department of applied mathematics and physics (dept. amp) · 2015-03-06 · mathematical...
TRANSCRIPT
Foundation In 1950s, as a subdivision of engineering with
much emphasis on applications. Dept. AMP was founded in 1959 to study interdisciplinary & traversal areas, basics & fundamentals. (rich background in math. & phys.)
Department of Applied Mathematicsand Physics (Dept. AMP)
数理工学専攻
8 groups: Applied Mathematical Analysis Discrete Mathematics System Optimization Control Systems Theory Physical Statistics Dynamical Systems Theory Mathematical Finance (Instutute of Economic Research) Applied Mathematical Modeling (Operated jointly with Industry)
22 faculty members 22-26 master course students each year 4-6 doctor course students each year
Current State of the Department
Research and Education
Research (keywords):Control, Optimization, Discrete mathematics, Dynamical system, Computational science, Simulations, Finance, Econo-physics, ...
Educational aims:Flexible conception and high competence for searching solutions with profound attainments in mathematics and physics and computer sciences.
Group 1: Applied Mathematical Analysis
New Singular Value Decomposition Algorithms
faster computing of singular value decompositionComputer Science
Discrete and ultradiscrete systemOrthogonal polynomials. Special functionsEnumerative combinatorics. Graph
Mathematics
Discrete Integrable systems
Research subjects are integrable systems and their applications to engineering science. Based on the theory of the discrete integrable systems, we are capable of developing new numerical algorithms.
Construction of Theoretical Foundation for New and Efficient Algorithms
Discrete Mathematics, Graph Theory, Optimization Theory, Complexity Theory, Data Structure, Algorithm Design
Chemical Graphs
2D, 3D Packing
Scheduling
Graph DrawingInf. Visualization
Network Design
CC
H H
H
OH
Polynomial TimeAlgorithms
Branch-and-Bound Method
ApproximationAlgorithms
Metaheuristics
Group 2: Discrete Mathematics1. Establishment of Theoretical Foundation: Develop the theoretical foundation on optimization theory and complexity theory by applying the results in graph theory and discrete mathematics.2. Design and Analysis of Algorithms: Design new efficient algorithms under the algorithm frameworks suitably selected according to the complexity hardness and required solution quality.
3. Construction of Solver Systems: Formulate new mathematical models,and construct a solver system by integrating related algorithms effectively .
2D Irregular Strip Packing
3D Molecule Packing Road Label Layout
Packing and Related Problems
2D & 3D objects, container
Non-overlap layout
Sphere Approximation Nonlinear Optimization Metaheuristics
There are a lot of objectives which we want to optimize.
Finance People want to maximize (optimize) the profit.
Transportation People try to minimize (optimize)the time or distance to the destination.
Physical system The nature tends to minimize (optimize)the potential energy of materials.
How can we find the optimal solution?
“Application” and “Theory “of Optimization
Research interests
Group 3: System Optimization
Group 4: Control Systems Theory
New paradigm of Control theoryconstraint of channel capacity and data compression
convex optimization and control theorybehavior approaches
Analysis and design of control systemconstrained controlnetworked control
hybrid system
System Identification(Modeling)
State-space representationsdifferential equations
transfer functions
Modeling and System Identification
State-space representations, differential equations, and transfer functions are models of dynamical systems. In this study, we are interested in deriving dynamical models from input-output (or output) data. Techniques such as prediction error methods, subspace methods, and stochastic realization are primal tools. Recent study includes modeling of continuous-time systems, time varying systems, and nonlinear systems.
Input
System
Model
NoiseOutput
Data-processingData-processing
A conceptual illustration of a multi-element coupled system. Each unit influences one another with different coupling strength, which is described as a weighted undirected arrow. The causality of effect is represented as a blue weighted arrow.
Group 5: Physical Statistics
A conceptual illustration of a scale free network, a kind of complex network. A black filled circle represents a node and a curve between two nodes a link. Colored nodes (blue, red, and green one), which are called hubs, have a lot of neighbors. Such a network, where a few nodes have a lot of neighbors and most nodes have a few neighbors emerges frequently in socio-economic and biological systems, and ecosystems.
Group 6: Dynamical Systems Theory
1. Geometric Mechanics: classical and quantum2. Hamiltonian Dynamics with computer simulation3. Geometric model of quantum computation4. Applications of differential geometry
Model of falling cat
The falling cat can be modeled by jointed two cylinders with two torques as control inputs under the condition of the vanishing total angular momentum.
Mathematical modeling/analysis of financial markets. Stochastic models are intensively used because of “randomness” in real financial market.
Group 7: Mathematical Finance
Some Historical References:Nobel Prize in Economics:
Markowitz, Miller, Sharpe (1990), (Black), Merton, Scholes (1997)
Gauss Prize (by IMU, 2006)Ito
Examples:Black-Scholes Stock Price model:
dS(t)= S(t) { μdt + σdW(t) }stochastic differential equation(W: Brownian motion, source of
randomness)
Derivatives pricing/hedging:A stochastic control problem* PDE approach* Probabilistic approach
(+ stochastic numerics)
Group 8: Applied Mathematical Modeling(Operated Jointly with Industry)
Information systems valued for our better life and superior productivity shall be equipped with mathematical models describing dynamical behavior of human and objects in the systems. Ranging from conceptual to precise numerical forms, modeling technology is studied including utilization of expert knowledge (structural modeling) as well as observed data (multivariate analysis) with practical industrial case studies.
■Conceptual Model
■Numerical Model∑ ∑ −+−= )()()( jtubityaty ji
Agent i
Agent j
Shop p BulletinBoard q
Visit
Buy Contributor
Read
-1.5%
-1.0%
-0.5%
0.0%
0.5%
1.0%
1.5%
2.0%
2.5%
3.0%
3.5%
02/11/5
02/11/1
2
02/11/1
9
02/11/2
6
02/12/3
02/12/1
0
02/12/1
7
02/12/2
4
02/12/3
1
03/1/7
03/1/14
03/1/21