Sequential Investment, Universal Portfolio Algos and Log-loss

Size: px
Start display at page:

Download "Sequential Investment, Universal Portfolio Algos and Log-loss"

Transcription

1 1/37 Sequential Investment, Universal Portfolio Algos and Log-loss Chaitanya Ryali, ECE UCSD March 3, 2014

2 Table of contents 2/

3 Definitions and Notations 3/37 A market vector x = {x 1,x 2,...,x m } for m assets is a vector of nonnegative real numbers representing price relatives for a given trading period. x i 0 denotes the ratio of closing to opening price of the ith asset for that period. An initial wealth invested in m assets according to the fractions Q 1,Q 2,...,Q m multiplies by a factor of m i=1 x iq i at the end of the period. Market behavior during n trading periods is represented by a sequence of market vectors x n = ( x 1, x 2,..., x n ).

4 Definitions and Notations 4/37 The probability simplex in R m is denoted by m 1. An investment strategy Q for n trading periods is a sequence Q 1,..., Q n of vector valued functions t : R Q t 1 + m 1 ith component Q i,t ( x t 1 )of vector t ( x Q t 1 ) denotes the fraction of the current wealth invested in the i th asset at the beginning of the tth period on the basis of the past market behavior x t 1

5 Wealth Factor 5/37 S n (Q, x n ) = n m ( x i,t Q i,t ( x t 1 )) (1) t=1 denotes the wealth factor of strategy Q after n trading periods. i=1 Q t has nonnegative component summing to one expresses no short sales and no buying on margin.

6 Examples 6/37 Buy-and-Hold: S n (Q, x n ) = m n Q j,1 j=1 max t=1 n x j,t x j,t j=1,...,m t=1

7 Examples 7/37 Constantly Rebalanced Portfolios: Parametrized by a probability vector B = (B 1,B 2,...,B m ) m 1 Q t ( x t 1 ) = regardless of t and B x t 1 n m S n ( B, x n ) = ( x i,t B i ). t=1 i=1 Example: (1, 1 2 ),(1,2),(1, 1 2 ),(1,2),... Buy and Hold No profit, No loss CRP : B = ( 1 2, 1 2 ) (9 8 )n/2, exponentially increasing wealth.

8 Minimax Wealth Ratio 8/37 Given a class Q of investment strategies, the worst case logarithmic wealth ratio of a strategy P is given by W n (P,Q) = sup sup xn Q Q ln S n(q, x n ) S n (P, x n ). Minimax logarithmic wealth ratio is defined as: W n (Q) = inf P W n(p,q). W n (P,Q) = o(n) means strategy P achieves the same exponent of growth as the best reference strategy in class Q for all market behaviors.

9 Prediction under log-loss and Investment 9/37 Any investment strategy Q can be used define a forecaster that predicts elements y t Y{1,...,m} of a sequence y n Y n with probability vectors ˆp t m 1 Kelly Market Vectors: Market vectors x with a single component equal to 1 and all other components equal to zero. If x 1,..., x n are Kelly market vectors, we denote the index of the only non zero component of each vector x t by y t, we may define a forecaster f by f t (y y t 1 ) = Q y,t ( x t 1 ). f is induced by investment strategy Q.

10 Prediction under log-loss and Investment 10/37 When x n is a sequence of Kelly vectors determined by the indices y n, we write S n (Q,y n ) for S n (Q, x n ). Note that S n (Q,y n ) = f n (y n ), where f is the forecaster induced by Q, where f n (y n ) = n t=1 f t(y t y t 1 ), where y n Y n f n(y n ) = 1. Conversely, given a f n (y n ), we may define f t (y t y t 1 ) = f t(y t ) f t 1 (y t 1 ), where f t (y t ) = y n t+1 Yn t f n(y n ).

11 Prediction under log-loss and Investment 11/37 Log-loss: l(f t,y t ) = lnf t (y t y t 1 ) Regret against a reference forecaster f is ˆL n L f,n = ln f n(y n ) ˆp n (y n ) = ln Q(yn ) P(y n ), where Q and P are the investment strategies induced by f and ˆp.

12 12/37 Lemma Let Q be a class of investment strategies, and let F denote the class of forecasters induced by the strategies in Q. Then, the minimax regret satisfies W n (Q) V n (F). V n (F) = infsupsup p n y n f F ln f n(y n ) p n (y n )

13 13/37 Proof. Let P be any investment strategy and let p be it s induced forecaster. Then sup sup ln S n(q, x n ) xn Q Q S n (P, x n ) max sup ln S n(q,y n ) y n Y n Q Q S n (P,y n ) = max sup ln f n(y n ) y n Y n f F p n (y n ) = V n (p,f) V n (F).

14 14/37 Given a prediction p, we define an investment strategy P as follows: P j,t ( x t 1 ) = y t 1 Y t 1p t(j y t 1 )p t 1 (y t 1 )( t 1 s=1 x y s,s) y t 1 Y t 1p t 1(y t 1 )( t 1 s=1 x y s,s) The obtained investment strategy induces p, and so we say p and P induce each other. t 1 s=1 x y s,s may be viewed as the return of the extremal investment strategy that, on each trading period t, invests everything on the y t th asset.

15 15/37 Theorem Let P be an investment strategy induced by a forecaster p, and let Q be an arbitrary class of investment strategies. Then for any market sequence x n, sup ln S n(q, x n ) Q Q S n (P, x n ) max sup ln y n Y n Q Q n t=1 Q y,t( x t 1 ) p n (y n )

16 16/37 Lemma let a 1,...,a n, b 1,...,b n be non negative numbers. Then, where we define 0/0 = 0. n i=1 a i n i=1 b i a j max, j=1,...,n b j

17 17/37 Lemma The wealth factor achieved by an investment strategy Q may be written as S n (Q, x n ) = ( y n Y n t=1 n n x yt,t)( Q yt,t( x t 1 )). If the investment strategy P is induced by a forecaster p n, then S n (P, x n ) = ( y n Y n t=1 t=1 n x yt,t)p n (y n ).

18 18/37 Proof. S n (Q, x n ) = n m ( x j,t Q j,t ( x t 1 )) t=1 j=1 = y n Y n ( = y n Y n ( n x yt,tq yt,t( x t 1 )) t=1 n n x yt,t)( Q yt,t( x t 1 )). t=1 t=1

19 19/37 Proof. S n (P, x n ) = = n m ( x j,t P j,t ( x t 1 )) t=1 n t=1 j=1 m j=1 y t 1 Y t 1p t(y t 1 j)x j,t ( t 1 s=1 x y s,s) y t 1 Y t 1p t(y t 1 j)x j,t ( t 1 n y = t Y t( t s=1 x y s,s)p t (y t ) t=1 y t 1 Y t 1( t 1 s=1 x y s,s)p t 1 (y t 1 ) = n x yt,t)p n (y n ). ( y n Y n t=1 s=1 x y s,s)

20 20/37 Proof. Fix any market sequence x n and choose any reference strategy Q Q. Denote by S n (y n, x n ) = n t=1 x y t,t, then S n (Q, x n ) S n (P, x n ) = y n Y n S n(y n, x n )( n t=1 Q y t,t ( xt 1 )) y n Y n S n(y n, x n )p n (y n ) S n (y n, x n )( n t=1 max Q y t,t( x t 1 )) S n (y n, x n )p n (y n ) n t=1 Q y t,t ( xt 1 ) y n :S n(y n, x n )>0 = max y n Y n max y n Y n sup q Q p n (y n ) n t=1 Q y t,t( x t 1 ) p n (y n. )

21 21/37 Theorem Let Q be a class of static investment strategies, and let F denote the class of forecasters induced by strategies in Q. Then W n (Q) = V n (F). Furthermore, the minimax optimal investment strategy is defined by P j,t( x t 1 ) = y t 1 Y t 1p t (j yt 1 )p t 1 (yt 1 )( t 1 s=1 x y s,s) y t 1 Y t 1p t 1 (yt 1 )( t 1 s=1 x y s,s) where p is the normalized maximum likelihood forecaster p n(y n ) = sup n Q Q t=1 Q y t,t y n Y n sup n Q Q t=1 Q. y t,t

22 Normalized Maximum Likelihood Forecaster 22/37 Definition The normalized maximum likelihood forecaster is defined by the following: pn sup (yn ) = f F f n (y n ) x n Y n sup f F f n (y n )

23 Normalized Maximum Likelihood Forecaster 23/37 Theorem For any class F of experts and integer n > 0, the normalized maximum likelihood forecaster p is the unique forecaster such that sup (ˆL(y n ) inf L f(y n )) = V n (F). y n Y n f F Moreover, p is an equalizer that is, for all y n Y n, ln sup f F f n (y n ) p n (yn ) = ln x n Y n sup f F f n (x n ) = V n (F).

24 24/37 Proof. The normalized maximum likelihood forecaster p is minimax optimal for the class F; that is, max y n Ynln sup Q Q n t=1 Q y t,t p n (yn ) = V n (F). Now, let P be the investment strategy induced by minimax forecaster p for Q. By theorem, we get W n (Q) sup sup ln S n(q, x n ) xn Q Q S n (P, x n ) max sup ln y n Y n Q Q n t=1 Q y t,t p n (yn ) = V n (F)

25 Constantly Rebalanced Portfolios 25/37 W n (Q) = m 1 2 lnn+ln Γ(1/2)m Γ(m/2) +o(1)

26 26/37 We restrict our attention to class Q of all constantly rebalanced portfolios. Each strategy Q in this class is determined by a vector B = {B 1,B 2,...,B m } m 1 The Universal Portfolio strategy P is given by P j,t ( x t 1 ) = m 1 B j S t 1 ( B, x t 1 )µ( B)d B m 1 S t 1 ( B, x t 1 )µ( B)d B, j = 1,2...,m, t = 1,...,n, and µ is a density function on m 1.

27 27/37 The wealth achieved by the universal portfolio is just the average of the wealths achieved by the individual strategies in the class. S n (P, x n ) = = n t=1 j=1 n t=1 m P j,t ( x t 1 )x j,t m m 1 j=1 x j,tb j S t 1 ( B, x t 1 )µ( B)d B m 1 S t 1 ( B, x t 1 )µ( B)d B n = m 1 S t ( B, x t )µ( B)d B t=1 m 1 S t 1 ( B, x t 1 )µ( B)d B = S n ( B, x n )µ( B)d B m 1

28 28/37 It s like a Buy-and-Hold on all Constantly Rebalanced Portfolios (CRP) S n (P, x n ) = S n ( B, x n )µ( B)d B m 1 If it helps, think of it as: (Riemann sum approximation) S n (P, x n ) = i Q i S n ( B i, x n ), where, given the elements i of a fine partition of the simplex m 1, we assume that B i m 1 and Q i = i µ( B)d B.

29 29/37 Theorem If µ is the uniform density on the probability density simplex m 1 R m, then the wealth achieved by the universal portfolio satisfies sup xn sup ln S n( B, x n ) B m 1 S n (P, x n (m 1)ln(n+1). ) If the universal portfolio is defined using the Dirichlet(1/2,...,1/2) density µ, then sup xn sup ln S n( B, x n ) B m 1 S n (P, x n ) m 1 2 lnn+ln Γ(1/2)m Γ(m/2) +m 1 ln2+o(1). 2

30 30/37 Universal Portfolio involves integration over m-dimensional simplex. EG s computational cost is linear in m invests at a time t using the vector P t = (P 1,t,...,P m,t ) where t = (1/m,...,1/m) and P P i,t 1 exp(η(x i,t 1 / P t 1 x t 1 )) P i,t = m j=1 P j,t 1exp(η(x j,t 1 / P t 1 x t 1 )) where i = 1,2,...,m and t = 2,3,...

31 31/37 Special case of gradient-based forecaster: P i,t = P i,t 1 exp(η l t 1 ( P t 1 ) i ) m j=1 P j,t 1exp(η l t 1 ( P t 1 ) j ) when the loss function is set as l t 1 ( P t 1 ) = ln P t 1 x t 1.

32 32/37 Theorem Assume that the price relatives x i,t all fall between two positive constants c < C. Then the worst-case logarithmic wealth ratio of the with η = (c/c) (8lnm)/n is bounded by lnm η + nη C 2 8 c 2 = C n c 2 lnm.

33 Simple proof without transaction costs 33/37 Main Idea: Portfolios that are near each other perform similarly, and there is a large fraction of portfolios near the optimal one. Suppose in hindsight B is the optimal CRP. Let B = (1 α) B +α z, for some m 1.(Meaning, is z B close to B ). For a single period Over n periods, gain of CRP B (1 α)(gain ofcrp B ). wealth of CRP B (1 α) n (wealth of CRP B ).

34 Simple proof without transaction costs wealth of UNIVERSAL wealth of best CRP E B m 1 [(1 α)n ] = = = n = Prob B m 1 [(1 α) n x]dx (1 x 1/n ) m 1 dx 0 y n 1 (1 y) m 1 dy (m 1)!(n 1)! = n(!) n+m 2 1 = ) ( n+m 1 m 1 34/37

35 Commission 35/37 The following assumptions are made: The costs paid changing from distribution B 1 to B 3 is no more than the costs paid changing from B 1 to B 2 and then from B 2 to B 3. The cost, per dollars, of changing from a distribution B to a distribution (1 α) B 1 +α B is no more than αc, because at most an α fraction of the money is being moved. An investment strategy I which invests an initial fraction α of it s money according to investment strategy I 1 and an initial 1 α of it s money according to I 2, will achieve at least α times the wealth of I 1 plus 1 α times the wealth of I 2.

36 Result with Commission 36/37 Theorem In the presence of commission 0 c 1, wealth of UNIVERSAL c wealth of best CRP ( (1+c)n+m 1 m 1 1 ((1+c)n+1) m 1. ) 1

37 Result with Commission 37/37 Proof. Based on the properties we assumed, if B j (1 α)b j, then single-period profit of CRP single-period profit of CRP Over n periods, this gives B B (1 α)(1 cα). wealth of CRP B (1 α) (1+c)n (wealth of CRP B ). The previous proof can be applied and we can replace n by (1+c)nin the final guarantee.

Learning Methods for Online Prediction Problems. Peter Bartlett Statistics and EECS UC Berkeley

Learning Methods for Online Prediction Problems. Peter Bartlett Statistics and EECS UC Berkeley Learning Methods for Online Prediction Problems Peter Bartlett Statistics and EECS UC Berkeley Course Synopsis A finite comparison class: A = {1,..., m}. Converting online to batch. Online convex optimization.

More information

Learning Methods for Online Prediction Problems. Peter Bartlett Statistics and EECS UC Berkeley

Learning Methods for Online Prediction Problems. Peter Bartlett Statistics and EECS UC Berkeley Learning Methods for Online Prediction Problems Peter Bartlett Statistics and EECS UC Berkeley Online Learning Repeated game: Aim: minimize ˆL n = Decision method plays a t World reveals l t L n l t (a

More information

Internal Regret in On-line Portfolio Selection

Internal Regret in On-line Portfolio Selection Internal Regret in On-line Portfolio Selection Gilles Stoltz (gilles.stoltz@ens.fr) Département de Mathématiques et Applications, Ecole Normale Supérieure, 75005 Paris, France Gábor Lugosi (lugosi@upf.es)

More information

Internal Regret in On-line Portfolio Selection

Internal Regret in On-line Portfolio Selection Internal Regret in On-line Portfolio Selection Gilles Stoltz gilles.stoltz@ens.fr) Département de Mathématiques et Applications, Ecole Normale Supérieure, 75005 Paris, France Gábor Lugosi lugosi@upf.es)

More information

Efficient Algorithms for Universal Portfolios

Efficient Algorithms for Universal Portfolios Efficient Algorithms for Universal Portfolios Adam Kalai CMU Department of Computer Science akalai@cs.cmu.edu Santosh Vempala y MIT Department of Mathematics and Laboratory for Computer Science vempala@math.mit.edu

More information

Sequential prediction with coded side information under logarithmic loss

Sequential prediction with coded side information under logarithmic loss under logarithmic loss Yanina Shkel Department of Electrical Engineering Princeton University Princeton, NJ 08544, USA Maxim Raginsky Department of Electrical and Computer Engineering Coordinated Science

More information

Meta Algorithms for Portfolio Selection. Technical Report

Meta Algorithms for Portfolio Selection. Technical Report Meta Algorithms for Portfolio Selection Technical Report Department of Computer Science and Engineering University of Minnesota 4-192 EECS Building 200 Union Street SE Minneapolis, MN 55455-0159 USA TR

More information

Portfolio Optimization

Portfolio Optimization Statistical Techniques in Robotics (16-831, F12) Lecture#12 (Monday October 8) Portfolio Optimization Lecturer: Drew Bagnell Scribe: Ji Zhang 1 1 Portfolio Optimization - No Regret Portfolio We want to

More information

MATH 236 ELAC FALL 2017 CA 9 NAME: SHORT ANSWER. Write the word or phrase that best completes each statement or answers the question.

MATH 236 ELAC FALL 2017 CA 9 NAME: SHORT ANSWER. Write the word or phrase that best completes each statement or answers the question. MATH 236 ELAC FALL 207 CA 9 NAME: SHORT ANSWER. Write the word or phrase that best completes each statement or answers the question. ) 27 p 3 27 p 3 ) 2) If 9 t 3 4t 9-2t = 3, find t. 2) Solve the equation.

More information

Tutorial: PART 1. Online Convex Optimization, A Game- Theoretic Approach to Learning.

Tutorial: PART 1. Online Convex Optimization, A Game- Theoretic Approach to Learning. Tutorial: PART 1 Online Convex Optimization, A Game- Theoretic Approach to Learning http://www.cs.princeton.edu/~ehazan/tutorial/tutorial.htm Elad Hazan Princeton University Satyen Kale Yahoo Research

More information

Shadow prices and well-posedness in the problem of optimal investment and consumption with transaction costs

Shadow prices and well-posedness in the problem of optimal investment and consumption with transaction costs Shadow prices and well-posedness in the problem of optimal investment and consumption with transaction costs Mihai Sîrbu, The University of Texas at Austin based on joint work with Jin Hyuk Choi and Gordan

More information

ln(9 4x 5 = ln(75) (4x 5) ln(9) = ln(75) 4x 5 = ln(75) ln(9) ln(75) ln(9) = 1. You don t have to simplify the exact e x + 4e x

ln(9 4x 5 = ln(75) (4x 5) ln(9) = ln(75) 4x 5 = ln(75) ln(9) ln(75) ln(9) = 1. You don t have to simplify the exact e x + 4e x Math 11. Exponential and Logarithmic Equations Fall 016 Instructions. Work in groups of 3 to solve the following problems. Turn them in at the end of class for credit. Names. 1. Find the (a) exact solution

More information

The Multi-Arm Bandit Framework

The Multi-Arm Bandit Framework The Multi-Arm Bandit Framework A. LAZARIC (SequeL Team @INRIA-Lille) ENS Cachan - Master 2 MVA SequeL INRIA Lille MVA-RL Course In This Lecture A. LAZARIC Reinforcement Learning Algorithms Oct 29th, 2013-2/94

More information

STA 256: Statistics and Probability I

STA 256: Statistics and Probability I Al Nosedal. University of Toronto. Fall 2017 My momma always said: Life was like a box of chocolates. You never know what you re gonna get. Forrest Gump. There are situations where one might be interested

More information

Online Aggregation of Unbounded Signed Losses Using Shifting Experts

Online Aggregation of Unbounded Signed Losses Using Shifting Experts Proceedings of Machine Learning Research 60: 5, 207 Conformal and Probabilistic Prediction and Applications Online Aggregation of Unbounded Signed Losses Using Shifting Experts Vladimir V. V yugin Institute

More information

Partial derivatives, linear approximation and optimization

Partial derivatives, linear approximation and optimization ams 11b Study Guide 4 econ 11b Partial derivatives, linear approximation and optimization 1. Find the indicated partial derivatives of the functions below. a. z = 3x 2 + 4xy 5y 2 4x + 7y 2, z x = 6x +

More information

Chapter 8. Exponential and Logarithmic Functions

Chapter 8. Exponential and Logarithmic Functions Chapter 8 Eponential and Logarithmic Functions Lesson 8-1 Eploring Eponential Models Eponential Function The general form of an eponential function is y = ab. Growth Factor When the value of b is greater

More information

Some Aspects of Universal Portfolio

Some Aspects of Universal Portfolio 1 Some Aspects of Universal Portfolio Tomoyuki Ichiba (UC Santa Barbara) joint work with Marcel Brod (ETH Zurich) Conference on Stochastic Asymptotics & Applications Sixth Western Conference on Mathematical

More information

1 Cost, Revenue and Profit

1 Cost, Revenue and Profit MATH 104 - SECTION 101 FIN AL REVIEW 1 Cost, Revenue and Profit C(x), R(x), and P(x); marginal cost MC(x), marginal revenue MR(x), and marginal profit M P(x). 1. Profit is the difference between cost and

More information

Adaptive Sampling Under Low Noise Conditions 1

Adaptive Sampling Under Low Noise Conditions 1 Manuscrit auteur, publié dans "41èmes Journées de Statistique, SFdS, Bordeaux (2009)" Adaptive Sampling Under Low Noise Conditions 1 Nicolò Cesa-Bianchi Dipartimento di Scienze dell Informazione Università

More information

= 2 = 1.5. Figure 4.1: WARP violated

= 2 = 1.5. Figure 4.1: WARP violated Chapter 4 The Consumer Exercise 4.1 You observe a consumer in two situations: with an income of $100 he buys 5 units of good 1 at a price of $10 per unit and 10 units of good 2 at a price of $5 per unit.

More information

Ambiguity in portfolio optimization

Ambiguity in portfolio optimization May/June 2006 Introduction: Risk and Ambiguity Frank Knight Risk, Uncertainty and Profit (1920) Risk: the decision-maker can assign mathematical probabilities to random phenomena Uncertainty: randomness

More information

The main purpose of this chapter is to prove the rst and second fundamental theorem of asset pricing in a so called nite market model.

The main purpose of this chapter is to prove the rst and second fundamental theorem of asset pricing in a so called nite market model. 1 2. Option pricing in a nite market model (February 14, 2012) 1 Introduction The main purpose of this chapter is to prove the rst and second fundamental theorem of asset pricing in a so called nite market

More information

Ordered Sample Generation

Ordered Sample Generation Ordered Sample Generation Xuebo Yu November 20, 2010 1 Introduction There are numerous distributional problems involving order statistics that can not be treated analytically and need to simulated through

More information

Coding on Countably Infinite Alphabets

Coding on Countably Infinite Alphabets Coding on Countably Infinite Alphabets Non-parametric Information Theory Licence de droits d usage Outline Lossless Coding on infinite alphabets Source Coding Universal Coding Infinite Alphabets Enveloppe

More information

MTAEA Implicit Functions

MTAEA Implicit Functions School of Economics, Australian National University February 12, 2010 Implicit Functions and Their Derivatives Up till now we have only worked with functions in which the endogenous variables are explicit

More information

Online Prediction Peter Bartlett

Online Prediction Peter Bartlett Online Prediction Peter Bartlett Statistics and EECS UC Berkeley and Mathematical Sciences Queensland University of Technology Online Prediction Repeated game: Cumulative loss: ˆL n = Decision method plays

More information

Geometry and Optimization of Relative Arbitrage

Geometry and Optimization of Relative Arbitrage Geometry and Optimization of Relative Arbitrage Ting-Kam Leonard Wong joint work with Soumik Pal Department of Mathematics, University of Washington Financial/Actuarial Mathematic Seminar, University of

More information

Chapter 4 Notes, Calculus I with Precalculus 3e Larson/Edwards

Chapter 4 Notes, Calculus I with Precalculus 3e Larson/Edwards 4.1 The Derivative Recall: For the slope of a line we need two points (x 1,y 1 ) and (x 2,y 2 ). Then the slope is given by the formula: m = y x = y 2 y 1 x 2 x 1 On a curve we can find the slope of a

More information

1 Functions and Graphs

1 Functions and Graphs 1 Functions and Graphs 1.1 Functions Cartesian Coordinate System A Cartesian or rectangular coordinate system is formed by the intersection of a horizontal real number line, usually called the x axis,

More information

Online Convex Optimization: From Gambling to Minimax Theorems by Playing Repeated Games

Online Convex Optimization: From Gambling to Minimax Theorems by Playing Repeated Games Online Convex Optimization: From Gambling to Minimax heorems by Playing Repeated Games im van Erven Nederlands Mathematisch Congres, April 4, 2018 Example: Betting on Football Games Before every match

More information

Simplifying Radical Expressions

Simplifying Radical Expressions Simplifying Radical Expressions Product Property of Radicals For any real numbers a and b, and any integer n, n>1, 1. If n is even, then When a and b are both nonnegative. n ab n a n b 2. If n is odd,

More information

LEGENDRE POLYNOMIALS AND APPLICATIONS. We construct Legendre polynomials and apply them to solve Dirichlet problems in spherical coordinates.

LEGENDRE POLYNOMIALS AND APPLICATIONS. We construct Legendre polynomials and apply them to solve Dirichlet problems in spherical coordinates. LEGENDRE POLYNOMIALS AND APPLICATIONS We construct Legendre polynomials and apply them to solve Dirichlet problems in spherical coordinates.. Legendre equation: series solutions The Legendre equation is

More information

1 Markov decision processes

1 Markov decision processes 2.997 Decision-Making in Large-Scale Systems February 4 MI, Spring 2004 Handout #1 Lecture Note 1 1 Markov decision processes In this class we will study discrete-time stochastic systems. We can describe

More information

converges as well if x < 1. 1 x n x n 1 1 = 2 a nx n

converges as well if x < 1. 1 x n x n 1 1 = 2 a nx n Solve the following 6 problems. 1. Prove that if series n=1 a nx n converges for all x such that x < 1, then the series n=1 a n xn 1 x converges as well if x < 1. n For x < 1, x n 0 as n, so there exists

More information

We now impose the boundary conditions in (6.11) and (6.12) and solve for a 1 and a 2 as follows: m 1 e 2m 1T m 2 e (m 1+m 2 )T. (6.

We now impose the boundary conditions in (6.11) and (6.12) and solve for a 1 and a 2 as follows: m 1 e 2m 1T m 2 e (m 1+m 2 )T. (6. 158 6. Applications to Production And Inventory For ease of expressing a 1 and a 2, let us define two constants b 1 = I 0 Q(0), (6.13) b 2 = ˆP Q(T) S(T). (6.14) We now impose the boundary conditions in

More information

11.6: Ratio and Root Tests Page 1. absolutely convergent, conditionally convergent, or divergent?

11.6: Ratio and Root Tests Page 1. absolutely convergent, conditionally convergent, or divergent? .6: Ratio and Root Tests Page Questions ( 3) n n 3 ( 3) n ( ) n 5 + n ( ) n e n ( ) n+ n2 2 n Example Show that ( ) n n ln n ( n 2 ) n + 2n 2 + converges for all x. Deduce that = 0 for all x. Solutions

More information

2.6 Logarithmic Functions. Inverse Functions. Question: What is the relationship between f(x) = x 2 and g(x) = x?

2.6 Logarithmic Functions. Inverse Functions. Question: What is the relationship between f(x) = x 2 and g(x) = x? Inverse Functions Question: What is the relationship between f(x) = x 3 and g(x) = 3 x? Question: What is the relationship between f(x) = x 2 and g(x) = x? Definition (One-to-One Function) A function f

More information

Advanced Topics in Machine Learning and Algorithmic Game Theory Fall semester, 2011/12

Advanced Topics in Machine Learning and Algorithmic Game Theory Fall semester, 2011/12 Advanced Topics in Machine Learning and Algorithmic Game Theory Fall semester, 2011/12 Lecture 4: Multiarmed Bandit in the Adversarial Model Lecturer: Yishay Mansour Scribe: Shai Vardi 4.1 Lecture Overview

More information

Probabilistic Graphical Models

Probabilistic Graphical Models Probabilistic Graphical Models Brown University CSCI 2950-P, Spring 2013 Prof. Erik Sudderth Lecture 12: Gaussian Belief Propagation, State Space Models and Kalman Filters Guest Kalman Filter Lecture by

More information

THE WEIGHTED MAJORITY ALGORITHM

THE WEIGHTED MAJORITY ALGORITHM THE WEIGHTED MAJORITY ALGORITHM Csaba Szepesvári University of Alberta CMPUT 654 E-mail: szepesva@ualberta.ca UofA, October 3, 2006 OUTLINE 1 PREDICTION WITH EXPERT ADVICE 2 HALVING: FIND THE PERFECT EXPERT!

More information

Hölder s and Minkowski s Inequality

Hölder s and Minkowski s Inequality Hölder s and Minkowski s Inequality James K. Peterson Department of Biological Sciences and Department of Mathematical Sciences Clemson University September 1, 218 Outline Conjugate Exponents Hölder s

More information

Math 116: Business Calculus Chapter 4 - Calculating Derivatives

Math 116: Business Calculus Chapter 4 - Calculating Derivatives Math 116: Business Calculus Chapter 4 - Calculating Derivatives Instructor: Colin Clark Spring 2017 Exam 2 - Thursday March 9. 4.1 Techniques for Finding Derivatives. 4.2 Derivatives of Products and Quotients.

More information

Optimal exit strategies for investment projects. 7th AMaMeF and Swissquote Conference

Optimal exit strategies for investment projects. 7th AMaMeF and Swissquote Conference Optimal exit strategies for investment projects Simone Scotti Université Paris Diderot Laboratoire de Probabilité et Modèles Aléatories Joint work with : Etienne Chevalier, Université d Evry Vathana Ly

More information

Online learning CMPUT 654. October 20, 2011

Online learning CMPUT 654. October 20, 2011 Online learning CMPUT 654 Gábor Bartók Dávid Pál Csaba Szepesvári István Szita October 20, 2011 Contents 1 Shooting Game 4 1.1 Exercises...................................... 6 2 Weighted Majority Algorithm

More information

A Model of Optimal Portfolio Selection under. Liquidity Risk and Price Impact

A Model of Optimal Portfolio Selection under. Liquidity Risk and Price Impact A Model of Optimal Portfolio Selection under Liquidity Risk and Price Impact Huyên PHAM Workshop on PDE and Mathematical Finance KTH, Stockholm, August 15, 2005 Laboratoire de Probabilités et Modèles Aléatoires

More information

Week 10: Theory of the Firm (Jehle and Reny, Chapter 3)

Week 10: Theory of the Firm (Jehle and Reny, Chapter 3) Week 10: Theory of the Firm (Jehle and Reny, Chapter 3) Tsun-Feng Chiang* *School of Economics, Henan University, Kaifeng, China November 22, 2015 First Last (shortinst) Short title November 22, 2015 1

More information

UNIVERSAL PORTFOLIO GENERATED BY IDEMPOTENT MATRIX AND SOME PROBABILITY DISTRIBUTION LIM KIAN HENG MASTER OF MATHEMATICAL SCIENCES

UNIVERSAL PORTFOLIO GENERATED BY IDEMPOTENT MATRIX AND SOME PROBABILITY DISTRIBUTION LIM KIAN HENG MASTER OF MATHEMATICAL SCIENCES UNIVERSAL PORTFOLIO GENERATED BY IDEMPOTENT MATRIX AND SOME PROBABILITY DISTRIBUTION LIM KIAN HENG MASTER OF MATHEMATICAL SCIENCES FACULTY OF ENGINEERING AND SCIENCE UNIVERSITI TUNKU ABDUL RAHMAN APRIL

More information

Dynamic Discrete Choice Structural Models in Empirical IO

Dynamic Discrete Choice Structural Models in Empirical IO Dynamic Discrete Choice Structural Models in Empirical IO Lecture 4: Euler Equations and Finite Dependence in Dynamic Discrete Choice Models Victor Aguirregabiria (University of Toronto) Carlos III, Madrid

More information

Expectation maximization

Expectation maximization Expectation maximization Subhransu Maji CMSCI 689: Machine Learning 14 April 2015 Motivation Suppose you are building a naive Bayes spam classifier. After your are done your boss tells you that there is

More information

Fundamentals in Optimal Investments. Lecture I

Fundamentals in Optimal Investments. Lecture I Fundamentals in Optimal Investments Lecture I + 1 Portfolio choice Portfolio allocations and their ordering Performance indices Fundamentals in optimal portfolio choice Expected utility theory and its

More information

SOME RESULTS AND PROBLEMS IN PROBABILISTIC NUMBER THEORY

SOME RESULTS AND PROBLEMS IN PROBABILISTIC NUMBER THEORY Annales Univ. Sci. Budapest., Sect. Comp. 43 204 253 265 SOME RESULTS AND PROBLEMS IN PROBABILISTIC NUMBER THEORY Imre Kátai and Bui Minh Phong Budapest, Hungary Le Manh Thanh Hue, Vietnam Communicated

More information

Sensitivity analysis of the expected utility maximization problem with respect to model perturbations

Sensitivity analysis of the expected utility maximization problem with respect to model perturbations Sensitivity analysis of the expected utility maximization problem with respect to model perturbations Mihai Sîrbu, The University of Texas at Austin based on joint work with Oleksii Mostovyi University

More information

Online Learning and Sequential Decision Making

Online Learning and Sequential Decision Making Online Learning and Sequential Decision Making Emilie Kaufmann CNRS & CRIStAL, Inria SequeL, emilie.kaufmann@univ-lille.fr Research School, ENS Lyon, Novembre 12-13th 2018 Emilie Kaufmann Online Learning

More information

Keywords: Acinonyx jubatus/cheetah/development/diet/hand raising/health/kitten/medication

Keywords: Acinonyx jubatus/cheetah/development/diet/hand raising/health/kitten/medication L V. A W P. Ky: Ayx j//m// ///m A: A y m "My" W P 1986. S y m y y. y mm m. A 6.5 m My.. A { A N D R A S D C T A A T ' } T A K P L A N T A T { - A C A S S T 0 R Y y m T ' 1986' W P - + ' m y, m T y. j-

More information

Prediction and Playing Games

Prediction and Playing Games Prediction and Playing Games Vineel Pratap vineel@eng.ucsd.edu February 20, 204 Chapter 7 : Prediction, Learning and Games - Cesa Binachi & Lugosi K-Person Normal Form Games Each player k (k =,..., K)

More information

MATH 412 Fourier Series and PDE- Spring 2010 SOLUTIONS to HOMEWORK 5

MATH 412 Fourier Series and PDE- Spring 2010 SOLUTIONS to HOMEWORK 5 MATH 4 Fourier Series PDE- Spring SOLUTIONS to HOMEWORK 5 Problem (a: Solve the following Sturm-Liouville problem { (xu + λ x u = < x < e u( = u (e = (b: Show directly that the eigenfunctions are orthogonal

More information

DUALITY AND INTEGER PROGRAMMING. Jean B. LASSERRE

DUALITY AND INTEGER PROGRAMMING. Jean B. LASSERRE LABORATOIRE d ANALYSE et d ARCHITECTURE des SYSTEMES DUALITY AND INTEGER PROGRAMMING Jean B. LASSERRE 1 Current solvers (CPLEX, XPRESS-MP) are rather efficient and can solve many large size problems with

More information

Thomas Knispel Leibniz Universität Hannover

Thomas Knispel Leibniz Universität Hannover Optimal long term investment under model ambiguity Optimal long term investment under model ambiguity homas Knispel Leibniz Universität Hannover knispel@stochastik.uni-hannover.de AnStAp0 Vienna, July

More information

Integration and Differentiation Limit Interchange Theorems

Integration and Differentiation Limit Interchange Theorems Integration and Differentiation Limit Interchange Theorems James K. Peterson Department of Biological Sciences and Department of Mathematical Sciences Clemson University March 11, 2018 Outline A More General

More information

Order book modeling and market making under uncertainty.

Order book modeling and market making under uncertainty. Order book modeling and market making under uncertainty. Sidi Mohamed ALY Lund University sidi@maths.lth.se (Joint work with K. Nyström and C. Zhang, Uppsala University) Le Mans, June 29, 2016 1 / 22 Outline

More information

Introduction to Bandit Algorithms. Introduction to Bandit Algorithms

Introduction to Bandit Algorithms. Introduction to Bandit Algorithms Stochastic K-Arm Bandit Problem Formulation Consider K arms (actions) each correspond to an unknown distribution {ν k } K k=1 with values bounded in [0, 1]. At each time t, the agent pulls an arm I t {1,...,

More information

Asset Pricing. Chapter IX. The Consumption Capital Asset Pricing Model. June 20, 2006

Asset Pricing. Chapter IX. The Consumption Capital Asset Pricing Model. June 20, 2006 Chapter IX. The Consumption Capital Model June 20, 2006 The Representative Agent Hypothesis and its Notion of Equilibrium 9.2.1 An infinitely lived Representative Agent Avoid terminal period problem Equivalence

More information

Portfolio Optimization in discrete time

Portfolio Optimization in discrete time Portfolio Optimization in discrete time Wolfgang J. Runggaldier Dipartimento di Matematica Pura ed Applicata Universitá di Padova, Padova http://www.math.unipd.it/runggaldier/index.html Abstract he paper

More information

Lecture 8: Information Theory and Statistics

Lecture 8: Information Theory and Statistics Lecture 8: Information Theory and Statistics Part II: Hypothesis Testing and I-Hsiang Wang Department of Electrical Engineering National Taiwan University ihwang@ntu.edu.tw December 23, 2015 1 / 50 I-Hsiang

More information

Handout 1: Introduction to Dynamic Programming. 1 Dynamic Programming: Introduction and Examples

Handout 1: Introduction to Dynamic Programming. 1 Dynamic Programming: Introduction and Examples SEEM 3470: Dynamic Optimization and Applications 2013 14 Second Term Handout 1: Introduction to Dynamic Programming Instructor: Shiqian Ma January 6, 2014 Suggested Reading: Sections 1.1 1.5 of Chapter

More information

QMI Lesson 19: Integration by Substitution, Definite Integral, and Area Under Curve

QMI Lesson 19: Integration by Substitution, Definite Integral, and Area Under Curve QMI Lesson 19: Integration by Substitution, Definite Integral, and Area Under Curve C C Moxley Samford University Brock School of Business Substitution Rule The following rules arise from the chain rule

More information

Real Analysis Math 131AH Rudin, Chapter #1. Dominique Abdi

Real Analysis Math 131AH Rudin, Chapter #1. Dominique Abdi Real Analysis Math 3AH Rudin, Chapter # Dominique Abdi.. If r is rational (r 0) and x is irrational, prove that r + x and rx are irrational. Solution. Assume the contrary, that r+x and rx are rational.

More information

Course Handouts: Pages 1-20 ASSET PRICE BUBBLES AND SPECULATION. Jan Werner

Course Handouts: Pages 1-20 ASSET PRICE BUBBLES AND SPECULATION. Jan Werner Course Handouts: Pages 1-20 ASSET PRICE BUBBLES AND SPECULATION Jan Werner European University Institute May 2010 1 I. Price Bubbles: An Example Example I.1 Time is infinite; so dates are t = 0,1,2,...,.

More information

Asymptotic Minimax Regret for Data Compression, Gambling, and Prediction

Asymptotic Minimax Regret for Data Compression, Gambling, and Prediction IEEE TRANSACTIONS ON INFORMATION THEORY, VOL. 46, NO. 2, MARCH 2000 431 Asymptotic Minimax Regret for Data Compression, Gambling, Prediction Qun Xie Andrew R. Barron, Member, IEEE Abstract For problems

More information

ECE 314 Signals and Systems Fall 2012

ECE 314 Signals and Systems Fall 2012 ECE 31 ignals and ystems Fall 01 olutions to Homework 5 Problem.51 Determine the impulse response of the system described by y(n) = x(n) + ax(n k). Replace x by δ to obtain the impulse response: h(n) =

More information

Meta Optimization and its Application to Portfolio Selection

Meta Optimization and its Application to Portfolio Selection Meta Optimization and its Application to Portfolio Selection Puja Das Dept of Computer Science & Engg Univ of Minnesota, win Cities pdas@cs.umn.edu Arindam Banerjee Dept of Computer Science & Engg Univ

More information

Functions. A function is a rule that gives exactly one output number to each input number.

Functions. A function is a rule that gives exactly one output number to each input number. Functions A function is a rule that gives exactly one output number to each input number. Why it is important to us? The set of all input numbers to which the rule applies is called the domain of the function.

More information

Decision Models Lecture 5 1. Lecture 5. Foreign-Currency Trading Integer Programming Plant-location example Summary and Preparation for next class

Decision Models Lecture 5 1. Lecture 5. Foreign-Currency Trading Integer Programming Plant-location example Summary and Preparation for next class Decision Models Lecture 5 1 Lecture 5 Foreign-Currency Trading Integer Programming Plant-location example Summary and Preparation for next class Foreign Exchange (FX) Markets Decision Models Lecture 5

More information

Pareto Efficiency in Robust Optimization

Pareto Efficiency in Robust Optimization Pareto Efficiency in Robust Optimization Dan Iancu Graduate School of Business Stanford University joint work with Nikolaos Trichakis (HBS) 1/26 Classical Robust Optimization Typical linear optimization

More information

CS281B/Stat241B. Statistical Learning Theory. Lecture 1.

CS281B/Stat241B. Statistical Learning Theory. Lecture 1. CS281B/Stat241B. Statistical Learning Theory. Lecture 1. Peter Bartlett 1. Organizational issues. 2. Overview. 3. Probabilistic formulation of prediction problems. 4. Game theoretic formulation of prediction

More information

Economics 205 Exercises

Economics 205 Exercises Economics 05 Eercises Prof. Watson, Fall 006 (Includes eaminations through Fall 003) Part 1: Basic Analysis 1. Using ε and δ, write in formal terms the meaning of lim a f() = c, where f : R R.. Write the

More information

EC5555 Economics Masters Refresher Course in Mathematics September 2013

EC5555 Economics Masters Refresher Course in Mathematics September 2013 EC5555 Economics Masters Refresher Course in Mathematics September 013 Lecture 3 Differentiation Francesco Feri Rationale for Differentiation Much of economics is concerned with optimisation (maximise

More information

Generalized Linear Models and Exponential Families

Generalized Linear Models and Exponential Families Generalized Linear Models and Exponential Families David M. Blei COS424 Princeton University April 12, 2012 Generalized Linear Models x n y n β Linear regression and logistic regression are both linear

More information

Financial Factors in Economic Fluctuations. Lawrence Christiano Roberto Motto Massimo Rostagno

Financial Factors in Economic Fluctuations. Lawrence Christiano Roberto Motto Massimo Rostagno Financial Factors in Economic Fluctuations Lawrence Christiano Roberto Motto Massimo Rostagno Background Much progress made on constructing and estimating models that fit quarterly data well (Smets-Wouters,

More information

Short correct answers are sufficient and get full credit. Including irrelevant (though correct) information in an answer will not increase the score.

Short correct answers are sufficient and get full credit. Including irrelevant (though correct) information in an answer will not increase the score. Economics 200A Part 2 UCSD Fall 2012 Prof. R. Starr, Mr. Troy Kravitz Final Exam 1 Your Name: Please answer all questions. Each of the six questions marked with a big number counts equally. Designate your

More information

Logarithms. Professor Richard Blecksmith Dept. of Mathematical Sciences Northern Illinois University

Logarithms. Professor Richard Blecksmith Dept. of Mathematical Sciences Northern Illinois University Logarithms Professor Richard Blecksmith richard@math.niu.edu Dept. of Mathematical Sciences Northern Illinois University http://math.niu.edu/ richard/math211 1. Definition of Logarithm For a > 0, a 1,

More information

Lecture : The Indefinite Integral MTH 124

Lecture : The Indefinite Integral MTH 124 Up to this point we have investigated the definite integral of a function over an interval. In particular we have done the following. Approximated integrals using left and right Riemann sums. Defined the

More information

0.1 Motivating example: weighted majority algorithm

0.1 Motivating example: weighted majority algorithm princeton univ. F 16 cos 521: Advanced Algorithm Design Lecture 8: Decision-making under total uncertainty: the multiplicative weight algorithm Lecturer: Sanjeev Arora Scribe: Sanjeev Arora (Today s notes

More information

40.530: Statistics. Professor Chen Zehua. Singapore University of Design and Technology

40.530: Statistics. Professor Chen Zehua. Singapore University of Design and Technology Singapore University of Design and Technology Lecture 9: Hypothesis testing, uniformly most powerful tests. The Neyman-Pearson framework Let P be the family of distributions of concern. The Neyman-Pearson

More information

Parametric Equations, Function Composition and the Chain Rule: A Worksheet

Parametric Equations, Function Composition and the Chain Rule: A Worksheet Parametric Equations, Function Composition and the Chain Rule: A Worksheet Prof.Rebecca Goldin Oct. 8, 003 1 Parametric Equations We have seen that the graph of a function f(x) of one variable consists

More information

3 a = 3 b c 2 = a 2 + b 2 = 2 2 = 4 c 2 = 3b 2 + b 2 = 4b 2 = 4 b 2 = 1 b = 1 a = 3b = 3. x 2 3 y2 1 = 1.

3 a = 3 b c 2 = a 2 + b 2 = 2 2 = 4 c 2 = 3b 2 + b 2 = 4b 2 = 4 b 2 = 1 b = 1 a = 3b = 3. x 2 3 y2 1 = 1. MATH 222 LEC SECOND MIDTERM EXAM THU NOV 8 PROBLEM ( 5 points ) Find the standard-form equation for the hyperbola which has its foci at F ± (±2, ) and whose asymptotes are y ± 3 x The calculations b a

More information

Comparative Statics. Autumn 2018

Comparative Statics. Autumn 2018 Comparative Statics Autumn 2018 What is comparative statics? Contents 1 What is comparative statics? 2 One variable functions Multiple variable functions Vector valued functions Differential and total

More information

1 Definition of the Riemann integral

1 Definition of the Riemann integral MAT337H1, Introduction to Real Analysis: notes on Riemann integration 1 Definition of the Riemann integral Definition 1.1. Let [a, b] R be a closed interval. A partition P of [a, b] is a finite set of

More information

Entropy, Inference, and Channel Coding

Entropy, Inference, and Channel Coding Entropy, Inference, and Channel Coding Sean Meyn Department of Electrical and Computer Engineering University of Illinois and the Coordinated Science Laboratory NSF support: ECS 02-17836, ITR 00-85929

More information

MAT 570 REAL ANALYSIS LECTURE NOTES. Contents. 1. Sets Functions Countability Axiom of choice Equivalence relations 9

MAT 570 REAL ANALYSIS LECTURE NOTES. Contents. 1. Sets Functions Countability Axiom of choice Equivalence relations 9 MAT 570 REAL ANALYSIS LECTURE NOTES PROFESSOR: JOHN QUIGG SEMESTER: FALL 204 Contents. Sets 2 2. Functions 5 3. Countability 7 4. Axiom of choice 8 5. Equivalence relations 9 6. Real numbers 9 7. Extended

More information

FACULTY OF ARTS AND SCIENCE University of Toronto FINAL EXAMINATIONS, APRIL 2012 MAT 133Y1Y Calculus and Linear Algebra for Commerce

FACULTY OF ARTS AND SCIENCE University of Toronto FINAL EXAMINATIONS, APRIL 2012 MAT 133Y1Y Calculus and Linear Algebra for Commerce FACULTY OF ARTS AND SCIENCE University of Toronto FINAL EXAMINATIONS, APRIL 0 MAT 33YY Calculus and Linear Algebra for Commerce Duration: Examiners: 3 hours N. Francetic A. Igelfeld P. Kergin J. Tate LEAVE

More information

Math Refresher Course

Math Refresher Course Math Refresher Course Columbia University Department of Political Science Fall 2007 Day 2 Prepared by Jessamyn Blau 6 Calculus CONT D 6.9 Antiderivatives and Integration Integration is the reverse of differentiation.

More information

7.1. Calculus of inverse functions. Text Section 7.1 Exercise:

7.1. Calculus of inverse functions. Text Section 7.1 Exercise: Contents 7. Inverse functions 1 7.1. Calculus of inverse functions 2 7.2. Derivatives of exponential function 4 7.3. Logarithmic function 6 7.4. Derivatives of logarithmic functions 7 7.5. Exponential

More information

CS261: A Second Course in Algorithms Lecture #12: Applications of Multiplicative Weights to Games and Linear Programs

CS261: A Second Course in Algorithms Lecture #12: Applications of Multiplicative Weights to Games and Linear Programs CS26: A Second Course in Algorithms Lecture #2: Applications of Multiplicative Weights to Games and Linear Programs Tim Roughgarden February, 206 Extensions of the Multiplicative Weights Guarantee Last

More information

Minimax Redundancy for Large Alphabets by Analytic Methods

Minimax Redundancy for Large Alphabets by Analytic Methods Minimax Redundancy for Large Alphabets by Analytic Methods Wojciech Szpankowski Purdue University W. Lafayette, IN 47907 March 15, 2012 NSF STC Center for Science of Information CISS, Princeton 2012 Joint

More information

BMIR Lecture Series on Probability and Statistics Fall 2015 Discrete RVs

BMIR Lecture Series on Probability and Statistics Fall 2015 Discrete RVs Lecture #7 BMIR Lecture Series on Probability and Statistics Fall 2015 Department of Biomedical Engineering and Environmental Sciences National Tsing Hua University 7.1 Function of Single Variable Theorem

More information

Order book resilience, price manipulation, and the positive portfolio problem

Order book resilience, price manipulation, and the positive portfolio problem Order book resilience, price manipulation, and the positive portfolio problem Alexander Schied Mannheim University Workshop on New Directions in Financial Mathematics Institute for Pure and Applied Mathematics,

More information

Reformulation of chance constrained problems using penalty functions

Reformulation of chance constrained problems using penalty functions Reformulation of chance constrained problems using penalty functions Martin Branda Charles University in Prague Faculty of Mathematics and Physics EURO XXIV July 11-14, 2010, Lisbon Martin Branda (MFF

More information