Tail Factor Convergence in Sherman s Inverse Power Curve Loss Development Factor Model

Similar documents
Tail Factor Convergence in Sherman s Inverse Power Curve Loss Development Factor Model

Chapter Gauss-Seidel Method

Numerical Differentiation and Integration

Mathematically, integration is just finding the area under a curve from one point to another. It is b

this is the indefinite integral Since integration is the reverse of differentiation we can check the previous by [ ]

Problem Set 4 Solutions

CS 4758 Robot Kinematics. Ashutosh Saxena

In Calculus I you learned an approximation method using a Riemann sum. Recall that the Riemann sum is

under the curve in the first quadrant.

Chapter 1 Vector Spaces

Chapter Simpson s 1/3 Rule of Integration. ( x)

Strategies for the AP Calculus Exam

The Z-Transform in DSP Lecture Andreas Spanias

An Introduction to Robot Kinematics. Renata Melamud

10.2 Series. , we get. which is called an infinite series ( or just a series) and is denoted, for short, by the symbol. i i n

Matrix. Definition 1... a1 ... (i) where a. are real numbers. for i 1, 2,, m and j = 1, 2,, n (iii) A is called a square matrix if m n.

Exercise # 2.1 3, 7, , 3, , -9, 1, Solution: natural numbers are 3, , -9, 1, 2.5, 3, , , -9, 1, 2 2.5, 3, , -9, 1, , -9, 1, 2.

Numerical Analysis Topic 4: Least Squares Curve Fitting

Lecture 3: Review of Linear Algebra and MATLAB

ON JENSEN S AND HERMITE-HADAMARD S INEQUALITY

3/20/2013. Splines There are cases where polynomial interpolation is bad overshoot oscillations. Examplef x. Interpolation at -4,-3,-2,-1,0,1,2,3,4

Probabilistic Graphical Models

Optimality of Strategies for Collapsing Expanded Random Variables In a Simple Random Sample Ed Stanek

The linear system. The problem: solve

Definition Integral. over[ ab, ] the sum of the form. 2. Definite Integral

ON THE LAWS OF LARGE NUMBERS FOR DEPENDENT RANDOM VARIABLES

Preliminary Examinations: Upper V Mathematics Paper 1

COMPLEX NUMBERS AND DE MOIVRE S THEOREM

European Journal of Mathematics and Computer Science Vol. 3 No. 1, 2016 ISSN ISSN

Data Compression Techniques (Spring 2012) Model Solutions for Exercise 4

St John s College. UPPER V Mathematics: Paper 1 Learning Outcome 1 and 2. Examiner: GE Marks: 150 Moderator: BT / SLS INSTRUCTIONS AND INFORMATION

Analysis of error propagation in profile measurement by using stitching

Math 1313 Final Exam Review

Support vector machines for regression

The formulae in this booklet have been arranged according to the unit in which they are first

Area and the Definite Integral. Area under Curve. The Partition. y f (x) We want to find the area under f (x) on [ a, b ]

Asymptotic Dominance Problems. is not constant but for n 0, f ( n) 11. 0, so that for n N f

Advanced Algorithmic Problem Solving Le 3 Arithmetic. Fredrik Heintz Dept of Computer and Information Science Linköping University

Dual-Matrix Approach for Solving the Transportation Problem

GENERALIZED OPERATIONAL RELATIONS AND PROPERTIES OF FRACTIONAL HANKEL TRANSFORM

1 4 6 is symmetric 3 SPECIAL MATRICES 3.1 SYMMETRIC MATRICES. Defn: A matrix A is symmetric if and only if A = A, i.e., a ij =a ji i, j. Example 3.1.

Section 2.2. Matrix Multiplication

The z-transform. LTI System description. Prof. Siripong Potisuk

PART ONE. Solutions to Exercises

Ruin Probability-Based Initial Capital of the Discrete-Time Surplus Process

Module 3: Element Properties Lecture 5: Solid Elements

An Improvement on Disc Separation of the Schur Complement and Bounds for Determinants of Diagonally Dominant Matrices

MTH 146 Class 7 Notes

Modeling uncertainty using probabilities

Analyzing Control Structures

Johns Hopkins University Department of Biostatistics Math Review for Introductory Courses

Study on the Normal and Skewed Distribution of Isometric Grouping

The Computation of Common Infinity-norm Lyapunov Functions for Linear Switched Systems

Analele Universităţii din Oradea, Fascicula: Protecţia Mediului, Vol. XIII, 2008

Johns Hopkins University Department of Biostatistics Math Review for Introductory Courses

ICS141: Discrete Mathematics for Computer Science I

19 22 Evaluate the integral by interpreting it in terms of areas. (1 x ) dx. y Write the given sum or difference as a single integral in

20 23 Evaluate the integral by interpreting it in terms of areas. (1 x ) dx. y 2 2

Diagonally Implicit Runge-Kutta Nystrom General Method Order Five for Solving Second Order IVPs

Section 2:00 ~ 2:50 pm Thursday in Maryland 202 Sep. 29, 2005

Coding Theorems on New Fuzzy Information Theory of Order α and Type β

Lecture 6: Coding theory

On Several Inequalities Deduced Using a Power Series Approach

n 1 n 2 n 2 n n 1 ln n 2 1 n Express the number as a ratio of integers

means the first term, a2 means the term, etc. Infinite Sequences: follow the same pattern forever.

Density estimation II

CHAPTER 7 SOLVING FUZZY LINEAR FRACTIONAL PROGRAMMING PROBLEM BY USING TAYLOR'S SERIES METHOD

4.1 Schrödinger Equation in Spherical Coordinates

Extension Formulas of Lauricella s Functions by Applications of Dixon s Summation Theorem

Graphical rules for SU(N)

Gauss Quadrature Rule of Integration

Chapter 7. Bounds for weighted sums of Random Variables

M2S1 - EXERCISES 8: SOLUTIONS

Chapter Trapezoidal Rule of Integration

STATISTICS 13. Lecture 5 Apr 7, 2010

4. When is the particle speeding up? Why? 5. When is the particle slowing down? Why?

5 - Determinants. r r. r r. r r. r s r = + det det det

He e e some s nd t icks o emoving g om v iety o su ces

a f(x)dx is divergent.

Calculating the Values of Multiple Integrals by Replacing the Integrands Interpolation by Interpolation Polynomial

Cooper and McGillem Chapter 4: Moments Linear Regression

Sequences and summations

Chapter Gauss Quadrature Rule of Integration

Design maintenanceand reliability of engineering systems: a probability based approach

Technical Appendix for Inventory Management for an Assembly System with Product or Component Returns, DeCroix and Zipkin, Management Science 2005.

14.2 Line Integrals. determines a partition P of the curve by points Pi ( xi, y

Maximum Walk Entropy Implies Walk Regularity

Introduction of Fourier Series to First Year Undergraduate Engineering Students

Stats & Summary

( x y ) x y. a b. a b. Chapter 2Properties of Exponents and Scientific Notation. x x. x y, Example: (x 2 )(x 4 ) = x 6.

ME 501A Seminar in Engineering Analysis Page 1

Mu Sequences/Series Solutions National Convention 2014

Classification of Rational Homotopy Type for 8-Cohomological Dimension Elliptic Spaces

On Testing Simple and Composite Goodness-of-Fit Hypotheses When Data are Censored

Section IV.6: The Master Method and Applications

SUM PROPERTIES FOR THE K-LUCAS NUMBERS WITH ARITHMETIC INDEXES

Answer: First, I ll show how to find the terms analytically then I ll show how to use the TI to find them.

Gauss Quadrature Rule of Integration

1. (25 points) Use the limit definition of the definite integral and the sum formulas to compute. [1 x + x2

Data Structures LECTURE 10. Huffman coding. Example. Coding: problem definition

Transcription:

T tor Covergee Sherm s Iverse Power Curve Loss Deveopmet tor Moe Jo Evs Astrt The fte prout of the ge-to-ge eveopmet ftors Sherm s verse power urve moe s prove to overge to fte umer whe the power prmeter s ess th - tertvey to verge to fty whe the power prmeter s - or greter or the overget prmeter vues smpe formu s erve terms of y fte prout of ge-to-ge ftors for the epots of terv otg the mt of the fte prout These epots overge to the mt s the fte tme utoff pot reses or y fte tme utoff the prout of ge-to-ge ftors es eow the terv thus the ower epot of the terv s wys etter estmte of the mt th the fte prout tsef Sever umer empes re ue for ustrto Keywors T tor Iverse Power Curve BACKGROUND AND INTRODUCTION t ft I 984 Sherm [5] fou tht verse power urve of the form empr ge-to-ge oss eveopmet ftors etter th sever other s futo forms he teste Lowe Mohrm 985 [3] epresse oer out the overgee of the prout of the ge-to-ge ftors Boor 2006 [ p 373] the CAS T tor Workg Prty 203 [2 p 52] ote tht there hs ee o kow ose form epresso tht ppromtes the t geerte y the verse power urve I prte the ge-to-ge eveopmet ftors proue y the urve re mutpe together out to some fte ge utoff suh s t = 80 to proue umutve eveopmet ftor The mpt of ftors eyo tht ge to utmte or the t ftor eyo the utoff ths se t = 8 s ssume to e egge Atertvey f the mpt of the t ftor s ot egge the some other moeg oserto must form the seeto of the utoff tme The potet ger the ssumpto of egge t ftor mpt s ustrte Te Two fferet sets of prmeters shre the sme t ge-to-ge ftor of 0 t t = the sme umutve ftor of 30 from t = to 00 However whe the umutve ftor for Empe usg power prmeter = 40 grows oy tte pst t = 00 Empe 2 usg = 05 ppers to zoom towr fty the t Csuty Atur Soety E-orum 204-Voume Copyrght 204 Nto Cou o Compesto Isure I A Rghts Reserve

Covergee Of Sherm s Iverse Power Curve T tor Te : Empes of Apprety Coverget Dverget T tors for the Iverse Power Curve Moe Prmeter Vues Prmeters Empe Empe 2 545540243359093 0050047502457 40 05 849422458022239 2504425242429 Cumutve Deveopmet tors rom to Empe Empe 2 00 00 0 085 065 00 300 300 000 337 2482 0000 338 9293 00000 338 27E+04 000000 338 03E+3 0000000 338 54E+4 Ths pper uses s re yss ([4] eg str tetook referee to prove tht the fte prout of the ge-to-ge ftors overges to fte umer whe the power prmeter s ess th - verges to + whe - Note ths pper we refer to sequee tht reses wthout y upper ou s vergg to + or hvg mt of + urthermore whe < - for y fte prout of the ge-to-ge ftors up to spef ge there s smpe formu for terv otg the mt of the fte prout As reses the terv eomes tghter the epots eh overge to the mt of the fte prout The ower epot of ths terv s wys etter estmte of the fte prout th the fte prout of the ge-to-ge ftors whh s wys ess th the ower epot It s worth otg g tht t vergee oes ot eessry me the moe s v ut smpy tht y spef fte utoff pot shou e otherwse justfe or overget t ether utoff pot must st e justfe y some other oserto or re must e tke tht the t ftor pst the utoff s resoy ose to The terv 2 Csuty Atur Soety E-orum 204-Voume Copyrght 204 Nto Cou o Compesto Isure I A Rghts Reserve

Covergee Of Sherm s Iverse Power Curve T tor estmte erve ths pper hep swer the tter questo The proof of overgee/vergee s out Seto 2 wth the proof of sever usefu emms Appe A The terv estmte s erve Seto 22 Numer empes of the progressve overgee/vergee of the fte prout the terv estmte of the fte prout for sever sets of prmeters re show Seto 23 2 CONVERGENCE THEOREM AND LIMIT ESTIMATION oowg the otto ovetos of the reet CAS workg prty [2] the remer of ths pper ste of t s use for ge or tme 2 Sttemet Proof of Prmry Theorem rst we w set up efto for the fte prout of the ge-to-ge ftors the verse power urve moe Defto: s postve teger where > 0 0 re re umers Note ths efto ues ses where egs t hgher vue th s the prmeter e rese to he suh ses It s so worth otg tht 0 tht w e use susequet ervtos Theorem key ft ( If - the m Csuty Atur Soety E-orum 204-Voume 3 Copyrght 204 Nto Cou o Compesto Isure I A Rghts Reserve

Covergee Of Sherm s Iverse Power Curve T tor ( If < - the m ests Proof: ( or y sequee of umers 0 where 2 the equty hos org to Lemm A3 Appyg ths we hve If - the m org to Lemm A osequety m ( By Lemm A2 Summg over gves og for y 0 so og og og If < - the L m ests s ess th + org to Lemm A Now ote tht og s resg sequee euse og mpes tht 0 s oue y L Cosequety m og ess th + So m ests s ess th + ests s 4 Csuty Atur Soety E-orum 204-Voume Copyrght 204 Nto Cou o Compesto Isure I A Rghts Reserve

Covergee Of Sherm s Iverse Power Curve T tor Csuty Atur Soety E-orum 204-Voume 5 Copyrght 204 Nto Cou o Compesto Isure I A Rghts Reserve 22 A Iterv Estmte for the Ifte Prout Lmt or the overget se of < - t s posse to ostrut usefu terv estmte for the fte prout Defto: The t upper ou ftor s ep U Defto: The t ower ou ftor s L Theorem 2 Let m If < - the: ( ( m U ( ( m L ( U L Proof: ( + < 0 mpes tht 0 m osequety m ep

Covergee Of Sherm s Iverse Power Curve T tor 6 Csuty Atur Soety E-orum 204-Voume Copyrght 204 Nto Cou o Compesto Isure I A Rghts Reserve ( + < 0 mpes m m ( Tkg the ogrthm of the t ftor ppyg oug tehques esre Lemms A A2 og Epoettg proues ep Cosequety ( ( ( U Smry usg tehques from Lemms A A3 proues Cosequety ( ( ( L Ths ompetes the proof of Theorem 2 The ower epot of the estmto terv s wys etter estmte of the fte prout se L osequety L The t ou ftors re omputtoy smpe eve for rge vues of gve mesure of the retve wth of the estmto terv pror to og the omputtoy tese uto of the fte prout or empe to heve ert trget U for the

Covergee Of Sherm s Iverse Power Curve T tor upper ou requres ogu A more reevt mesure of retve error ut wthout y smpe formu for tht the uthor s wre of s the rto of the t upper ou ftor to the t ower ou ftor U / L ep Empe : A upper ou ftor trget set t U = 0 for the prmeter vues = 545540 = -40 = 849422 requres 78 However y 29 the rto of the t upper ou ftor to the t ower ou ftor s out 0 23 More Numer Empes Te 2 shows s fferet sets of prmeters eh of whh proues ge-to-ge ftor t = of 0 umutve ftor from = to 00 of 30 The prmeter sets re ee y set of vues {-20-5 - -0-09 -06} for the power prmeter or = - the vergee hppes very sowy ut for = - the overgee hppes remrky sowy To heve 0 U for the = - prmeter set wou requre 22 27 0 though y 0 530 U / L 0 st stroomy sow rte of overgee Csuty Atur Soety E-orum 204-Voume 7 Copyrght 204 Nto Cou o Compesto Isure I A Rghts Reserve

Covergee Of Sherm s Iverse Power Curve T tor Te 2: Empes of te Deveopmet tor Prouts Iterv Estmtes or Ifte Deveopmet tor Prouts Prmeter Vues Prmeters Empe 3 Empe 4 Empe 5 22097485352 0774730055099 0744567689596 20 5 338590439679 264389382624 24522704340826 Cumutve Deveopmet tor Prout (Ifte Prout Lower Bou Ifte Prout Upper Bou Empe 3 Empe 4 Empe 5 00 (352 43 00 (458 589 00 (2359 3877 0 083 (375 428 08 (488 585 078 (2449 3868 00 300 (47 423 300 (553 58 300 (273 3858 000 406 (423 423 477 (576 580 60 (307 3856 0000 42 (423 423 546 (579 580 926 (3263 3856 00000 423 (423 423 569 (580 580 222 (3447 3856 000000 423 (423 423 576 (580 580 2488 (3578 3856 0000000 423 (423 423 579 (580 580 2723 (3670 3856 00000000 423 (423 423 580 (580 580 2925 (3733 3856 000000000 423 (423 423 580 (580 580 3096 (3776 3856 Prmeter Vues Prmeters Empe 6 Empe 7 Empe 8 02899790370 0073738447594275 0029230646958 0 09 06 0289979090266 8206704807852 269970572509898 Cumutve Deveopmet tor Prout Empe 6 Empe 7 Empe 8 00 00 00 0 077 075 069 00 300 300 300 000 668 744 285 0000 26 2550 88 00000 2803 49 29782 000000 3635 7534 872E+05 0000000 474 6 955E+4 00000000 63 4946 486E+37 000000000 7928 3999 527E+94 8 Csuty Atur Soety E-orum 204-Voume Copyrght 204 Nto Cou o Compesto Isure I A Rghts Reserve

Covergee Of Sherm s Iverse Power Curve T tor Akowegmet The uthor s grety thkfu to Joh Roertso D Corro Le Herk for revew ommets o ths pper Appe A - Lemms Lemm A Let e postve teger 0 k ( If the m k k ( If the m k ests Proof: It suffes to show overgee or vergee for m k se k s fte umer k or k 0 k therefore m k or k 0 k s strty eresg futo of k therefore there s swh equty k k t t k k t k t osequety t t k k t t Sovg the tegrs whe Csuty Atur Soety E-orum 204-Voume 9 Copyrght 204 Nto Cou o Compesto Isure I A Rghts Reserve

Covergee Of Sherm s Iverse Power Curve T tor k k or tkg mts proues m k k I ths se the upper ou of the equty s fte umer mpes overgee to fte umer se the sequee of prt sums the me s o-eresg k or 0 tkg mts resuts m k se ths se the ower ou of the erer equty verges m or the se tegrto of the erer equty es to og k k og Oe g tkg mts es to m k k from the ower ou of the equty vergg m og Lemm A2 If > 0 the og 0 Csuty Atur Soety E-orum 204-Voume Copyrght 204 Nto Cou o Compesto Isure I A Rghts Reserve

Covergee Of Sherm s Iverse Power Curve T tor Csuty Atur Soety E-orum 204-Voume Copyrght 204 Nto Cou o Compesto Isure I A Rghts Reserve Proof: If t the 0 / t osequety 0 / t t So 0 / t t t sovg the tegrs proues 0 og Lemm A3 for y sequee of umers 0 teger 2 Proof: We proee y uto or 2 se 0 2 t foows tht 2 2 2 Assume the ouso of the emm s true for where 2 We w show tht the emm s the true for + I geer But so

whh estshes the emm Covergee Of Sherm s Iverse Power Curve T tor REERENCES [] Boor J Estmtg T Deveopmet tors: Wht to o Whe the Trge Rus Out Csuty Atur Soety orum Wter 2006 [2] CAS T tor Workg Prty The Estmto of Loss Deveopmet T tors: A Summry Report Csuty Atur Soety orum 203 [3] Lowe S P Mohrm D Dsusso of Etrpotg Smoothg Iterpotg Deveopmet tors PCAS LXXII 985 [4] Ru W Prpes of Mthemt Ayss 3r e MGrw-H I 976 [5] Sherm R E Etrpotg Smoothg Iterpotg Deveopmet tors PCAS LXXI 984 Bogrphy of the Author Joth Evs CAS SA CA CERA MAAA WCP s tury t the Nto Cou o Compesto Isure Bo Rto L Hs work prmry voves reserh eveopmet of NCCI s rtemkg reservg tstrophe moeg proeures 2 Csuty Atur Soety E-orum 204-Voume Copyrght 204 Nto Cou o Compesto Isure I A Rghts Reserve