THEORETICAL COMPUTER SCIENCE

Real World Real World ProblemsProblems

Abstract ModelsAbstract Models

Modeling

SolutionsSolutions

Math Tools

InspireMore

Problems

CAREER DIRECTIONS

Study Study AlgorithmsAlgorithms

CollegeCollege

ResearchResearchInstituteInstitute

IndustryIndustry

Network Algorithm Team

Hubert Chan

Postgraduates:

Li Ning 宁立Fei Chen 陈飞Mingfei Li 李明飞Xiaowei Wu 吴晓伟

Natural Algorithm: to Achieve Consensus

5

12

5

3

2

35

2

4

1

Observation: An individual’s opinion will be influenced by its friends/neighbors. In some cases, people will achieve a consensus finally, even when their neighborhoods keep changing.

Researcher: This kind of natural behavior is modeled by dynamic systems. By our results, it is very fast for people to achieve consensus even under dynamic networks.

“Fast convergence for consensus in dynamic networks”, Hubert Chan, Li Ning, ICALP 2011.

33.3

2.6

3

3.3

2.7 3.4

3

3

42.6

3.1

2.9

2.9

3.1

3.1

3

3.33.1

2.9

33

3

3

3

3

33

3

3

3

3

Spanners: “Building the Roads”

Government: We need roads to connect cities, and we want save the cost.Traveler: Travelling along the roads should save

time.

Researcher: This is a spanner problem. Use our methods, then you don’t need many roads, and travelling distance along the roads is almost the same with travelling directly.

“Sparse fault-tolerant spanners for doubling metrics with boundedhop-diameter or degree”, Hubert Chan, Mingfei Li, Li Ning, ICALP 2012.

Observation: People will spread information, e.g., news, via talking to their friends. Moreover, the news talked by more friends attract more attention.

Information Spreading in Social Networks

News

…

News

…News

…

News

…

Not a big deal

All my friends are talking about this news

Not a big deal

News

…

News

…

Advertiser: Persuade one person, then he will spread the advertisement… Researcher: To spread information efficiently,

which person should be picked initially? …

“Maximizing influence in information networks under the non-progressive linear threshold model”, Hubert Chan, Li Ning, ongoing project.

Bargaining

Rent sharing: The total cost to rent a flat is 100. It can be shared by two persons.

I’ll pay 50.

I’ll pay 50.

I’ll pay 40.

I’ll pay 40. I’ll pay 30. I’ll pay 40.

I’ll pay 45.

I’ll pay more... 60.

Good. I’ll pay less... 40.

I have to pay more… 60.

Sociologist: Human behavior is so complicated. What are the stable deals?

Researcher: This is studied as the problem called bargaining game.

“Optimizing Social Welfare for Network Bargaining Games in the Face of Unstability Greed and Spite”, Hubert Chan, Fei Chen, Li Ning, ESA 2012.

ONLINE ALGORITHMS: JOB SCHEDULING & POWER MANAGEMENT

Scheduling research team: T.W. Lam 林德华 H.F. Ting 田慶豐H.L. Chan 陳昊樑L.K. Lee 李立基 S.H. Chan R. Li

ONLINE PROBLEMS ARE CHALLENGING.

Robot motion

Bin Packing

Paging

Load balancing

Deadline

FlowTrading

Broadcasting

Scheduling

Responsetime

Without knowing the future, one often makes the wrong decision.

“What matters most to the computer designers at Google is not speed, but power, low power, because data centers can consume as much electricity as a city.“

Eric Schmidt, CEO Google NY Times, 2002

REPRESENTATIVE PUBLICATIONS Sleep Management on Multiple Machines for

Energy and Flow Time. ICALP 2011: 219-231 Non-clairvoyant Speed Scaling for Weighted Flow

Time. ESA 2010: 23-35 Sleep with Guilt and Work Faster to Minimize Flow

Plus Energy. ICALP 2009: 665-676 Speed scaling with an arbitrary power function.

SODA 2009: 693-701 Scheduling for Speed Bounded Processors. ICALP

2008: 409-420 Energy efficient online deadline scheduling. SODA

2007: 795-804

BIOINFORMATICS

F. Chin T.W. LamH.F. TingS.M. Yiu

Bioinformatics involves the analysis of biological and genetic information, the results of which can then be applied to gene-based drug discovery and development to cure illnesses

Our objective is to develop better software tools to aid bioinformatics research.

Next Generation Sequencing

NGS is the new & cornerstone technology for today’s biological research and tomorrow’s medical care.

Nowadays NGS equipment (Solexa, SOLiD, Illumina) can produce short fragments (read) of a DNA sequence (genome) efficiently and at a low cost.

Our research work is to base on our algorithmic expertise to develop efficient and effective software tools for NGS data.

Research Publications and FundingPublications: Bioinformatics, J of Computational Biology, RECOMB, ISMB, ECCB

GRF GrantsStructural Alignment and prediction for non-coding RNAs with triple helix structure (2011-2014, HK$681,195)

Algorithms for Inferring k-articulated Phylogenetic Network (2009-2012, HK$591,080)

Combinatorial Phenotype Testing (2009-2012, HK$394,053)

Finding Conserved Patterns In Biological Networks (2008-2012, HK$777,108)

A New Motif Representation Based on Position Specific Patterns (2006-2009, HK$775,008)

Compressed Indexes for Approximate String Matching, with Applications to Biological Sequences (2006-2008, HK$654,000)

Design and Analysis of Algorithms for Constrained Structure Comparison (2006-2008, HK$612,816)

Computationally Haplotyping Pedigree Data (2005-2007, HK$359,224)

Algorithms for Uncovering Conserved Genes on Whole Genomes (2004-2007, HK$650,000)

Finding Motifs for Sequences with Weak Binding Sites (2004-2006, HK$339,414)

Programming Languages

Programming languages research team: B. C. d. S. OliveiraX. BiH. Zhang

Programming Languages

PROGRAMMING LANGUAGES ARE FUNDAMENTAL TO PROGRAMMER PRODUCTIVITY

PROGRAMMING LANGUAGE RESEARCH AIMS AT:

- ALLOWING FASTER DEVELOPMENT CYCLES

- SUPPORTING LARGE-SCALE PROGRAMMING

- PREVENTING MORE BUGS

BY CREATING NEW PROGRAMMING LANGUAGES/ABSTRACTIONS

RESEARCH TOPICS

- BETTER PROGRAMMING MODELS FOR MULTI-CORE COMPUTING, GPU PROGRAMMING

- BETTER MODULARITY ABSTRACTIONS FOR LARGE-SCALE SOFTWARE

- FUNCTIONAL PROGRAMMING (SCALA, HASKELL, OCAML, SCHEME …)