RELATION BETWEEN SOLVABILITY OF SOME MULTIVARIATE INTERPOLATION PROBLEMS AND THE VARIETY OF SUBSPACE ARRANGEMENT. The subspace arrangement is π= π π=1 π»π where π»π are linear subspaces of β π . π»π = π π=1{ π§1 , … , π§π : π π=1 ππ π§π = 0}, dim π»π = π − π π ⊆ β π is an affine variety πππ π = πππ₯{πππ π»π }=d-s Algebraic geometry studies the property π George David Birkhoff Birkhoff Interpolation (in one variable) Birkhoff interpolation is an extension of Hermite interpolation. It involves matching of values and derivatives of a function at certain points without the requirement that the derivatives are consecutive. Example: find a polynomial π π₯ = π0 + π1 π₯ + π2 π₯ 2 + π3 π₯ 3 : π π§1 = π1 , π ′ (π§1 ) = π2 , π′ π§2 = π3 π′′′ π§2 = π4 If this equations have unique solution for all distinct π§1 , π§2 ∈ π and all π1 , π2 , π3 , π4 ∈ π the problem (ΰ€3 , 0,1 , {1,3}) is regular. det 1 π§1 π§1 2 π§1 3 0 1 2π§1 3π§1 2 0 1 2π§2 3π§2 2 0 0 0 3 =π (π§1 , π§2 ) = 6π§2 −6π§1 ? π΅ φ : = {(π§1 , π§2 ) :π (π§1 , π§2 )} ⊆ π = {(π§1 , π§2 ) :π§1 =π§2 )} Example: find a polynomial π π₯ = π0 + π1 π₯ + π2 π₯ 2 + π3 π₯ 3 π π§1 = π1 , π ′ ′(π§1 ) = π2 , π′ π§2 = π3 , det 1 π§1 π§1 2 π§1 3 0 1 2π§2 3π§2 2 0 0 2 6π§1 0 0 0 6 π′′′ π§2 = π4 =π (π§1 , π§2 ) =12 the problem (ΰ€3 , 0,2 , {1,3}) is regular, completely regular. π΅ φ : = {(π§1 , π§2 ) :π (π§1 , π§2 )=0}=∅ ⊆ π΅ π§1 − π§2 = {(π§1 , π§2 ) :π§1 =π§2 )} 0,2 {1,3})= ∅ πππ 0,2 ∪ {1,3}={0,1,2,3,} Birkhoff Interpolation problem: (ΰ€π ,π1 , … ππ ), ππ ⊆ {0, … , π + 1} where ππ ⊆ {0, … , π + 1}; π1 #ππ = π + 1 = πππ ΰ€π is regular if for any set of k district points {π§1 , … , π§π } ⊆ π the interpolation problem and any f∈ ΰ€ there exists unique p∈ ΰ€π such that π (π) (π§π )=π(π) (π§π ) for all k∈ ππ and all j=1,…,k. George Pólya Isaac Schoenberg The problem (ΰ€π ,π1 , … ππ ), ππ ⊆ 0, … , π + 1 is regular if 1) k=n+1, ππ ={0}, Lagrange interpolation 2) For k>0, k∈ ππ (π − 1) ∈ ππ , Hermite interpolation 3) ππ ππ = ∅, ( ∪ ππ = {0, … , π + 1} ). β π, π >1 Take (ΰ€=span{1,x,y,xy}, π1 , π2 ), π1 , π2 ⊆ 0,1 × 0,1 πππ ππ£πππ’ππ‘π π‘βπ ππππππ πππππππ πππ‘ππππππππ‘ ππ‘ π‘π€π πππππ‘π : π§1 =(π§1,1 , π§1,2 ), π§2 =(π§2,1 , π§2,2 ): πππ‘ = π (π§1 , π§2 )= π₯1 =π (π§1,1 , π§1,2 , π§2,1 , π§2,2 ) ∈ β[π§1,1 , π§1,2 , π§2,1 , π§2,2 ], π΅ φ ⊆ β4 ; dim π΅ φ =3 if π ≠ ππππ π‘. π = { π§1,1 , π§1,2 , π§2,1 , π§2,2 : π§1 = π§2 , π. π. , π§1,1 = π§2,1 , π§1,2 = π§2,2} dim π =2 If the scheme is regular then π΅ φ ⊆π George G. Lorentz If π ≠ ππππ π‘. If π = ππππ π‘. ≠ 0 ππππ΅ φ ≤ ππππ πππ π΅ φ =3≤ ππππ = 2 π πβπππ ππ πππππππ‘πππ¦ ππππ’πππ πΌπ π1 π2 ≠ ∅ and π§1 = π§2 then the determinant φ contains two identical rows (columns), hence π1 π2 = ∅ The same is true in real case Birkhoff Interpolation (in several variables) Carl de Boor Given a subspace ΰ€ ⊆ π[π₯1 , … , π₯π ], a collection of subspaces π1 , … , ππ ⊆ ΰ€, a set of points π§1 , … , π§π ∈ ππ and a function f ∈ π[π₯1 , … , π₯π ] we want to find a polynomial p ∈ ΰ€ such that π(π·1 , … , π·π )f(π§π )= π (π·1 , … , π·π )f(π§π ) for all q∈ ππ . Amos Ron The scheme (ΰ€,π1 , … ππ ) is regular if the problem has a unique solutions for all f ∈ π[π₯1 , … , π₯π ] and all distinct π§1 , … , π§π ∈ ππ , it is completely regular if it has unique solution for all π§1 , … , π§π ∈ ππ . Birkhoff Interpolation (in several variables) πππ‘ = π (π§1 ,…, π§π ) is a polynomial in π × π π£ππππππππ π= π≠π π»π,π ; π»π,π = (π§1 , … , π§π ∈ βππ : π§π = π§π } dim π =max{dimπ»π,π }=kd-d ππππ’πππππ‘π¦ πππππ π΅ φ ⊆ π If π ≠ ππππ π‘. πππ π΅ φ =dk-1≤ ππ − π= ππππ Theorem: If (ΰ€,π1 , … ππ ) is regular then ππ ∩ ππ = {0} Conjecture: it is true in the real case False Rong-Qing Jia A. Sharma HAAR SUBSPACES AND HAAR COVERINGS Alfréd Haar Definition: H=span { β1 , … , βπ } ππ πππ πππ¦ πππ π‘ππππ‘ πππππ‘π {π§1 , … , π§π } ⊆ ππ the determinant πππ‘ βπ (π§π ) ≠ 0 (The Lagrange interpolation problem is well posed) ΰ€π−1 ππ π π»πππ π π’ππ ππππ ππ π π₯ πππ π‘βπ π’ππππ’π π»πππ π π’π ππππ ππ β[x]. For d>1, n>1 there are no n-dimensional Haar subspaces in π π₯1 , … , π₯π or even in C(ππ ) J. C. Mairhuber in real case and I. Schoenberg in the complex case;. or from the previous discussion: (ΰ€,{0},{0},…,{0}) Definition: A family of n-dimensional subspaces {π»1 , … , π»π } ππ C(ππ ) is a Haar covering of πd if for any distinct points {π§1 , … , π§π } ⊆ ππ the Lagrange interpolation problem is well posed in one of these spaces. π»π =span { β1 (π) , … , βπ (π) } {π»1 , … , π»π }is a Haar covering of πd if for any distinct points {π§1 , … , π§π } ⊆ ππ πππ ππ π‘βπ πππ‘ππππππππ‘π Δπ π§1 , … , π§π = πππ‘ β1 (π) (π§π ) ≠ 0 Question: What is the minimal number s:=ηπ (ππ ) of n-dimensional subspaces {π»1 , … , π»π } ππ π π₯1 , … , π₯π that forms a Haar covering of πd and what are those subspaces? Kyungyong Lee Conjecture: : ηπ β2 = π π¦ η 2 π2 = 2 π»1 = π πππ 1, π₯ , π»2 = π πππ 1, π¦ π₯ Theorem (Stefan Tohaneanu& B.S.): η3 β2 = 3 The three spaces π»1 = π πππ 1, π₯, π¦ , π»2 = π πππ 1, π₯, π₯ 2 , π»3 = π πππ 1, π¦, π¦ 2 will do: It remains to show that no two subspaces can do the job. π»1 =span { β1 (1) , β2 (1) , β3 (1) } ; π»2 =span { β1 (2) , β2 (2) , β3 (2) } I want show that there are three distinct points π§1 , π§2 , π§3 = 0 π π’πβ π‘βππ‘ Δ1 π§1 , π§2 , 0 =0, Δ2 π§1 , π§2 , 0 =0 π§1 = (π§1,1 , π§1,2), π§2 = (π§2,1 , π§2,2 ), πβ 0,0, a, b ∪ {(a, b, 0,0) ∪ {(a, b, a, b)} π΅ Δ1 , Δ2 ={(π§1,1 , π§1,2 , π§2,1 , π§2,2 ): Δ1 = 0 πππ Δ2 =0} π΅ Δ1 , Δ2 ⊆ β4 πβ 0,0, a, b ∪ {(a, b, 0,0) ∪ {(a, b, a, b)} (0,0,1,1) (0,0,0,0) π π΅ Δ1 , Δ2 = π ? dim π΅ Δ1 , Δ2 =2=dim π (-1,-1,0,0) Not every two-dimensional variety in β4 can be formed as a set of common zeroes of two polynomials. The ones that can are called (set theoretic) complete intersections. As luck would have it, π is not a complete intersection: Robin Hartshorne π\(0,0,0,0) is not connected… Alexander Grothendieck A very deep theorem states that a two dimensional complete intersection in β4 can not be disconnected by removing just one point (Tom McKinley and B.S.): η3 (π2 )=3 η4 (π2 ) ≤ 4 d ≤ η π ππ < ∞ π»1 = π πππ 1, π₯, π¦ , π»2 = π πππ 1, π₯, π₯ 2 , π»3 = π πππ 1, π¦, π¦ 2 The family of D-invariant subspaces spanned by monomials form a finite Haar covering. There are too many of them: for n=4 in 2 dimensions there are five: 1 π¦2 π¦ 1 π₯ π₯2 π₯3 π¦ 1 π₯ π¦ 1 π₯2 Yang tableaux π₯ Cojecture: ηπ (ππ )= π¦3 π¦2 π¦ 1 π+π−2 π−1 π₯π¦ π₯ köszönöm Thank You Thank You köszönöm