and opte0 (e) = opt(e). Additionally, the following always
holds: opteθe (e) = xeθe (e) = y eθe (e), ∀e ∈ E.
 A Binary Search Double Greedy Algorithm 9

Let us analyze the approximation ratio of the randomized algorithm. We

consider the subsequence E[f (opte0 )], . . . , E[f (opteθe )] for any dimension e ∈ E,
and a whole sequence which is a combination of every such subsequence for each
element e ∈ E. This sequence starts with f (opt) and ends with the expected
value of the algorithm’s output. The following lemma upper bounds the loss
between every two consecutive elements in the sequence. Formally, E[f (optei−1 )−
f (optei )] is upper bounded by the average expected change in the value of the two
solutions maintained by the algorithm, i.e., 12 E[f (xei )−f (xei−1 )+f (y ei )−f (y ei−1 )].

Lemma 5. For every 1 ≤ i ≤ θe ,

1
E[f (optei−1 ) − f (optei )] ≤ E[f (xei ) − f (xei−1 ) + f (y ei ) − f (y ei−1 )] (3)
2
where expectations are taken over the random choices of the algorithm.

Proof. Notice that it suffices to prove the inequality conditioned on any event of
the form xei−1 = sei−1 , where sei−1 ∈ ZE+ , (sei−1 (e) ≤ σ1 +· · ·+σi−1 ), for which the
probability that xei−1 = sei−1 is nonzero. Hence, fix such an event corresponding
to an integer vector sei−1 . The rest of the proof implicitly assumes everything
is conditioned on this event. Since the analysis is same for every coordinate,
we omit the superscript e in xei , yie , sei and optei in the following proof. After an
iteration i on coordinate e, denote by δi the distance between xi (e) and yi (e),
i
which can be calculate as δi = yi (e) − xi (e) = B − k=1 σk . The parameter σi

B− i−1 σ
can be obtained iteratively as σ1 =  B2 , σi =  2
k=1 k
 =  δ2i . Due to the
conditioning, the following random variables become constants:

1. y i−1 , where y i−1 (e) = si−1 (e) + δi−1

2. opti−1  (opt ∨ xi−1 ) ∧ y i−1 , where opti−1 (e) = si−1 (e) + min(opt(e) −
si−1 (e), δi−1 )
3. αi and βi , which refer to α, β at the iteration i .

By Lemma 1, αi + βi ≥ 0. Thus at most one of αi and βi is strictly less than

zero. We need to consider the following three cases for the value of αi and βi :
Case 1: (αi ≥ 0 and βi ≤ 0). In this case the vector y does not change:
y i = y i−1 . The vector x changes. xi ← xi−1 +σi χe . Hence, f (y i )−f (y i−1 ) = 0.
Also, by our definition opti  (opt ∨ xi ) ∧ y i = (opt ∨ (xi−1 + σi χe )) ∧ y i . Thus,
we are left to prove that
1 αi
f ((opt ∨ x i−1 ) ∧ y i−1 ) − f ((opt ∨ (x i−1 + σi χe )) ∧ y i ) ≤ [f (x i ) − f (x i−1 )] =
2 2

We prove it considering the relationship among xi−1 (e), xi (e) and opti (e).
Case 1.1: xi (e) = si−1 (e) + σi ≤ opt(e).
This condition implies xi−1 (e) ≤ opt(e). Since y i = y i−1 , the left-hand side
of the last inequality is 0, which is definitely not greater than the nonnegative
αi
2 .
10 S. Gu et al.

Case 1.2: si−1 (e) ≥ opt(e).

This condition implies that xi (e) = si−1 (e) + σi > opt(e).
We can see that (opt ∨ (xi−1 + σi χe )) ∧ y i = (opt ∨ xi−1 ) ∧ y i−1 + σi χe =
opti−1 + σi χe ≤ y i−1 , by diminish return submodularity we have

f ((opt ∨ xi−1 ) ∧ y i−1 ) − f ((opt ∨ (xi−1 + σi χe )) ∧ y i )

αi
= f (−σi χe |opti−1 + σi χe ) ≤ f (−σi χe |y i−1 ) = β ≤ 0 ≤
2

Case 1.3: si−1 (e) ≤ opt(e) and xi (e) = si−1 (e) + σi > opt(e).
Then we have

((opt ∨ x i−1 ) ∧ y i−1 ) − (opt ∨ (x i−1 + σi χ e ) ∧ y i ) = −(s i−1 (e) + σi − opt(e))χ e

Let si−1 (e) + σi − opt(e) = δ, then

((opt ∨ xi−1 ) ∧ y i−1 ) − (opt ∨ (xi−1 + σi χe ) ∧ y i ) = −δχe .

And
f ((opt ∨ xi−1 ) ∧ y i−1 ) − f ((opt ∨ (xi−1 + σi χe )) ∧ y i )
= f (−δχe |(opt ∨ (xi−1 + σi χe )) ∧ y i ) (4)
= f (−δχe |opti ).

By the definition of the random variable opti , we have

xi ≤ opti ≤ y i

Note that 0 ≤ δ < σi . Since ψ(b) := f (−χe |y − bχe ) is a non-increasing function

on b. f (−σi χe |y i ) ≤ 0 implies f (−δχe |y i − σi χe ) ≤ 0. We note that in the ithe
iteration, y i = y i−1 and due to the condition of case 1.3, we have y i − σi χe ≥
opti , thus f (−δχe |opti ) ≤ f (−δχe |y i − σi χe ) ≤ 0 ≤ α2i .
Case 2: (αi < 0 and βi > 0). This case is analogous to the previous one, and
therefore we omit its proof.
αi
Case 3: (αi ≥ 0 and βi > 0). With probability αi +βi the following events
βi
happen: xi ← xi−1 + σi χe and y i ← y i−1 ; while with probability αi +βi the
following events happen: xi ← xi−1 and y i ← y i−1 − σi χe Thus,
αi
E[f (xi ) − f (xi−1 ) + f (y i ) − f (y i−1 )] = [f (xi−1 + σi χe ) − f (xi−1 )]
αi + βi
βi
+ [f (y i−1 − σi χe ) − f (y i−1 )](5)
αi + βi
α2 + βi2
= i
αi + βi
 A Binary Search Double Greedy Algorithm 11

Next, we upper bound E[f (opti−1 ) − f (opti )].

E[f (opti−1 ) − f (opti )]

αi
= [f ((opt ∨ xi−1 ) ∧ y i−1 ) − f ((opt ∨ (xi−1 + σi χe )) ∧ y i )]
αi + βi
βi (6)
+ [f ((opt ∨ xi−1 ) ∧ y i−1 ) − f ((opt ∨ xi ) ∧ (y i−1 − σi χe ))]
αi + βi
αi βi
≤
αi + βi
The final inequality follows by considering two cases. The first case is: y i (e) =
y i−1 (e) − σi , xi (e) = xi−1 (e), the first term of the left-hand side of the last
inequality equals zero. There are three subcases,
Case 3.1: (xi (e) ≤ y i (e) = y i−1 (e) − σi < y i−1 (e) ≤ opt(e)).
Thus (opt ∨ xi−1 ) ∧ y i−1 = (opt ∨ xi ) ∧ (y i−1 − σi χe ) + σi χe , and (opt ∨
xi ) ∧ (y i−1 − σi χe ) ≥ xi−1 , hence

f ((opt ∨ xi−1 ) ∧ y i−1 ) − f ((opt ∨ xi ) ∧ (y i−1 − σi χe ))

= f (σi χe |(opt ∨ xi ) ∧ (y i−1 − σi χe )) ≤ f (σi χe |xi−1 ) = αi

Case 3.2: (opt(e) ≤ y i (e) = y i−1 (e) − σi < y i−1 (e)).

In this case, the second term of the left-hand side of inequality (6) also equals
zero, thus inequality (6) follows.
Case 3.3: (xi (e) ≤ y i (e) = y i−1 (e) − σi ≤ opt(e) < y i−1 (e)).
We have (opt∨xi−1 )∧y i−1 = (opt∨xi )∧(y i−1 −σi χe )+(opt(e)−y i−1 (e)+
σi )χe . Let μ = opt(e) − y i−1 (e) + σi , then 0 < μ ≤ σi , and

f ((opt ∨ xi−1 ) ∧ y i−1 ) − f ((opt ∨ xi ) ∧ (y i−1 − σi χe ))

= f (μχe |(opt ∨ xi ) ∧ (y i−1 − σi χe ))
≤ f (μχe |y i−1 − σi χe ) ≤ 0 ≤ αi
The line 18–23 in the algorithm guarantees that f (−χe |y i−1 − (σ − 1)χe )) ≥ 0.
By Lemma 3, we have f (−σχe |y i−1 ) ≥ 0. Also f (−μχe |y i−1 − (σ − μ)χe ) ≥ 0.
Thus,
f (−μχe |y i−1 − (σ − μ)χe ) = −f (μχe |y i−1 − σi χe ) ≥ 0
By now, we show that the inequality (6) holds for the case y i (e) = y i−1 (e) − σi ,
xi (e) = xi−1 (e). The other case is that y i (e) = y i−1 (e), xi (e) = xi−1 (e) + σi ,
which is analogous to the previous case, we omit the proof here.
Now we show inequality (6) follows for all situations. By (5) and (6) inequality
(3) holds if
αi βi 1 α2 + βi2
≤ · i
αi + βi 2 αi + βi
which can easily be verified.
12 S. Gu et al.

Theorem 1. Algorithm 1 is a randomized O(n log B) time (1/2)-approximation

algorithm for the DR-Submodular Maximization problem.

Proof. Summing up Lemma 4 for every 1 ≤ i ≤ θe for each e ∈ E gives

θe

E[f (optei−1 ) − f (optei )]
e∈E i=1
(7)
θe
1 
≤ E[f (xei ) − f (xei−1 ) + f (y ei ) − f (y ei−1 )]
2 i=1
e∈E

The above sum is telescopic. We define that the algorithm executes on the vector
coordinates ordered by e1 , . . . , en . Collapse the inequality, we get
1
f (opte01 ) − f (opteθne ) ≤ E[f (xeθne ) − f (xe01 ) + f (y eθne ) − f (y e01 )]
2 n n
(8)
1
≤ E[f (xeθne ) + f (y eθne )]
2 n n

Recalling the definitions of optei , opte01 = opt, opteθne = xeθne = y eθne is the
n n
output solution, thus E[f (xeθne )] = E[f (y eθne )] ≥ f (opt)/2. It is clear that the
n n
algorithm makes O(n log B) oracle calls since for each coordinate e ∈ E the
number of oracle calls is at most log B and there are n = |E| coordinates.

5 Conclusions
In this paper, we propose a binary search double greedy algorithm for non-
monotone DR-submodular function maximization over bounded integer lattice.
Our algorithm improves the approximation ratio and significantly reduces the
time complexity.
One interesting direction for our future work is to explore the problems
in social networks that fall into non-monotone DR-submodular maximization,
exploit our proposed algorithm and test with real datasets. Another direction
we are interested in is maximizing DR-submodular function with some types
of constraints such as cardinality constraint, matroid constraint and knapsack
constraint.

References
1. Alon, N., Gamzu, I., Tennenholtz, M.: Optimizing budget allocation among chan-
nels and influencers. In: Proceedings of the 21st International Conference on World
Wide Web, pp. 381–388 (2012)
2. Gottschalk, C., Peis, B.: Submodular function maximization on the bounded inte-
ger lattice. In: Sanità, L., Skutella, M. (eds.) WAOA 2015. LNCS, vol. 9499, pp.
133–144. Springer, Cham (2015). https://doi.org/10.1007/978-3-319-28684-6_12
 A Binary Search Double Greedy Algorithm 13

3. Feige, U., Mirrokni, V.S., Vondrák, J.: Maximizing non-monotone submodular

functions. SIAM J. Comput. 40(4), 1133–1153 (2011)
4. Soma, T., Yoshida, Y.: A generalization of submodular cover via the diminishing
return property on the integer lattice. In: Advances in Neural Information Process-
ing Systems, vol. 28 (2015)
5. Buchbinder, N., Feldman, M., Seffi, J., Schwartz, R.: A tight linear time (1/2)-
approximation for unconstrained submodular maximization. SIAM J. Comput.
44(5), 1384–1402 (2015)
6. Soma, T., Yoshida, Y.: Non-monotone DR-submodular function maximization. In:
Proceedings of the AAAI Conference on Artificial Intelligence, vol. 31, no. 1 (2017)
7. Niazadeh, R., Roughgarden, T., Wang, J.: Optimal algorithms for continuous non-
monotone submodular and DR-submodular maximization. In: Advances in Neural
Information Processing Systems, vol. 31 (2018)
8. Bian, A., Levy, K., Krause, A., Buhmann, J.M.: Continuous DR-submodular maxi-
mization: structure and algorithms. In: Advances in Neural Information Processing
Systems, vol. 30 (2017)
9. Soma, T., Kakimura, N., Inaba, K., Kawarabayashi, K.-I.: Optimal budget alloca-
tion: theoretical guarantee and efficient algorithm. In: International Conference on
Machine Learning, pp. 351–359. PMLR (2014)
10. Soma, T., Yoshida, Y.: Maximizing monotone submodular functions over the inte-
ger lattice. Math. Program. 172(1), 539–563 (2018)
11. Zhang, Z., Du, D., Jiang, Y., Wu, C.: Maximizing DR-submodular+ supermodular
functions on the integer lattice subject to a cardinality constraint. J. Glob. Optim.
80(3), 595–616 (2021)
12. Lai, L., Ni, Q., Lu, C., Huang, C., Wu, W.: Monotone submodular maximiza-
tion over the bounded integer lattice with cardinality constraints. Discrete Math.
Algorithms Appl. 11(06), 1950075 (2019)
13. Sahin, A., Buhmann, J., Krause, A.: Constrained maximization of lattice submod-
ular functions. In: ICML 2020 Workshop on Negative Dependence and Submodu-
larity for ML, Vienna, Austria, PMLR, vol. 119 (2020)
14. Zhang, Z., Guo, L., Wang, Y., Xu, D., Zhang, D.: Streaming algorithms for max-
imizing monotone DR-submodular functions with a cardinality constraint on the
integer lattice. Asia-Pacific J. Oper. Res. 38(05), 2140004 (2021)
15. Hassani, H., Soltanolkotabi, M., Karbasi, A.: Gradient methods for submodular
maximization. In: Advances in Neural Information Processing Systems, vol. 30
(2017)
16. Kuhnle, A., Smith, J.D., Crawford, V., Thai, M.: Fast maximization of non-
submodular, monotonic functions on the integer lattice. In: International Con-
ference on Machine Learning, pp. 2786–2795. PMLR (2018)
17. Nong, Q., Fang, J., Gong, S., Du, D., Feng, Y., Qu, X.: A 1/2-approximation
algorithm for maximizing a non-monotone weak-submodular function on a bounded
integer lattice. J. Comb. Optim. 39(4), 1208–1220 (2020)
18. Hartline, J., Mirrokni, V., Sundararajan, M.: Optimal marketing strategies over
social networks. In: Proceedings of the 17th International Conference on World
Wide Web, pp. 189–198 (2008)
19. Goemans, M.X., Williamson, D.P.: Improved approximation algorithms for max-
imum cut and satisfiability problems using semidefinite programming. J. ACM
(JACM) 42(6), 1115–1145 (1995)
20. Ageev, A.A., Sviridenko, M.I.: An 0.828-approximation algorithm for the uncapac-
itated facility location problem. Discret. Appl. Math. 93(2–3), 149–156 (1999)
14 S. Gu et al.

21. Gharan, S.O., Vondrák, J.: Submodular maximization by simulated annealing. In:
Proceedings of the 22nd Annual ACM-SIAM Symposium on Discrete Algorithms,
pp. 1098–1116. SIAM (2011)
22. Buchbinder, N., Feldman, M.: Deterministic algorithms for submodular maximiza-
tion problems. ACM Trans. Algorithms (TALG) 14(3), 1–20 (2018)
23. Pan, X., Jegelka, S., Gonzalez, J.E., Bradley, J.K., Jordan, M.I.: Parallel double
greedy submodular maximization. In: Advances in Neural Information Processing
Systems, vol. 27 (2014)
 An Approximation Algorithm
for the Clustered Path Travelling
Salesman Problem

Jiaxuan Zhang, Suogang Gao, Bo Hou, and Wen Liu(B)

Hebei Key Laboratory of Computational Mathematics and Applications,

School of Mathematical Sciences, Hebei Normal University,
Shijiazhuang 050024, People’s Republic of China
[email protected]

Abstract. In this paper, we consider the clustered path travelling sales-

man problem. In this problem, we are given a complete graph G = (V, E)
with edge weight satisfying the triangle inequality. In addition, the vertex
set V is partitioned into clusters V1 , · · · , Vk . The objective of the problem
is to find a minimum Hamiltonian path in G, and in the path all vertices
of each cluster are visited consecutively. We provide a polynomial-time
approximation algorithm for the problem.

Keywords: Travelling salesman problem · Stacker crane problem ·

Path · Cluster

1 Introduction

The travelling salesman problem (TSP) is a best-known combinatorial optimiza-

tion problem. In this problem, we are given a complete graph G = (V, E) with
vertex set V and edge set E, and there is an edge weight function ω satisfying
the triangle inequality. The task of the TSP is to find a minimum Hamiltonian
cycle. This problem is NP-hard and has multitudes of applications [4,6].
Meanwhile, TSP has quite a lot variants [11,18]. Among these variants there is
an important one called the path travelling salesman problem (PTSP), whose task
is to find a minimum Hamiltonian path. For the PTSP, Hoogeveen [15] presented √
a 53 -approximation algorithm. An et al. [1] improved this result and gave a 1+2 5 -
approximation algorithm. Recently, Zenklusen [22] developed the best known 32 -
approximation algorithm. For more work on this problem, one can see [12,19,20].

Supported by the NSF of China (No. 11971146), the NSF of Hebei Province of China
(No. A2019205089, No. A2019205092), Overseas Expertise Introduction Program of
Hebei Auspices (25305008) and the Graduate Innovation Grant Program of Hebei
Normal University (No. CXZZSS2022052).
c The Author(s), under exclusive license to Springer Nature Switzerland AG 2022
Q. Ni and W. Wu (Eds.): AAIM 2022, LNCS 13513, pp. 15–27, 2022.
https://doi.org/10.1007/978-3-031-16081-3_2
16 J. Zhang et al.

Another best-studied variant of the TSP is the clustered travelling salesman

problem (CTSP). In this problem, the vertex set V is partitioned into clusters
V1 , · · · , Vk . The goal of the CTSP is to find a minimum Hamiltonian cycle in G,
and in the cycle all vertices of each cluster are visited consecutively. Note that if
k = 1, the CTSP is exactly the TSP. So in the following we assume k ≥ 2. Chis-
man [5] first introduced the CTSP and gave some applications about it. Arkin et al.
[3] developed the first approximation algorithm for the CTSP with a performance
guarantee of 72 . Guttmann-Beck et al. [13] designed approximation algorithms for
several cases of the CTSP by decomposing them into the PTSP together with the
stacker crane problem, or the PTSP together with the rural postman problem.
Then, Kawasaki and Takazawa [17] improved approximation ratios by applying an
improved approximation algorithm for the PTSP given by Zenklusen [22]. Appli-
cations and other related work for the CTSP may be found in [9,16].
Motivated by the work of Kawasaki and Takazawa [17] and Anily et al. [2], we
study the clustered path travelling salesman problem (CPTSP). In the CPTSP,
we are given a complete graph G = (V, E) with edge weight satisfying the
triangle inequality, and the vertex set V is partitioned into clusters V1 , · · · , Vk .
The goal is to find a minimum Hamiltonian path in G, and in the path all
vertices of each cluster are visited consecutively. For the CPTSP, we get three
corresponding problems when we specify neither, one or both endpoints of the
Hamiltonian path. But we only deal with the case that both endpoints of the path
are specified. The reason is by guessing an endpoint, one can use an algorithm
for the case with two specified endpoints to solve the case with only one specified
endpoint, and by the algorithm for the case with one specified endpoint to solve
that with no specified endpoint. Specifically, let s (t) be the start (end) vertex of
the Hamiltonian path. The goal for the CPTSP is to find a minimum Hamiltonian
path in G from s to t, and in the path all vertices of each cluster are visited
consecutively. Note that s must be the start vertex for one cluster and t must
be the end vertex for another. We might as well assume s ∈ V1 and t ∈ Vk .
Then V1 and Vk are the first and the last clusters to be visited respectively. For
other clusters, we visit them in any order. Note that there are k clusters in the
CPTSP and a Hamiltonian path of this problem induces a Hamiltonian path
in each cluster. For simplicity, we assume that in each cluster the start vertex
and the end vertex for the Hamiltonian path are both specified. In this paper,
we design an approximation algorithm with an approximation ratio 83 for the
CPTSP by decomposing it into the path travelling salesman problem and the
path version of the stacker crane problem.
The paper is organized as follows. In Sect. 2, we give some definitions and
results. In Sect. 3, we design two approximation algorithms for the path stacker
crane problem. Based on these two algorithms, we design an algorithm for the
clustered path travelling salesman problem and analyze its approximation ratio
in Sect. 4. In Sect. 5, we provide several future research problems.
 An Approximation Algorithm 17

2 Preliminaries
In this section, we introduce some terminology, concepts and related results.
A graph G is a pair (V, E) where V is a set of objects called vertices and
E is a set of edges. Each edge is a pair {v, w} of vertices v, w ∈ V . If an edge
{v, w} is associated with a direction, a directed edge (arc) is obtained and we
denote it by (v, w) if the direction is from v to w or (w, v) otherwise. Vertices v
and w are called the tail and the head of the arc (v, w), respectively. A directed
graph has a set of vertices and a set of directed edges. In a directed graph, the
outdegree (indegree) of a vertex is the number of edges directed out of (into) the
vertex. A mixed graph in which both directed and undirected edges may exist.
A mixed multigraph is a graph, possibly with parallel edges, each of which is
either undirected or directed.
A walk in a graph G from vertex v1 to vl is a sequence
(v1 , {v1 , v2 }, v2 , {v2 , v3 }, v3 , · · · , vl−1 , {vl−1 , vl }, vl ), in which all vi  s are vertices
and {vi−1 , vi } ∈ E(G) for i = 2, 3, · · · , l. We also call v1 , vl the start vertex
and the end vertex of the walk. A path is a walk with no repeated vertices.
A trail is a walk with no repeated edges. An Eulerian trail is a walk passing
through every edge of G exactly once. If this walk is closed (starts and ends
at the same vertex), it is called an Eulerian tour. A Hamiltonian path/cycle in
G is a path/cycle visiting every vertex of G exactly once. A directed walk in
a directed graph is a sequence (v1 , (v1 , v2 ), v2 , (v2 , v3 ), v3 , · · · , vl−1 , (vl−1 , vl ), vl ).
A directed path is a directed walk with no repeated vertices. A directed trail
is a directed walk with no repeated arcs. A directed Eulerian trail is a directed
walk passing through every arc of G exactly once. Similarly, we can define that
on the mixed multigraph.
In the stacker crane problem (SCP), we are given a mixed multigraph G =
(V , E  , D), where V  = {si , ti | i ∈ [k]}, (V  , E  ) is an undirected complete


graph with an edge weight function satisfying the triangle inequality, and D =
{(si , ti ) | i ∈ [k]}. Here [k] is the notation for the set {1, 2, · · · , k}. The objective
is to find a minimum Hamiltonian cycle that traverses each arc (si , ti ) in the
specified direction from si to ti , i ∈ [k], where we identify the weight of the arc
(si , ti ) with that of the corresponding edge {si , ti }.
If the objective “Hamiltonian cycle” substitutes for “Hamiltonian path” in
the definition of the SCP, we get the path version of the stacker crane problem,
and we call it the path stacker crane problem (PSCP).
The following results are important for the discussion in Sects. 3 and 4.

Theorem 2.1 [7]. A connected multigraph has an Eulerian trail if and only if
it has either 0 or 2 vertices of odd degree.

By a simple deduction, we get the following result similar to that in [14].

18 J. Zhang et al.

Theorem 2.2. A connected directed multigraph has an Eulerian trail if and only
if every vertex has the same indegree and outdegree or the indegree of one vertex
is equal to the outdegree of this vertex plus one, the outdegree of another vertex
is equal to the indegree of this vertex plus one, and the outdegree of other vertices
is equal to the indegree of them.

Theorem 2.3 [17]. For the PTSP with u1 and u2 being the two given endpoints,
there exists a polynomial time approximation algorithm that finds Hamiltonian
paths S1 and S2 from u1 to u2 such that w(S1 ) ≤ 2OP T  − w{u1 , u2 }, w(S2 ) ≤
3  
2 OP T , where OP T denotes the weight of an optimal solution of the problem.

3 Approximation Algorithms for the PSCP

In this section, we give two polynomial-time algorithms for the PSCP which will
be used when we design the algorithm for the CPTSP in next section.
Recall that in the PSCP, we are given a mixed multigraph G = (V  , E  , D),
where V  = {si , ti | i ∈ [k]}, (V  , E  ) is an undirected complete graph with an
edge weight function ω satisfying the triangle inequality, and D = {(si , ti ) | i ∈
[k]}. The objective is to find a minimum Hamiltonian path from s1 to tk that
traverses each arc (si , ti ) in the specified direction from si to ti , i ∈ [k]. The first
algorithm for the PSCP is as follows.

Algorithm 1.
Step 1: Find a minimum bipartite matching between the head set T =
{t1 , t2 , · · · , tk−1 } and the tail set S = {s2 , s3 , · · · , sk }.
Step 2: Initialize E1 to be empty. For each edge included in the above match-
ing, associate a direction with it, going from T to S, and insert it into E1 . (This
results in m ≥ 1 disjoint connected components, each of which consists of edges
with the associated directions in the matching and arcs in D, and we denote
these m disjoint connected components by Ri , i ∈ [m].)
Step 3: Condense each Ri into a single node ni . Define

d{ni , nj } = min{ω{u, v} | u ∈ Ri  , v ∈ Rj  }

where Ri  represents the set of all vertices in Ri except for s1 , tk .

Step 4: Find a minimum spanning tree for the nodes {ni | i ∈ [m]}. Here the
minimum is with respect to the distance function d defined in Step 3.
Step 5: Firstly, make two copies of each edge in the spanning tree. Secondly,
associate one direction with one copy, and the opposite direction with the other.
Thirdly, insert these edges with the  associated directions into E1 . (This results
in a directed graph G1 = (V  , E1 D).)
Step 6: Find an Eulerian trail from s1 to tk in G1 .
Step 7: Using the triangle inequality, we get a Hamiltonian path PPSCP1 from
s1 to tk traversing each arc (si , ti ) in the specified direction from si to ti , i ∈ [k].
 Another Random Document on
Scribd Without Any Related Topics
*** END OF THE PROJECT GUTENBERG EBOOK THE CASTAWAY ***

Updated editions will replace the previous one—the old editions will
be renamed.

Creating the works from print editions not protected by U.S.

copyright law means that no one owns a United States copyright in
these works, so the Foundation (and you!) can copy and distribute it
in the United States without permission and without paying
copyright royalties. Special rules, set forth in the General Terms of
Use part of this license, apply to copying and distributing Project
Gutenberg™ electronic works to protect the PROJECT GUTENBERG™
concept and trademark. Project Gutenberg is a registered trademark,
and may not be used if you charge for an eBook, except by following
the terms of the trademark license, including paying royalties for use
of the Project Gutenberg trademark. If you do not charge anything
for copies of this eBook, complying with the trademark license is
very easy. You may use this eBook for nearly any purpose such as
creation of derivative works, reports, performances and research.
Project Gutenberg eBooks may be modified and printed and given
away—you may do practically ANYTHING in the United States with
eBooks not protected by U.S. copyright law. Redistribution is subject
to the trademark license, especially commercial redistribution.

START: FULL LICENSE

THE FULL PROJECT GUTENBERG LICENSE
 PLEASE READ THIS BEFORE YOU DISTRIBUTE OR USE THIS WORK

To protect the Project Gutenberg™ mission of promoting the free

distribution of electronic works, by using or distributing this work (or
any other work associated in any way with the phrase “Project
Gutenberg”), you agree to comply with all the terms of the Full
Project Gutenberg™ License available with this file or online at
www.gutenberg.org/license.

Section 1. General Terms of Use and

Redistributing Project Gutenberg™
electronic works
1.A. By reading or using any part of this Project Gutenberg™
electronic work, you indicate that you have read, understand, agree
to and accept all the terms of this license and intellectual property
(trademark/copyright) agreement. If you do not agree to abide by all
the terms of this agreement, you must cease using and return or
destroy all copies of Project Gutenberg™ electronic works in your
possession. If you paid a fee for obtaining a copy of or access to a
Project Gutenberg™ electronic work and you do not agree to be
bound by the terms of this agreement, you may obtain a refund
from the person or entity to whom you paid the fee as set forth in
paragraph 1.E.8.

1.B. “Project Gutenberg” is a registered trademark. It may only be

used on or associated in any way with an electronic work by people
who agree to be bound by the terms of this agreement. There are a
few things that you can do with most Project Gutenberg™ electronic
works even without complying with the full terms of this agreement.
See paragraph 1.C below. There are a lot of things you can do with
Project Gutenberg™ electronic works if you follow the terms of this
agreement and help preserve free future access to Project
Gutenberg™ electronic works. See paragraph 1.E below.
1.C. The Project Gutenberg Literary Archive Foundation (“the
Foundation” or PGLAF), owns a compilation copyright in the
collection of Project Gutenberg™ electronic works. Nearly all the
individual works in the collection are in the public domain in the
United States. If an individual work is unprotected by copyright law
in the United States and you are located in the United States, we do
not claim a right to prevent you from copying, distributing,
performing, displaying or creating derivative works based on the
work as long as all references to Project Gutenberg are removed. Of
course, we hope that you will support the Project Gutenberg™
mission of promoting free access to electronic works by freely
sharing Project Gutenberg™ works in compliance with the terms of
this agreement for keeping the Project Gutenberg™ name associated
with the work. You can easily comply with the terms of this
agreement by keeping this work in the same format with its attached
full Project Gutenberg™ License when you share it without charge
with others.

1.D. The copyright laws of the place where you are located also
govern what you can do with this work. Copyright laws in most
countries are in a constant state of change. If you are outside the
United States, check the laws of your country in addition to the
terms of this agreement before downloading, copying, displaying,
performing, distributing or creating derivative works based on this
work or any other Project Gutenberg™ work. The Foundation makes
no representations concerning the copyright status of any work in
any country other than the United States.

1.E. Unless you have removed all references to Project Gutenberg:

1.E.1. The following sentence, with active links to, or other

immediate access to, the full Project Gutenberg™ License must
appear prominently whenever any copy of a Project Gutenberg™
work (any work on which the phrase “Project Gutenberg” appears,
or with which the phrase “Project Gutenberg” is associated) is
accessed, displayed, performed, viewed, copied or distributed:
 This eBook is for the use of anyone anywhere in the United
States and most other parts of the world at no cost and with
almost no restrictions whatsoever. You may copy it, give it away
or re-use it under the terms of the Project Gutenberg License
included with this eBook or online at www.gutenberg.org. If you
are not located in the United States, you will have to check the
laws of the country where you are located before using this
eBook.

1.E.2. If an individual Project Gutenberg™ electronic work is derived

from texts not protected by U.S. copyright law (does not contain a
notice indicating that it is posted with permission of the copyright
holder), the work can be copied and distributed to anyone in the
United States without paying any fees or charges. If you are
redistributing or providing access to a work with the phrase “Project
Gutenberg” associated with or appearing on the work, you must
comply either with the requirements of paragraphs 1.E.1 through
1.E.7 or obtain permission for the use of the work and the Project
Gutenberg™ trademark as set forth in paragraphs 1.E.8 or 1.E.9.

1.E.3. If an individual Project Gutenberg™ electronic work is posted

with the permission of the copyright holder, your use and distribution
must comply with both paragraphs 1.E.1 through 1.E.7 and any
additional terms imposed by the copyright holder. Additional terms
will be linked to the Project Gutenberg™ License for all works posted
with the permission of the copyright holder found at the beginning
of this work.

1.E.4. Do not unlink or detach or remove the full Project

Gutenberg™ License terms from this work, or any files containing a
part of this work or any other work associated with Project
Gutenberg™.

1.E.5. Do not copy, display, perform, distribute or redistribute this

electronic work, or any part of this electronic work, without
prominently displaying the sentence set forth in paragraph 1.E.1
with active links or immediate access to the full terms of the Project
Gutenberg™ License.

1.E.6. You may convert to and distribute this work in any binary,
compressed, marked up, nonproprietary or proprietary form,
including any word processing or hypertext form. However, if you
provide access to or distribute copies of a Project Gutenberg™ work
in a format other than “Plain Vanilla ASCII” or other format used in
the official version posted on the official Project Gutenberg™ website
(www.gutenberg.org), you must, at no additional cost, fee or
expense to the user, provide a copy, a means of exporting a copy, or
a means of obtaining a copy upon request, of the work in its original
“Plain Vanilla ASCII” or other form. Any alternate format must
include the full Project Gutenberg™ License as specified in
paragraph 1.E.1.

1.E.7. Do not charge a fee for access to, viewing, displaying,

performing, copying or distributing any Project Gutenberg™ works
unless you comply with paragraph 1.E.8 or 1.E.9.

1.E.8. You may charge a reasonable fee for copies of or providing

access to or distributing Project Gutenberg™ electronic works
provided that:

• You pay a royalty fee of 20% of the gross profits you derive
from the use of Project Gutenberg™ works calculated using the
method you already use to calculate your applicable taxes. The
fee is owed to the owner of the Project Gutenberg™ trademark,
but he has agreed to donate royalties under this paragraph to
the Project Gutenberg Literary Archive Foundation. Royalty
payments must be paid within 60 days following each date on
which you prepare (or are legally required to prepare) your
periodic tax returns. Royalty payments should be clearly marked
as such and sent to the Project Gutenberg Literary Archive
Foundation at the address specified in Section 4, “Information
 about donations to the Project Gutenberg Literary Archive
Foundation.”

• You provide a full refund of any money paid by a user who

notifies you in writing (or by e-mail) within 30 days of receipt
that s/he does not agree to the terms of the full Project
Gutenberg™ License. You must require such a user to return or
destroy all copies of the works possessed in a physical medium
and discontinue all use of and all access to other copies of
Project Gutenberg™ works.

• You provide, in accordance with paragraph 1.F.3, a full refund of

any money paid for a work or a replacement copy, if a defect in
the electronic work is discovered and reported to you within 90
days of receipt of the work.

• You comply with all other terms of this agreement for free
distribution of Project Gutenberg™ works.

1.E.9. If you wish to charge a fee or distribute a Project Gutenberg™

electronic work or group of works on different terms than are set
forth in this agreement, you must obtain permission in writing from
the Project Gutenberg Literary Archive Foundation, the manager of
the Project Gutenberg™ trademark. Contact the Foundation as set
forth in Section 3 below.

1.F.

1.F.1. Project Gutenberg volunteers and employees expend

considerable effort to identify, do copyright research on, transcribe
and proofread works not protected by U.S. copyright law in creating
the Project Gutenberg™ collection. Despite these efforts, Project
Gutenberg™ electronic works, and the medium on which they may
be stored, may contain “Defects,” such as, but not limited to,
incomplete, inaccurate or corrupt data, transcription errors, a
copyright or other intellectual property infringement, a defective or
damaged disk or other medium, a computer virus, or computer
codes that damage or cannot be read by your equipment.

1.F.2. LIMITED WARRANTY, DISCLAIMER OF DAMAGES - Except for

the “Right of Replacement or Refund” described in paragraph 1.F.3,
the Project Gutenberg Literary Archive Foundation, the owner of the
Project Gutenberg™ trademark, and any other party distributing a
Project Gutenberg™ electronic work under this agreement, disclaim
all liability to you for damages, costs and expenses, including legal
fees. YOU AGREE THAT YOU HAVE NO REMEDIES FOR
NEGLIGENCE, STRICT LIABILITY, BREACH OF WARRANTY OR
BREACH OF CONTRACT EXCEPT THOSE PROVIDED IN PARAGRAPH
1.F.3. YOU AGREE THAT THE FOUNDATION, THE TRADEMARK
OWNER, AND ANY DISTRIBUTOR UNDER THIS AGREEMENT WILL
NOT BE LIABLE TO YOU FOR ACTUAL, DIRECT, INDIRECT,
CONSEQUENTIAL, PUNITIVE OR INCIDENTAL DAMAGES EVEN IF
YOU GIVE NOTICE OF THE POSSIBILITY OF SUCH DAMAGE.

1.F.3. LIMITED RIGHT OF REPLACEMENT OR REFUND - If you

discover a defect in this electronic work within 90 days of receiving
it, you can receive a refund of the money (if any) you paid for it by
sending a written explanation to the person you received the work
from. If you received the work on a physical medium, you must
return the medium with your written explanation. The person or
entity that provided you with the defective work may elect to provide
a replacement copy in lieu of a refund. If you received the work
electronically, the person or entity providing it to you may choose to
give you a second opportunity to receive the work electronically in
lieu of a refund. If the second copy is also defective, you may
demand a refund in writing without further opportunities to fix the
problem.

1.F.4. Except for the limited right of replacement or refund set forth
in paragraph 1.F.3, this work is provided to you ‘AS-IS’, WITH NO
OTHER WARRANTIES OF ANY KIND, EXPRESS OR IMPLIED,
INCLUDING BUT NOT LIMITED TO WARRANTIES OF
MERCHANTABILITY OR FITNESS FOR ANY PURPOSE.

1.F.5. Some states do not allow disclaimers of certain implied

warranties or the exclusion or limitation of certain types of damages.
If any disclaimer or limitation set forth in this agreement violates the
law of the state applicable to this agreement, the agreement shall be
interpreted to make the maximum disclaimer or limitation permitted
by the applicable state law. The invalidity or unenforceability of any
provision of this agreement shall not void the remaining provisions.

1.F.6. INDEMNITY - You agree to indemnify and hold the Foundation,

the trademark owner, any agent or employee of the Foundation,
anyone providing copies of Project Gutenberg™ electronic works in
accordance with this agreement, and any volunteers associated with
the production, promotion and distribution of Project Gutenberg™
electronic works, harmless from all liability, costs and expenses,
including legal fees, that arise directly or indirectly from any of the
following which you do or cause to occur: (a) distribution of this or
any Project Gutenberg™ work, (b) alteration, modification, or
additions or deletions to any Project Gutenberg™ work, and (c) any
Defect you cause.

Section 2. Information about the Mission

of Project Gutenberg™
Project Gutenberg™ is synonymous with the free distribution of
electronic works in formats readable by the widest variety of
computers including obsolete, old, middle-aged and new computers.
It exists because of the efforts of hundreds of volunteers and
donations from people in all walks of life.

Volunteers and financial support to provide volunteers with the

assistance they need are critical to reaching Project Gutenberg™’s
goals and ensuring that the Project Gutenberg™ collection will
remain freely available for generations to come. In 2001, the Project
Gutenberg Literary Archive Foundation was created to provide a
secure and permanent future for Project Gutenberg™ and future
generations. To learn more about the Project Gutenberg Literary
Archive Foundation and how your efforts and donations can help,
see Sections 3 and 4 and the Foundation information page at
www.gutenberg.org.

Section 3. Information about the Project

Gutenberg Literary Archive Foundation
The Project Gutenberg Literary Archive Foundation is a non-profit
501(c)(3) educational corporation organized under the laws of the
state of Mississippi and granted tax exempt status by the Internal
Revenue Service. The Foundation’s EIN or federal tax identification
number is 64-6221541. Contributions to the Project Gutenberg
Literary Archive Foundation are tax deductible to the full extent
permitted by U.S. federal laws and your state’s laws.

The Foundation’s business office is located at 809 North 1500 West,

Salt Lake City, UT 84116, (801) 596-1887. Email contact links and up
to date contact information can be found at the Foundation’s website
and official page at www.gutenberg.org/contact

Section 4. Information about Donations to

the Project Gutenberg Literary Archive
Foundation
Project Gutenberg™ depends upon and cannot survive without
widespread public support and donations to carry out its mission of
increasing the number of public domain and licensed works that can
be freely distributed in machine-readable form accessible by the
widest array of equipment including outdated equipment. Many
small donations ($1 to $5,000) are particularly important to
maintaining tax exempt status with the IRS.

The Foundation is committed to complying with the laws regulating

charities and charitable donations in all 50 states of the United
States. Compliance requirements are not uniform and it takes a
considerable effort, much paperwork and many fees to meet and
keep up with these requirements. We do not solicit donations in
locations where we have not received written confirmation of
compliance. To SEND DONATIONS or determine the status of
compliance for any particular state visit www.gutenberg.org/donate.

While we cannot and do not solicit contributions from states where

we have not met the solicitation requirements, we know of no
prohibition against accepting unsolicited donations from donors in
such states who approach us with offers to donate.

International donations are gratefully accepted, but we cannot make

any statements concerning tax treatment of donations received from
outside the United States. U.S. laws alone swamp our small staff.

Please check the Project Gutenberg web pages for current donation
methods and addresses. Donations are accepted in a number of
other ways including checks, online payments and credit card
donations. To donate, please visit: www.gutenberg.org/donate.

Section 5. General Information About

Project Gutenberg™ electronic works
Professor Michael S. Hart was the originator of the Project
Gutenberg™ concept of a library of electronic works that could be
freely shared with anyone. For forty years, he produced and
distributed Project Gutenberg™ eBooks with only a loose network of
volunteer support.
Project Gutenberg™ eBooks are often created from several printed
editions, all of which are confirmed as not protected by copyright in
the U.S. unless a copyright notice is included. Thus, we do not
necessarily keep eBooks in compliance with any particular paper
edition.

Most people start at our website which has the main PG search
facility: www.gutenberg.org.

This website includes information about Project Gutenberg™,

including how to make donations to the Project Gutenberg Literary
Archive Foundation, how to help produce our new eBooks, and how
to subscribe to our email newsletter to hear about new eBooks.