350 likes | 516 Views
Panel II. Logic, Scientific Computing, Computational Biology, Algorithms and Complexity, Information Science. Grad Visiting Day March 24, 2003. Scientific Computing. Algorithms and Complexity. Computational Biology. Information Science. Applied Logic. Panel Areas and Connections.
E N D
Panel II Logic, Scientific Computing, Computational Biology, Algorithms and Complexity, Information Science Grad Visiting DayMarch 24, 2003
Scientific Computing Algorithms and Complexity Computational Biology Information Science Applied Logic Panel Areas and Connections
Artificial Intelligence Graphics Security chemistry biology Data bases vision sociology Economics Machine Learning Algorithms and Complexity Distributed Systems Computational Biology psychology Programming Languages More Connections Operations Research Scientific Computing Applied Logic Information Science
Panel Areas: • Applied Logic: • Bob Constanble, Dexter Kozen, Joe Halpern • Scientific Computing: • Charlie Van Loan, Steve Vavasis, Tom Coleman • Computational Biology: • Ron Elber, Golan Yona • Algorithms and Complexity: • Juris Hartmanis, John Hopcroft, Jon Kleinberg, Dexter Kozen, David Shmoys, Éva Tardos • Information Science • Bill Arms, Phoebe Sengers
Applied Logic @ Cornell Robert L. Constable Cornell University Grad Visiting DayMarch 24, 2003
Professors Robert Constable – Computer Science Joe Halpern – Computer Science Dexter Kozen – Computer Science Christoph Kreitz – Computer Science (joint with Potsdam) Anil Nerode – Mathematics Richard Shore – Mathematics (joint with MIT) Researchers Stuart Allen – Computer Science Mark Bickford – ORA
What Dexter Kozen does Kleene algebras ECC (Efficient Certifying Compiler)– theory: applied programming logic– practice: an implemented system Recursive types What Joe Halpern does Epistemic logic applied to:– distributed systems– security protocols Reasoning about probability What Robert Constable does Constructive type theory applied to:– program verification and synthesis– process verification and synthesis Automated reasoning with Nuprl
An Example of Applied Logic circa 70’s circa Now • Constructive proofs as programs • – Stamps • Constructive proofs as processes • – two-phase handshake protocol
Stamps Proof *T si_thm7i:{8 } . m, n:N . 3 * m + 5 * n = i|BY D 0 THENA Auto.|1. i : {8 }m, n : N. 3 * m + 5 * n = i|BY NSubsetInd 1| THEN Auto|\| 1. i: Z| 2. 0 < i| 3. 8 = i| |1 BY DTerm [1] 0 THENM DTerm [1] 0 THEN Auto \ 1. i: Z 2. 8 < i 3. m, n : N. 3 * m + 5 * n = i - 1 | BY D 3 THEN D 4 | 3. m: N4. n: N 5. 3 * m + 5 * n = i - 1 |BY Decide [n > 0] THENA Auto |\ | 6. n > 0 | | 1 BY DTerm [m + 2] 0 THENM DTerm [n – 1] 0 THEN Auto \ 6. (n > 0) | BY DTerm [m – 3] 0 THENM Dterm [n + 2] 0 THEN Auto |0 m – 3 | BY SupInf THEN Auto
Two-Phase Handshake Protocol The extracted message automaton is:
Scientific Computing @ Cornell Charlie Van Loan Cornell University Grad Visiting DayMarch 24, 2003
Scientific Computing Tom Coleman Charlie Van Loan Steve Vavasis Complexity Issues in Optimization Large-Scale Optimization Matrix Computations Computational Geometry Computational Finance Fast transforms
Connections Automatic Differentiation <---> Compilers Mesh Generation <-------------- Comp Geom / Graphics Huge Eigenproblems <----------> Network structure Subspace Computations <------> Clustering Huge/structured Ax = b <----> Machine Learning Superfast Ax = b solvers <----> Optimizing Compilers
Crack Propagation: Physics + Geometry + CS In the above mesh of triangles, the red crack is energetically favored over the blue crack. The mesh forces the blue crack to follow the stair-step dashed line which artificially increases the energy of fracture. (Bad) This problem persists no matter how much the mesh is refined.
Consider the following subdivision of a 1:2:5 triangle into five congruent subtriangles proposed by Conway and Radin Radin and Sadun showed that if this subdivision is applied recursively like this: then in the limit as the tiling is refined, all directions are equally represented.
Why are proteins shaped like this: Ron Elber Scientific computing at the molecular level.
Computational Biology @ Cornell Ron Elber Cornell University Grad Visiting DayMarch 24, 2003
Computational Biology • Who are we, what do we work on, and who are our collaborators? • Ron Elber, protein dynamics, folding, annotation, and evolution • Work with Steve Tanksley (Plant Breeding), David Shalloway (Molecular Biology & Genetics), Harold Scheraga (Chemistry and Chemical Biology), Jack Freed (Chemistry) • Jon Kleinberg, algorithms, genome rearrangements, evolution • Work with Susan McCouch (Plant Breeding) • David Shmoys, algorithms, genetic maps, population genetics. • Work with Steve Tanksley (Plant Breeding), Rasmus Nielsen (BSCB) • Golan Yona, Machine Learning, Protein classification, Micro arrays • Work with David Lin (Biomedical Sciences)
Bio-spheres in CS • Golan Yona, Klara Kedem, Paul Chew (computational geometry: structural alignments) • Ron Elber, Richard Caruana, Thorsten Joachim (Machine Learning: Protein annotation) • Ron Elber, Jon Kleinberg (Algorithms: Temperature of evolution).
Protein structures and sequences aremarkers of evolution: Golan Yona, Jon Kleinberg and Ron Elber AVLICKGGNMRQWASP GVLICKGGNMKQWASG AVLICKPGNMDQWASG AVFICKGGNMRQWASG ALLICKGGNMDQWASP LVLLCKGGNMRQWASP MGLYTHYRCCSQWAN CGLYTHYKCCSQFAN CGLYTHFRCCSQWAN CGLYSHYRCCSQWAN NMHKTTREWQLPICVDS DMHKTTREWQLQICVDS
Clustering experimentally determined protein sequences: Golan Yona
Determining potential sizes of protein families and “fingerprints” of connectivity (temperature): Ron Elber and Jon Kleinberg with students Catherine Grasso and Leonid Meyerguz Randomized algorithms Temperature for protein > 200 amino acids roughly constant suggesting that these clusters are evolutionary connected
Algorithms and Complexity @ Cornell Éva Tardos Cornell University Grad Visiting DayMarch 24, 2003
Algorithms and Complexity Jon Kleinberg John Hopcroft Juris Hartmanis Dexter Kozen Éva Tardos David Shmoys
Some Current Areas of Interest • Approximation Algorithms and Combinatorial Optimization. • Models and Algorithms for Information Access and Complex Networks. • Algorithmic Game Theory. • Complexity
Connections to Other Areas in CS • Artificial intelligence and machine learning: • heuristic algorithms, probabilistic models, clustering. • Databases and data mining. • Information Science: • Information Access and the Word Wide Web. • Distributed Computing: • Network Algorithms. • Computational biology. • Vision and image processing.
Some Current Areas of Interest wide but long What happens when individuals share a network? Algorithms for users who are selfish optimizers • Nash equilibrium: no user wants to switch paths. • Theorem: [Roughgarden-Tardos] Delay at equilibrium no worse than optimal delay with half capacity. • Properties of equilibria in other optimization problems • [Anshelevich-Dasgupta-Tardos-Wexler] network design Short, but easily congested
Information Science @ Cornell Jon Kleinberg Cornell University Grad Visiting DayMarch 24, 2003
Information Science Computer Science Cognitive Studies Applications Society HCI
Computer Science Faculty William Arms Graeme Bailey Claire Cardie Robert Constable Johannes Gehrke Joseph Halpern Daniel Huttenlocher Thorsten Joachims Jon Kleinberg Carl Lagoze Lillian Lee Bart Selman Eva Tardos Charles Van Loan
Information Retrieval: Term Vector Space TermsDocuments c1 c2 c3 c4 c5 m1 m2 m3 m4 human 1 0 0 1 0 0 0 0 0 interface 1 0 1 0 0 0 0 0 0 computer 1 1 0 0 0 0 0 0 0 user 0 1 1 0 1 0 0 0 0 system 0 1 1 2 0 0 0 0 0 response 0 1 0 0 1 0 0 0 0 time 0 1 0 0 1 0 0 0 0 EPS 0 0 1 1 0 0 0 0 0 survey 0 1 0 0 0 0 0 0 1 trees 0 0 0 0 0 1 1 1 0 graph 0 0 0 0 0 0 1 1 1 minors 0 0 0 0 0 0 0 1 1
Latent Semantic Indexing • term document query --- cosine > 0.9