Verification of Inequalities (i) Four practical mechanisms The role of CAS in analysis (ii) Applications Kent Pearce Texas Tech University Presentation: January 2008 Question Given a function f on an interval (a, b), what does it take to show that f is non-negative on (a, b)? Proof by Picture Maple, Mathematica, Matlab, Mathcad, Excel, Graphing Calculators (P)Lots of Dots (P)Lots of Dots (P)Lots of Dots (P)Lots of Dots y f ( x) 1 2x 1 (P)Lots of Dots y f ( x) 1 2x 1 Blackbox Approximations Polynomial Blackbox Approximations Transcendental / Special Functions Practical Methods A. Sturm Sequence Arguments B. Linearity / Monotonicity Arguments C. Special Function Estimates D. Grid Estimates Applications "On a Coefficient Conjecture of Brannan," Complex Variables. Theory and Application. An International Journal 33 (1997) 51_61, with Roger W. Barnard and William Wheeler. "A Sharp Bound on the Schwarzian Derivatives of Hyperbolically Convex Functions," Proceeding of the London Mathematical Society 93 (2006), 395_417, with Roger W. Barnard, Leah Cole and G. Brock Williams. "The Verification of an Inequality," Proceedings of the International Conference on Geometric Function Theory, Special Functions and Applications (ICGFT) (accepted) with Roger W. Barnard. "Iceberg-Type Problems in Two Dimensions," with Roger.W. Barnard and Alex.Yu. Solynin Practical Methods A. Sturm Sequence Arguments B. Linearity / Monotonicity Arguments C. Special Function Estimates D. Grid Estimates Iceberg-Type Problems Iceberg-Type Problems Dual Problem for Class 0 Let D= {z : 0 | z | 1} and let 1 { f ( z ) a0 a1 z : f is analytic, z univalent on D}. For f let E f \ f (D) and 0 { f | 0 E f }. For 0 < h < 4, let H h {z | Re( z ) h}. Find A(h) max area( E f H h ) f 0 Iceberg-Type Problems Extremal Configuration Symmetrization Polarization Variational Methods Boundary Conditions Iceberg-Type Problems Iceberg-Type Problems We obtained explicit formulas for A = A(r) and h = h(r). However, the orginial problem was formulated to find A as a function of h, i.e. to find A = A(h). To find an explicit formulation giving A = A(h), we needed to verify that h = h(r) was monotone. Sturm Sequence Arguments General theorem for counting the number of distinct roots of a polynomial f on an interval (a, b) N. Jacobson, Basic Algebra. Vol. I., pp. 311315,W. H. Freeman and Co., New York, 1974. H. Weber, Lehrbuch der Algebra, Vol. I., pp. 301313, Friedrich Vieweg und Sohn, Braunschweig, 1898 Sturm Sequence Arguments Sturm’s Theorem. Let f be a non-constant polynomial with rational coefficients and let a < b be rational numbers. Let S f { f 0 , f1 , , f s } be the standard sequence for f . Suppose that f (a) 0, f (b) 0. Then, the number of distinct roots of f on (a, b) is Va Vb where Vc denotes the number of sign changes of S f (c) { f 0 (c), f1 (c), , f s (c)} Sturm Sequence Arguments Sturm’s Theorem (Generalization). Let f be a non-constant polynomial with rational coefficients and let a < b be rational numbers. Let S f { f 0 , f1 , , f s } be the standard sequence for f . Suppose that f (a) 0, f (b) 0. Then, the number of distinct roots of f on (a, b] is Va Vb where Vc denotes the number of sign changes of S f (c) { f 0 (c), f1 (c), , f s (c)} Sturm Sequence Arguments For a given f, the standard sequence S f is constructed as: f0 f f1 f f 2 : f 0 f1q1 f 2 f 3 : f1 f 2 q2 f 3 Sturm Sequence Arguments Polynomial Sturm Sequence Arguments Polynomial Linearity / Monotonicity Consider f ( x, Z ) c0 ( x) c1 ( x)Z where Z Let f ( x) f ( x, Z ) Z c0 ( x) c1 ( x) , f ( x) f ( x, Z ) Z c0 ( x) c1 ( x) Then, min { f ( x), f ( x)} f ( x, Z ) max{ f ( x), f ( x)} x( a ,b ) x( a ,b ) Iceberg-Type Problems We obtained explicit formulas for A = A(r) and h = h(r). However, the orginial problem was formulated to find A as a function of h, i.e. to find A = A(h). To find an explicit formulation giving A = A(h), we needed to verify that h = h(r) was monotone. Iceberg-Type Problems From the construction we explicitly found where Iceberg-Type Problems Iceberg-Type Problems where Iceberg-Type Problems It remained to show g g (r ) (c0 c1P) (d0 d1P)Q was non-negative. In a separate lemma, we showed 0 < Q < 1. Hence, using the linearity of Q in g, we needed to show g 0 (c0 c1P) (d 0 d1P) 0 g1 (c0 c1 P) (d 0 d1P) 1 were non-negative Iceberg-Type Problems In a second lemma, we showed s < P < t where Let g0,s g0 Ps , g0,t g0 Pt , g1, s g1 Ps , g1,t g1 Pt . Each g0, s , g0,t , g1, s , g1,t is a polynomial with rational coefficients for which a Sturm sequence argument show that it is non-negative. Practical Methods A. Sturm Sequence Arguments B. Linearity / Monotonicity Arguments C. Special Function Estimates D. Grid Estimates Notation & Definitions D {z :| z | 1} Notation & Definitions D {z :| z | 1} hyperbolic metric 2 | dz | ( z ) | dz | 2 1 | z | Notation & Definitions D {z :| z | 1} Hyberbolic Geodesics Notation & Definitions D {z :| z | 1} Hyberbolic Geodesics Hyberbolically Convex Set Notation & Definitions D {z :| z | 1} Hyberbolic Geodesics Hyberbolically Convex Set Hyberbolically Convex Function Notation & Definitions D {z :| z | 1} Hyberbolic Geodesics Hyberbolically Convex Set Hyberbolically Convex Function Hyberbolic Polygon o Proper Sides Examples k ( z ) 2 z (1 z ) (1 z ) 2 4 2 z k Examples f ( z ) tan where z (1 2 2 1 4 2 cos 2 ) 0 2 , 0 2 K (cos ) f d Schwarz Norm || S f ||D For f A(D) let Sf 2 f 1 f f 2 f and || S f ||D sup{D2 ( z) | S f ( z) |: z D} where 1 D ( z ) 1 | z |2 Extremal Problems for || S f ||D Euclidean Convexity Nehari (1976): f (D) convex || S f ||D 2 Extremal Problems for || S f ||D Euclidean Convexity Nehari (1976): f (D) convex || S f ||D 2 Spherical Convexity Mejía, Pommerenke (2000): f (D) convex || S f ||D 2 Extremal Problems for || S f ||D Euclidean Convexity Nehari (1976): f (D) convex || S f ||D 2 Spherical Convexity Mejía, Pommerenke (2000): f (D) convex || S f ||D 2 Hyperbolic Convexity Mejía, Pommerenke Conjecture (2000): f (D) convex || S f ||D 2.3836 Verification of M/P Conjecture "A Sharp Bound on the Schwarzian Derivatives of Hyperbolically Convex Functions," Proceeding of the London Mathematical Society 93 (2006), 395_417, with Roger W. Barnard, Leah Cole and G. Brock Williams. "The Verification of an Inequality," Proceedings of the International Conference on Geometric Function Theory, Special Functions and Applications (ICGFT) (accepted) with Roger W. Barnard. Verification of M/P Conjecture Invariance under disk automorphisms Reduction to hyperbolic polygonal maps Reduction to | S f (0) | Julia Variation Reduction to hyperbolic polygonal maps with at most two proper sides Reduction to (1 x2 )2 | S f ( x) |, 0 x 1, 0 1 Reduction to max S f (0) 0 1 Graph of S f ( ) * (0) 2(c ) 2 Two-sided Polygonal Map f ( z ) tan where z (1 2 2 1 4 2 cos 2 ) 0 2 , 0 2 K (cos ) f d Special Function Estimates Parameter /2 K ( y) where y cos Special Function Estimates Upper bound Special Function Estimates Upper bound Partial Sums Special Function Estimates (1 | z |2 )2 | S f ( z) | θ = 0.3π /2 (1 | z |2 )2 | S f ( z) | θ = 0.5π /2 Verification of M/P Conjecture Invariance under disk automorphisms Reduction to hyperbolic polygonal maps Reduction to | S f (0) | Julia Variation Reduction to hyperbolic polygonal maps with at most two proper sides Reduction to (1 x2 )2 | S f ( x) |, 0 x 1, 0 1 Reduction to max S f (0) 0 1 Verification where c cos 2 , /2 , 1 x 1 K (cos ) Graph of c 2 0 Verification where c cos 2 , /2 , 1 x 1 K (cos ) Verification c3 0 Straightforward to show that q p 0 In qm make a change of variable c 2 y 1 2 Verification c3 0 Obtain a lower bound for qm by estimating via an upper bound * q Sturm sequence argument shows m qm p8 is non-negative Grid Estimates Grid Estimates Given A) grid step size h B) global bound M for maximum of | f ( x) | Theorem Let f be defined on [a, b]. Let M max | f ( x) | . Let 0 and suppose that N is x[ a ,b ] chosen so that h (b a) / N . Let L be the f ( x) lattice L {a jh : 0 j N } . Let m min xL If m M , 2 then f is non-negative on [a, b]. Grid Estimates Maximum descent argument Grid Estimates Two-Dimensional Version Grid Estimates Maximum descent argument Verification where c cos 2 , /2 , 1 x 1 K (cos ) Verification c2 The problem was that the coefficient c2 ( , x) was not globally positive, specifically, it was not positive for 1 x 4 5 , 0 2 . We showed that p4 (t ) 0 by showing that q(t ) 0 where q(t ) c3 ( , x)t 2 c2 ( , x)t c1 ( , t ) 0 < t < 1/4. Verification c2 For the case, 4 2 expand q(t) in powers of q(t ) c4 (c, x, t ) 4 c2 (c, x, t ) 2 c0 (c, x, t ). Noting that c4 (c, x, t ) and c2 (c, x, t ) are negative, we replaced by an upper bound ( of 1) to obtain a lower bound q(t ) q0 (t ) q(t ) 1 e2 ( y, x)t 2 e1 ( y, x)t e0 ( y, x) where y cos Verification c2 Finally, we introduced a change of variable to obtain q0* (t ) q0 (t ) x 1 2 w /10 e2 ( y, w)t 2 e1 ( y, w)t e0 ( y, w) where the coefficients are polynomials (with rational coefficients) in w, y and 0 w 1, 0 y 2 2 , 0t 1 4 Verification c2 Used Lemma 3.3 to show that the endpoints q0* (0) e0 ( y, w) and q0* ( 1 4) are non-negative. We partition the parameter space into subregions: Verification c2 Application of Lemma 3.3 to q0* ( 1 ) 4 After another change of variable, we needed to show that r 0 where for 0 < w < 1, 0 < m < 1 Verification c2 Verification c2 [0,1] [0,1] [0,1/ 2] [0,1/ 2] [0,1/ 2] [1/ 2,1] [1/ 2,1] [0,1/ 2] [1/ 2,1] [1/ 2,1] Quarter Square [0,1/2]x[0,1/2] M max{M w , M m } 35 M w 21, M m 35 Grid 50 x 50 m min r ( w j , mk ) 0.400 ( w j , mk )L M 0.350 Verification c2 Application of Lemma 3.3 to non-negativity of q0* (t ) on the subregion D: We showed that the discriminant of a related quadratic function was negative on D. That computation amounted to showing that a polynomial hl (m, v), of degree 16 in m and degree 40 in w, was non-negative Conclusions There are “proof by picture” hazards CAS numerical computations are rational number calculations CAS “special function” numerical calculations are inherently finite approximations There is a role for CAS in analysis There are various useful, practical strategies for rigorously establishing analytic inequalities