Welcome and Thanks! - Julia · Welcome and Thanks! • Speakers (besides Chris, Joey and Miles):...

WelcomeandThanks!

• Organizingcommittee:

MilesLubin

JenniferChallisJoeyHuchette

ChrisCoey

WelcomeandThanks!

• Sponsor:

WelcomeandThanks!

• Speakers(besidesChris,JoeyandMiles):– CarletonCoffrin,LosAlamosNationalLab– StevenDiamond,Stanford– JoaquimDiasGarcia,PSR-Inc.&PUC-Rio– OscarDowson,UniversityofAuckland– JordanJalving,UniversityofWisconsin-Madison– Benoît Legat,Université Catholique deLouvain– YeeSianNg,MIT– JarrettRevels,MIT– NestorSepulveda,MIT– BartolomeoStellato,UniversityofOxford

WelcomeandThanks!

• AudienceandJulia/JuMP community:

JuliaandJuMP TutorialatUniversidadAdolfoIbáñez,Santiago,Chile.January,2014.

JuMP DevelopersWorkshop.MITSloan,Cambridge,MA.June,2017.

(My)QuickIntroto

3/9/16, 12:08 PMAcademic Page of Juan Pablo Vielma

Page 3 of 3http://www.mit.edu/~jvielma/

Jennifer ChallisE62-571, 100 Main Street,Cambridge, MA 02142(617) 324-4378jchallis at mit dot edu

CollaboratorsShabbir Ahmed, Daniel Bienstock, Daniel Dadush, Sanjeeb Dash, Santanu S. Dey, Iain Dunning, RodolfoCarvajal, Luis A. Cisternas, Miguel Constantino, Daniel Espinoza, Alexandre S. Freire, Marcos Goycoolea,Oktay Günlük, Joey Huchette, Nathalie E. Jamett, Ahmet B. Keha, Mustafa R. Kılınç, Guido Lagos, MilesLubin, Sajad Modaresi, Sina Modaresi, Eduardo Moreno, Diego Morán, Alan T. Murray, George L.Nemhauser, Luis Rademacher, David M. Ryan, Denis Saure, Alejandro Toriello, Andres Weintraub, SercanYıldız, Tauhid Zaman

Affiliations

Links

Back to top© 2013 Juan Pablo Vielma | Last updated: 02/29/2016 00:47:25 | Based on a template design by AndreasViklund

5/18/16, 12:21 PMJuliaOpt - Optimization packages for the Julia language

Page 1 of 3http://www.juliaopt.org/

JuliaOptOptimization packages for the Julia language. (http://julialang.org)

What is Julia?"Julia is a high-level, high-performance dynamic programming language for technicalcomputing" (http://julialang.org). It is free (open source) and supports Windows, OSX,and Linux. It has a familiar syntax, works well with external libraries, is fast, and hasadvanced language features like metaprogramming that enable interesting possibilitiesfor optimization software.

Case study: "Computing in Operations Research using Julia"(http://dx.doi.org/10.1287/ijoc.2014.0623). INFORMS Journal on Computing. [PDF](http://arxiv.org/abs/1312.1431).

What is JuliaOpt?JuliaOpt is an organization that brings together packages written in Julia that are relatedto optimization. All JuliaOpt packages should be high-quality, documented, tested,support the main operating systems, and interact with each other.

Overview: presentation (https://www.youtube.com/watch?v=7LNeR299q88) andworkshop (https://www.youtube.com/watch?v=nnL7yLMVu6c) from JuliaCon2015Code: github.com/JuliaOpt (http://github.com/JuliaOpt)Mailing list: julia-opt (https://groups.google.com/forum/#!forum/julia-opt)

Getting Started1. You can find downloads and installation instructions for Julia

(http://julialang.org/downloads/) for your operating system on the Julia site.2. To install a JuliaOpt package, simply use the Pkg.add command from inside

Julia:julia> Pkg.update() # Get latest package infojulia> Pkg.add("Optim")julia> Pkg.add("JuMP")julia> Pkg.add("Cbc") # and so on

3. Open-source solvers will automatically be downloaded and installed in your Juliapackage directory. If you want to use an external commercial solver, you willneed to download those before installing the corresponding JuliaOpt package.

Examples

We have a collection of JuliaOpt examples in the form of Jupyter(https://jupyter.org/)/IJulia (https://github.com/JuliaLang/IJulia.jl) notebooks, including:

Solving Sudoku puzzles with JuMP (/notebooks/JuMP-Sudoku.html),Controlling a rocket with JuMP (/notebooks/JuMP-Rocket.html),and many more (/notebooks/index.html).

The source for these notebooks is available here (https://github.com/JuliaOpt/juliaopt-notebooks).

Overview of PackagesJuMP Convex.jl

MathProgBase.jl

Cbc.jl Clp.jl CPLEX.jl

ECOS.jl GLPK.jl Gurobi.jl

Ipopt.jl KNITRO.jl Mosek.jl

NLopt.jl SCS.jl

Optim.jl

LsqFit.jl

CoinOptServices.jl

AmplNLWriter.jl

JuliaOpt's packages can be loosely grouped into two sets. The first set are standalone Julia packages:












Examples





MathProgBase.jl




NLopt.jl SCS.jl

Optim.jl

LsqFit.jl

CoinOptServices.jl

AmplNLWriter.jl


&

• User-friendlyalgebraicmodellinglanguages(AML):

• CommercialandOpen-SourceC/C++frameworks:

20th CenturyOptimizationModellingLanguages

StandaloneandFast BasedonGeneralLanguageandVersatile,butSlow

5/25/17, 11(03 AM

Page 1 of 1file:///Users/jvielma/delme/gams.svg

FastandVersatile,butcomplicated(andsomeproprietary)

SCIP

Gurobi 7.0 Performance Benchmarks

21st CenturyOptimizationModellingLanguage

• Modern,modular,easytoembed...– Withinasimulation,interactivevisualization,etc.

• Interactwithsolverswhiletheyarerunning

• Easytoextendtospecializedproblemclasses





Affiliations

Links






Affiliations

Links

Back to top© 2013 Juan Pablo Vielma | Last updated: 02/29/2016 00:47:25 | Based on a template design by AndreasViklundGreedyWishList:

LargeSoftwareStackandVibrantCommunity












Examples





MathProgBase.jl




NLopt.jl SCS.jl

Optim.jl

LsqFit.jl

CoinOptServices.jl

AmplNLWriter.jl













Examples





MathProgBase.jl




NLopt.jl SCS.jl

Optim.jl

LsqFit.jl

CoinOptServices.jl

AmplNLWriter.jl


SustainedGrowthandSomeMilesStones

• Version0.2:Solvercallbacks(December,2013)• Version0.5:Nonlinearoptimization(May,2014)• Version0.10:Semidefiniteoptimization(August,2015)• Version0.12:Rewrotenonlinearoptimization(February,2016)• Version0.13:RenamedeverythingfromcamelCase (April,2016)• Version0.15:sum{} becomessum() (December,2016)

Hits to JuMP’s documentation





Affiliations

Links


JuliaOpt inUniversityCourses

SeriousUseofJuliaOpt

Article outl ineArticle outl ine Show full outline

AbstractKeywords1. Introduction2. Methods3. Results4. ConclusionsFundingAppendix A. Mathematical Proofs of the …References

Figures and tablesFigures and tables

https://doi.org/10.1016/j.compbiolchem.2017.03.016

Computational Biology and ChemistryAvailable online 3 April 2017

In Press, Corrected Proof — Note to users

Research Article

Optimal hybrid sequencing and assembly: Feasibilityconditions for accurate genome reconstruction and costminimization strategy

Chun-Chi Chena, Noushin Ghaffarib, Xiaoning Qiana, , , Byung-Jun Yoona, ,

Show more

Abstract

Recent advances in high-throughput genome sequencing technologies have enabled thesystematic study of various genomes by making whole genome sequencing affordable.Modern sequencers generate a huge number of small sequence fragments called reads,where the read length and the per-base sequencing cost depend on the technologyused. To date, many hybrid genome assembly algorithms have been developed that cantake reads from multiple read sources to reconstruct the original genome. However,rigorous investigation of the feasibility conditions for complete genome reconstructionand the optimal sequencing strategy for minimizing the sequencing cost has beenconspicuously missing. An important aspect of hybrid sequencing and assembly is thatthe feasibility conditions for genome reconstruction can be satisfied by differentcombinations of the available read sources, opening up the possibility of optimallycombining the sources to minimize the sequencing cost while ensuring accurate genomereconstruction. In this paper, we derive the conditions for whole genome reconstructionfrom multiple read sources at a given confidence level and also introduce the optimalstrategy for combining reads from different sources to minimize the overall sequencingcost. We show that the optimal read set, which simultaneously satisfies the feasibilityconditions for genome reconstruction and minimizes the sequencing cost, can beeffectively predicted through constrained discrete optimization. Through extensiveevaluations based on several genomes and different read sets, we verify the derivedfeasibility conditions and demonstrate the performance of the proposed optimal hybridsequencing and assembly strategy.

Keywords

Assembly; Sequencing; Whole genome reconstruction

Choose an option to locate/access this article:

Get rights and content

Check if you have accessthrough your login credentialsor your institution

Sign In

PurchasePurchase

Search ScienceDirect Advanced searchPurchasePurchase Export

Journals Books Register Sign in

This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyrightSimultaneous identification of specifically interactingparalogs and interprotein contacts by directcoupling analysisThomas Gueudréa,1, Carlo Baldassia,b,1, Marco Zamparoa, Martin Weigtc,2, and Andrea Pagnania,b,2

aDepartment of Applied Science and Technology, Politecnico di Torino, 10129 Torino, Italy; bHuman Genetics Foundation, Molecular Biotechnology Center,10126 Torino, Italy; and cSorbonne Universités, UPMC Université Paris 06, CNRS, Biologie Computationnelle et Quantitative, Institut de Biologie Paris Seine,75005 Paris, France

Edited by Barry Honig, Howard Hughes Medical Institute, Columbia University, New York, NY, and approved September 16, 2016 (received for review May12, 2016)

Understanding protein−protein interactions is central to our under-standing of almost all complex biological processes. Computationaltools exploiting rapidly growing genomic databases to characterizeprotein−protein interactions are urgently needed. Such methodsshould connect multiple scales from evolutionary conserved interac-tions between families of homologous proteins, over the identifica-tion of specifically interacting proteins in the case of multipleparalogs inside a species, down to the prediction of residues beingin physical contact across interaction interfaces. Statistical inferencemethods detecting residue−residue coevolution have recently trig-gered considerable progress in using sequence data for quaternaryprotein structure prediction; they require, however, large joint align-ments of homologous protein pairs known to interact. The genera-tion of such alignments is a complex computational task on its own;application of coevolutionary modeling has, in turn, been restrictedto proteins without paralogs, or to bacterial systems with the corre-sponding coding genes being colocalized in operons. Here we showthat the direct coupling analysis of residue coevolution can be ex-tended to connect the different scales, and simultaneously to matchinteracting paralogs, to identify interprotein residue−residue con-tacts and to discriminate interacting from noninteracting families ina multiprotein system. Our results extend the potential applicationsof coevolutionary analysis far beyond cases treatable so far.

coevolution | protein−protein interaction networks | paralog matching |statistical inference | direct coupling analysis

Almost all biological processes depend on interacting proteins.Understanding protein−protein interactions is therefore key

to our understanding of complex biological systems. In this con-text, at least two questions are of interest: First, the question “whowith whom,” i.e., which proteins interact; this concerns the net-works connecting specific proteins inside one organism, but also—in the context of this article—the evolutionary perspective ofprotein−protein interactions, which are conserved across differentspecies. Their coevolution is at the basis of many modern com-putational techniques for characterizing protein−protein interac-tions. The second question is the question “how” proteins interactwith each other, in particular, which residues are involved in theinteraction interfaces, and which residues are in contact across theinterfaces. Such knowledge may provide important mechanisticinsight into questions related to interaction specificity or com-petitive interaction with partially shared interfaces.The experimental identification of protein−protein interactions

is an arduous task (for reviews, cf. refs. 1 and 2): High-throughputtechniques that aim to identify protein−protein interactions invivo or in vitro are well documented and include large-scale yeasttwo-hybrid assays and protein affinity mass spectrometry assays.Such large-scale efforts have revealed useful information but arehampered by high false positive and false negative error rates.Structural approaches based on protein cocrystallization are in-trinsically low-throughput and of uncertain outcome due to the

unphysiological treatment needed for protein purification, enrich-ment, and crystallization. It is therefore tempting to use the ex-ponentially increasing genomic databases to design in silicotechniques for identifying protein−protein interactions (cf. refs. 3and 4). Prominent techniques, to date, include the search forcolocalization of genes on the genome (e.g., operons in bacteria)(5, 6), the Rosetta stone method (domains fused to a single proteinin some genome are expected to interact in other genomes) (7, 8),and also coevolutionary techniques like phylogenetic profiling(correlated presence or absence of interacting proteins in ge-nomes) (9) or similarities between phylogenetic trees of groups oforthologous proteins (compare the mirrortree method) (10, 11).Despite the success of all these methods, their sensitivity is limiteddue to the use of relatively coarse global criteria (genomic location,phylogenetic distance) instead of full amino acid sequences.The availability of thousands of sequenced genomes (12),

thanks to next-generation sequencing techniques, enables theapplication of much finer-scale statistical modeling approaches,which take into account the full sequence (13). In this context,direct coupling analysis (DCA) (14) was developed to detect directinterprotein coevolution and, in turn, interprotein residue−residue

Significance

Most biological processes rely on specific interactions betweenproteins, but the experimental characterization of protein−pro-tein interactions is a labor-intensive task of frequently uncertainoutcome. Computational methods based on exponentially grow-ing genomic databases are urgently needed. It has recently beenshown that coevolutionary methods are able to detect correlatedmutations between residues in different proteins, which are incontact across the interaction interface, thus enabling the struc-ture prediction of protein complexes. Here we show that theapplicability of coevolutionary methods is much broader, con-necting multiple scales relevant in protein−protein interaction:the residue scale of interprotein contacts, the protein scale ofspecific interactions between paralogous proteins, and the evo-lutionary scale of conserved interactions between homologousprotein families.

Author contributions: T.G., C.B., M.W., and A.P. designed research; T.G., C.B., M.Z., M.W.,and A.P. performed research; T.G. and C.B. analyzed data; and M.W. and A.P. wrotethe paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Data deposition: Julia package for paralog matching is available at https://github.com/Mirmu/ParalogMatching.jl.1T.G. and C.B. contributed equally to this work.2To whom correspondence may be addressed. Email: [email protected] or [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1607570113/-/DCSupplemental.

12186–12191 | PNAS | October 25, 2016 | vol. 113 | no. 43 www.pnas.org/cgi/doi/10.1073/pnas.1607570113

NotSoSeriousUseofJuliaOpt

MIT Sloan Sports Analytics Conference 2012 March 2-3, 2012, Boston, MA, USA

1

Big 2’s and Big 3’s: Analyzing How a Team’s Best Players Complement Each Other

Robert Ayer MBA 2011

Massachusetts Institute of Technology Cambridge, MA

Email: [email protected]

Abstract One of the most important aspects of team construction is identifying and acquiring the most talented and productive players on your team, the players on whom a team’s fortunes most rely. Teams must decide which player-types, when combined, yield the best fit. As an example, suppose there is a team, whose current best player is a scoring, shoot-first point guard. Suppose this team is looking to bring in a top-flight free agent. What type of player should this team target? Should they bring in a defense-oriented big man? Should they acquire a multi-faceted, jack-of-all-trades wing? This paper aims to answer these questions. Analyzing player data and team season data from 1977, this paper first uses clustering techniques to group players into appropriate groups, then regression to determine the degree to which the composition of a team’s top 2 and top 3 players affect that team’s win total, while accounting for team quality and coaching ability. This paper shows that the composition of a team’s top 2 and top 3 players is a strongly statistically significant factor in the success of a team, and shows which combinations yield over-performance, and which combinations yield underperformance, relative to the team’s talent and coaching quality.

1 Introduction There have been many instances in basketball where a team, with perhaps a collection of new acquisitions, underperforms relative to the perceived talent on the team. Most observers will intuitively conclude that this team, while talented, just doesn’t fit well together. Conversely, there are also many instances where a team, with perhaps relatively modest top level talent, exceeds expectations. This team, most will conclude, is put together well, i.e., the players complement each other, they “fit.” Similarly, at the player level, there are many cases in which a new player on a team, perhaps acquired through free agency, who, though talented, fails to live up to expectations. This can be attributed to many things: lack of effort, erosion of skills, poor scouting. However, many times a player’s underperformance is attributed to a poor fit with the team. This is to say, that it is not enough for a player to have valuable basketball skills to fully reach his potential; the player must also be on a team which is constructed in a way that is complementary to those skills. This paper aims to provide some insight into team construction and player fit by analyzing combinations of player types (specifically, 2 and 3 man combinations, known commonly as “Big 2’s or Big 3’s), and determining which combinations lead to increased wins, while accounting for talent level and coaching skill. Along the way, the research uncovered some interesting insights regarding coaching ability; while those insights will be addressed briefly, this is not a focus of the paper.

Picking Winners Using Integer ProgrammingDavid Scott Hunter, Juan Pablo Vielma, Tauhid ZamanarXiv:1604.01455,https://github.com/dscotthunter/Fantasy-Hockey-IP-Code

2016ICSPrizeforJuMP

• “ComputinginoperationsresearchusingJulia”,M.Lubin,I.Dunning.INFORMSJournalonComputing27(2),238-248.

• “JuMP:Amodelinglanguageformathematicaloptimization”,I.Dunning,J.Huchette,M.Lubin.SIAMReview59(2),295-320

“JuMP’s designleveragesadvancedfeaturesoftheJulialanguagetoofferdistinctivefunctionalitywhileachievingperformanceininstancecreationoftensimilartocommercialmodelingtools.”

CongratulationsDr.Lubin!Office

E62-561, 100 Main Street,Cambridge, MA 02142

Support Staff


CollaboratorsShabbir Ahmed, Russell Bent, Daniel Bienstock, Daniel Dadush, Sanjeeb Dash, Santanu S. Dey, IainDunning, Rodolfo Carvajal, Luis A. Cisternas, Chris Coey, Miguel Constantino, Daniel Espinoza, AlexandreS. Freire, Marcos Goycoolea, Oktay Günlük, Joey Huchette, David Scott Hunter, Nathalie E. Jamett, AhmetB. Keha, Mustafa R. Kılınç, Guido Lagos, Miles Lubin, Sajad Modaresi, Sina Modaresi, Eduardo Moreno,Diego Morán, Carter Mundell, Alan T. Murray, George L. Nemhauser, Luis Rademacher, David M. Ryan,Denis Saure, Alejandro Toriello, Andres Weintraub, Sercan Yıldız, Emre Yamangil, Ilias Zadik, TauhidZaman

Affiliations

Links

Mixed-integer convex optimization: outerapproximation algorithms and modeling power

by

Miles LubinB.S., The University of Chicago (2011)M.S., The University of Chicago (2011)

Submitted to the Sloan School of Managementin partial fulfillment of the requirements for the degree of

Doctor of Philosophy

at the

MASSACHUSETTS INSTITUTE OF TECHNOLOGY

September 2017

c� Massachusetts Institute of Technology 2017. All rights reserved.

Author . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Sloan School of Management

June XX, 2017

Certified by. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Juan Pablo Vielma

Assistant Professor of Operations ResearchThesis Supervisor

Accepted by . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Dimitris Bertsimas

Boeing Professor of Operations ResearchCo-director, Operations Research Center

http://youtu.be/Vifryc-Iqao

WelcomeJarrett!

“Sponsored Research Technical Staff”





Affiliations

Links


SustainabilityPlanforJuMP (andJulia)

MIT Open Source CommercialAcademic Community Commercial

Remain(Primarily)Open-Source

MoreinLatinAmerica





Affiliations

Links


SantiagoinAugust

@J_P_Vielma

Welcome and Thanks! - Julia · Welcome and Thanks! • Speakers (besides Chris, Joey and Miles):...

Documents

Transcript of Welcome and Thanks! - Julia · Welcome and Thanks! • Speakers (besides Chris, Joey and Miles):...