A proof of the dimension conjecture and of Jacobi's bound. François Ollivier (CNRS) LIX UMR CNRS École polytechnique

Similar documents
Differential Polynomials

n α j x j = 0 j=1 has a nontrivial solution. Here A is the n k matrix whose jth column is the vector for all t j=0

Difference Equations

APPENDIX A Some Linear Algebra

8.4 COMPLEX VECTOR SPACES AND INNER PRODUCTS

Lectures - Week 4 Matrix norms, Conditioning, Vector Spaces, Linear Independence, Spanning sets and Basis, Null space and Range of a Matrix

Multivariate Subresultants

Chapter 5. Solution of System of Linear Equations. Module No. 6. Solution of Inconsistent and Ill Conditioned Systems

THE WEIGHTED WEAK TYPE INEQUALITY FOR THE STRONG MAXIMAL FUNCTION

Lecture 5 Decoding Binary BCH Codes

Errors for Linear Systems

DIFFERENTIAL FORMS BRIAN OSSERMAN

9 Characteristic classes

Inner Product. Euclidean Space. Orthonormal Basis. Orthogonal

Introduction to Vapor/Liquid Equilibrium, part 2. Raoult s Law:

THE CHINESE REMAINDER THEOREM. We should thank the Chinese for their wonderful remainder theorem. Glenn Stevens

Group Analysis of Ordinary Differential Equations of the Order n>2

1 Generating functions, continued

Salmon: Lectures on partial differential equations. Consider the general linear, second-order PDE in the form. ,x 2

Linear, affine, and convex sets and hulls In the sequel, unless otherwise specified, X will denote a real vector space.

where a is any ideal of R. Lemma Let R be a ring. Then X = Spec R is a topological space. Moreover the open sets

Econ107 Applied Econometrics Topic 3: Classical Model (Studenmund, Chapter 4)

Random Walks on Digraphs

Polynomial Regression Models

arxiv: v1 [math.ho] 18 May 2008

Determinants Containing Powers of Generalized Fibonacci Numbers

Problem Set 9 Solutions

Case A. P k = Ni ( 2L i k 1 ) + (# big cells) 10d 2 P k.

Exercise Solutions to Real Analysis

Asymptotics of the Solution of a Boundary Value. Problem for One-Characteristic Differential. Equation Degenerating into a Parabolic Equation

NUMERICAL DIFFERENTIATION

College of Computer & Information Science Fall 2009 Northeastern University 20 October 2009

Causal Diamonds. M. Aghili, L. Bombelli, B. Pilgrim

MEM 255 Introduction to Control Systems Review: Basics of Linear Algebra

Affine transformations and convexity

Example: (13320, 22140) =? Solution #1: The divisors of are 1, 2, 3, 4, 5, 6, 9, 10, 12, 15, 18, 20, 27, 30, 36, 41,

Implicit Integration Henyey Method

PHYS 705: Classical Mechanics. Canonical Transformation II

SL n (F ) Equals its Own Derived Group

AERODYNAMICS I LECTURE 6 AERODYNAMICS OF A WING FUNDAMENTALS OF THE LIFTING-LINE THEORY

Report on Image warping

arxiv: v1 [math.co] 1 Mar 2014

CSci 6974 and ECSE 6966 Math. Tech. for Vision, Graphics and Robotics Lecture 21, April 17, 2006 Estimating A Plane Homography

Advanced Quantum Mechanics

U.C. Berkeley CS294: Beyond Worst-Case Analysis Luca Trevisan September 5, 2017

Math 217 Fall 2013 Homework 2 Solutions

Transfer Functions. Convenient representation of a linear, dynamic model. A transfer function (TF) relates one input and one output: ( ) system

Week 2. This week, we covered operations on sets and cardinality.

LINEAR REGRESSION ANALYSIS. MODULE IX Lecture Multicollinearity

where a is any ideal of R. Lemma 5.4. Let R be a ring. Then X = Spec R is a topological space Moreover the open sets

Differentiating Gaussian Processes

FACTORIZATION IN KRULL MONOIDS WITH INFINITE CLASS GROUP

Canonical transformations

ON MECHANICS WITH VARIABLE NONCOMMUTATIVITY

Chapter Newton s Method

A CLASS OF RECURSIVE SETS. Florentin Smarandache University of New Mexico 200 College Road Gallup, NM 87301, USA

ON A DETERMINATION OF THE INITIAL FUNCTIONS FROM THE OBSERVED VALUES OF THE BOUNDARY FUNCTIONS FOR THE SECOND-ORDER HYPERBOLIC EQUATION

Ballot Paths Avoiding Depth Zero Patterns

The equation of motion of a dynamical system is given by a set of differential equations. That is (1)

The Geometry of Logit and Probit

8.6 The Complex Number System

p 1 c 2 + p 2 c 2 + p 3 c p m c 2

NP-Completeness : Proofs

HOPF ALGEBRAS WITH TRACE AND CLEBSCH-GORDAN COEFFICIENTS. 1. Recollections and the problem

Approximate Smallest Enclosing Balls

Subset Topological Spaces and Kakutani s Theorem

Lecture 12: Discrete Laplacian

DISCRIMINANTS AND RAMIFIED PRIMES. 1. Introduction A prime number p is said to be ramified in a number field K if the prime ideal factorization

However, since P is a symmetric idempotent matrix, of P are either 0 or 1 [Eigen-values

DOUBLE POINTS AND THE PROPER TRANSFORM IN SYMPLECTIC GEOMETRY

Structure from Motion. Forsyth&Ponce: Chap. 12 and 13 Szeliski: Chap. 7

Supplement: Proofs and Technical Details for The Solution Path of the Generalized Lasso

Buckingham s pi-theorem

In this section is given an overview of the common elasticity models.

More metrics on cartesian products

Hyper-Sums of Powers of Integers and the Akiyama-Tanigawa Matrix

Maximizing the number of nonnegative subsets

Bernoulli Numbers and Polynomials

Topic 5: Non-Linear Regression

Lecture 10 Support Vector Machines II

THERE ARE INFINITELY MANY FIBONACCI COMPOSITES WITH PRIME SUBSCRIPTS

Feature Selection: Part 1

3.1 Expectation of Functions of Several Random Variables. )' be a k-dimensional discrete or continuous random vector, with joint PMF p (, E X E X1 E X

Perron Vectors of an Irreducible Nonnegative Interval Matrix

= z 20 z n. (k 20) + 4 z k = 4

Smarandache-Zero Divisors in Group Rings

Module 2. Random Processes. Version 2 ECE IIT, Kharagpur

12. The Hamilton-Jacobi Equation Michael Fowler

1 Matrix representations of canonical matrices

20. Mon, Oct. 13 What we have done so far corresponds roughly to Chapters 2 & 3 of Lee. Now we turn to Chapter 4. The first idea is connectedness.

For now, let us focus on a specific model of neurons. These are simplified from reality but can achieve remarkable results.

Caps and Colouring Steiner Triple Systems

CALCULUS CLASSROOM CAPSULES

5 The Rational Canonical Form

Games of Threats. Elon Kohlberg Abraham Neyman. Working Paper

Mechanics Physics 151

Section 8.3 Polar Form of Complex Numbers

Additional Codes using Finite Difference Method. 1 HJB Equation for Consumption-Saving Problem Without Uncertainty

DECOUPLING OF LINEAR TIME-VARYING SYSTEMS WITH A BOND GRAPH APPROACH

PRIMES 2015 reading project: Problem set #3

Transcription:

A proof of the dmenson conjecture and of Jacob's bound Franços Ollver (CNRS) LIX UMR CNRS École polytechnque

Jacob s bound Proposton. Let P 1,..., P n <x 1,..., x n >, a,j = ord xj P and = max σ S n Σ =1 n a,σ() r, then the order of P s not greater than.

Orgn of Jacob's bound Frst appeared n two posthumous papers: De nvestgando ordne systemats aequatonum dfferentalum vulgarum cujuscunque, Crelle s Journal 64, 1865, 297-320, De aequatonum dfferentalum systemate non normal ad formam normalem revocando, Vorlesungen über Dynamk 550-578, 1866. Edted by Cohn, Borchardt and Clebsch from a collecton of manuscrpts possbly wrten around 1840 (kept n the Archv der Berln-Brandenburgsche Akademe der Wssenschaft). Generc case

Rtt s contrbutons Jacob s bound n the lnear case, and for two varables. The dmenson conjecture for a sngle equaton. Paradoxcal examples: the ntersecton of two components of dfferental codmenson r 1 and r 2 may be of codmenson greater than r 1 + r 2 ; the ntersecton of two components of order e 1 and e 2 may be of codmenson greater than e 1 + e 2. The general case remans conjectural: let P 1,..., P n {x 1,..., x n }, be a prme component of {P} of dff. dm. 0, ord?

State of the art Jacob s weak bound for frst order systems. Lando 1970. Jacob s bound mples the dmenson conjecture. Cohn 1983. Jacob s bound for systems satsfyng Johnson s regularty hypothess. Kondrateva, Mkhalev and Pankratev 1982. Ordnary case. Extenson to the partal case. KMP 2008.

Provng the dmenson conjecture by extendng Rtt s method Let P1,..., P r {x 1,..., x n }, and be a prme component of {P}, we want to prove that the dfferental codmenson of s at most r. We make a change of varables: y = x η, where η s a generc zero of and get a new system Q. We complete the system wth n r generc lnear equatons and try to buld a Puseux seres soluton for Q: ζ = ζ 1 c + ζ 1 c α 1 + ζ 2 c α 2 + ζ 3 c α 3 +

Basc dea In Rtt s proof, each P s decomposed as a sum of homogeneous polynomals P = Σ k M k. Frst ζ 1 s chosen to be a generc soluton of M 1. If we have many equatons P = Σ k M,k, thngs reman unchanged provded that the M,1, completed wth the lnear equatons defne a component of dfferental dmenson 0. New terms n the Puseux seres are recursvely defned, so that the equatons could be satsfed up to a gven degree.

What must be changed Frst, the component may be of a smaller codmenson. Ths poses no dffculty for the proof, but we need to add more lnear relatons. The component of the deal {M,1 } may have a smaller codmenson. Then, e.g., M 1,1 (r) belongs to the radcal of the algebrac deal [M 1,1,M 1,1,..., M 1,1 (r) ] + {M,1 }. Some relaton between the Σ k>1 M,k must be used nstead of P 1.

The new polynomal s decomposed agan n homogeneous components, and we terate the process. As n Rtt s proof the soluton for the homogegeous system s njected n the orgnal system, and correctng terms must be ntroduced. Ths s here that the homogeneous sytem must have the rght codmenson. We have then a set of soluton dependng of at most n r arbtrary functons (due to the generc lnear equatons) and contanng η (takng c=0).

From the dmenson conjecture to the order bound We use Jacob s shortest reducton method. Let the l be the mnmal ntegers such that n the matrx a,j + l we may fnd n transversal maxma. We may compute characterstc sets (normal forms) for components of dff. Dm. 0 and order usng dervatves of P up to order l f Jacob s truncated determnant does not vansh. They are defned accordng to orderng on dervatve that are orderly wth respect to Jacob order defned by J. ord x = m.

Proof by recurrence on The case = 0 s straghtforward : the generators must be all of order 0. Let s assume the result s true for < s and consder a system P wth = s.

Geometrcal verson of Jacob s method. We start wth the equatons wth maxmal dervatves of smaller J. ord (those wth the maxmal l ). We take the radcal deal Q 0 they defne and for each prme component I n t... We take the radcal of the deal generated by the next set of equatons and I,..., I (l 1 -l 2 ) and for each prme component n t... We may terate the process unless the component contans elements of smaller J. ord not already n I,..., I (l 1 -l 2 ).

fle:///c:/users/ollver/desktop/101msdcf/dsc06336.jpg l 6 =0 l 5 l 4 l 3 l 2 l 1

The good case We have been able to terate the process up to the last set of generators wthout dscoverng new elements of non maxmal J. ord. The components of dff. Dm. 0 we have found are all of order.

fle:///c:/users/ollver/desktop/101msdcf/dsc06335.jpg

Another good case Some lower J. order element Q appeared... We may replace some P wth Q. The components of dff. Dm. 0 we look for wll be ncluded n those of ths new system... That has a smaller. Then, we use the recurrence hypothess.

A more complcated stuaton Sngular components : how to make them appear? Dfferentate more than the l. Very dffcult to control the order... Add extra equatons (separants). Usng some elmnaton, we may reduce to an extra equaton Q of smaller J. order than some P.

fle:///c:/users/ollver/desktop/101msdcf/dsc06337.jpg

Where we use the dmensonal property Consder the P j wth j. We may neglect the components they defne that are of codmenson smaller than ther number: they are ncluded n components of a system defned by droppng some equatons and of the same dmenson, so they cannot be used to make appear components of P of dff. dm. 0. Assume some dervatve of Q s n the deal generated by some of the remanng components I and lower dervatves of Q. Then t s enough to add only these lower dervatves to some regular component of {P}: the order must be less than.

The end If not, we may replace P, whch s greater than Q, by Q. The component we consder wll be n some component of the new system. And as Q has smaller J. order, the new system has smaller.