Sparse Approximation of Pdes Based on Compressed Sensing
Total Page:16
File Type:pdf, Size:1020Kb
Sparse Approximation of PDEs based on Compressed Sensing Simone Brugiapaglia Department of Mathematics Simon Fraser University Retreat for Young Researchers in Stochastics September 24, 2016 1 Introduction We address the following question Can we employ Compressed Sensing to solve a PDE? In particular, we consider the weak formulation of a PDE find u 2 U : a(u; v) = F(v); 8v 2 V; focusing on the Petrov-Galerkin (PG) discretization method [Aziz and Babuška, 1972]. Motivation: I reduce the computational cost associated with a classical PG discretization; I situations with a limited budget of evaluations of F(·); I better theoretical understanding of the PG method. Case study: . Advection-diffusion-reaction (ADR) equation, with 1 d U = V = H0 (Ω), Ω = [0; 1] , and a(u; v) = (ηru; rv) + (b · ru; v) + (ρu, v); F(v) = (f; v): 2 Compressed Sensing CORSING (COmpRessed SolvING) A theoretical study of CORSING 3 Compressed Sensing (CS) [D. Donoho, 2006; E. Candès, J. Romberg, and T. Tao, 2006] A sparse vector u N Consider a signal s 2 C , sparse w.r.t. 2 Ψ 2 CN×N : 1.5 1 i u 0.5 s = Ψu and kuk0 =: s N; 0 −0.5 −1 where kuk0 := #fi : ui 6= 0g. 0 20 40 60 80 100 component i It can be acquired by means of m N linear and non-adaptive measurements hs; 'ii =: fi; for i = 1; : : : ; m: N×m If we consider the matrix Φ = ['i] 2 C , we have Au = f; H where A = Φ Ψ 2 Cm×N and f 2 Cm. 4 Sensing phase A picture to have in mind f ΦtH* u = sensing matrix measurements vector sparsity basis unknown sparse signal Since m N, the system Au = f is highly underdetermined. How to recover the right u among its infinite solutions? 5 Recovery: finding a needle in a haystack Thanks to the sparsity hypothesis, we can resort to sparse recovery techniques. We aim at approximating the solution to (P0) min kuk0; s.t. Au = f: N u2C Unfortunately, in general (P0) is a NP-hard problem... / There are computationally tractable strategies to approximate it! In, particular, we employ the greedy algorithm Orthogonal Matching Pursuit (OMP) to approximate " s (P0) min kuk0 (P0) min kAu − fk2 N or N u2C u2C s.t. kAu − fk2 ≤ " s.t. kuk0 ≤ s: Another valuable option is convex relaxation (not discussed here) (P1) min kuk1; s.t. Au = f: N u2C 6 Orthogonal Matching Pursuit (OMP) Input: m×N 2 Matrix A 2 C , with ` -normalized columns m Vector f 2 C Tolerance on the residual " > 0 (or else, sparsity s 2 [N]) Output: " s Approximate solution u to (P0) (or else, (P0)) Procedure: 1: S ; . Initialization 2: u 0 3: while kAu − fk2 > " (or else, kuk0 < s) do H 4: j arg max j[A (Au − f)]j j . Select new index j2[N] 5: S S[f jg . Enlarge support 6: u arg min kAz − fk2 s.t. supp(z) ⊆S . Minimize residual N z2C 7: end while 8: return u s I The computational cost for the (P0) formulation is in general O(smN). 7 Recovery results based on the RIP Many important recovery results in CS are based on the Restricted Isometry Property (RIP). Definition (RIP) m×N A matrix A 2 C satisfies the RIP(s; δ) iff 2 2 2 N N (1 − δ)kuk2 ≤ kAuk2 ≤ (1 + δ)kuk2; 8u 2 Σs := fv 2 C : kvk0 ≤ sg: Among many others, the RIP implies the following recovery result for OMP. [T. Zhang, 2011; A. Cohen, W. Dahmen, R. DeVore, 2015] Theorem (RIP ) OMP recovery) There exist K 2 N, C > 0 and δ 2 (0; 1) s.t. for every s 2 N, the following holds: if A 2 RIP((K + 1)s; δ); m then, for any f 2 C , the OMP algorithm computes in Ks iterationsa solution u that fulfills kAu − fk2 ≤ C inf kAw − fk2: N w2Σs > 8 Compressed Sensing CORSING (COmpRessed SolvING) A theoretical study of CORSING 9 The reference problem Given two Hilbert spaces U; V , consider the following problem find u 2 U : a(u; v) = F(v); 8v 2 V; (1) where a : U × V ! R is a bilinear form and F 2 V ∗. We will assume a(·; ·) to fulfill a(u; v) 9α > 0 : inf sup ≥ α; (2) u2U v2V kukU kvkV ja(u; v)j 9β > 0 : sup sup ≤ β; (3) u2U v2V kukU kvkV sup a(u; v) > 0; 8v 2 V n f0g: (4) u2U . (2) + (3) + (4) =) 9! solution to (1). [J. Nečas, 1962] . We will focus on advection-diffusion-reaction (ADR) equations. 10 The Petrov-Galerkin method d Given Ω ⊆ R , consider the weak formulation of an ADR equation: 1 1 find u 2 H0 (Ω) : (ηru; rv) + (bru; v) + (ρu, v) = (f; v); 8v 2 H0 (Ω): | {z } | {z } a(u;v) F(v) (ADR) N 1 M 1 Choose U ⊆ H0 (Ω) and V ⊆ H0 (Ω) with N M U = spanf 1; : : : ; N g;V = spanf'1;:::;'M g | {z } | {z } trials tests Then we can discretize (ADR) as Aub = f;Aij = a( j;' i); fi = F('i) M×N M with A 2 C , f 2 C . A common choice is M = N. I Examples of Petrov-Galerkin methods: Finite elements, spectral methods, collocation methods, etc. 11 The main analogy A fundamental analogy guided us through the development of our method... Petrov-Galerkin method: Sampling: solution of a PDE () signal tests (bilinear form) measurements (inner product) Reference: Compressed solving: a numerical approximation technique for elliptic PDEs based on compressed sensing S. B., S. Micheletti, S. Perotto Comput. Math. Appl. 2015; 70(6):1306-1335 12 Related literature Ancestors: PDE solvers based on `1-minimization 1988 [J. Lavery, 1988; J. Lavery, 1989] Inviscid Burgers’ equation, conservation laws 2004 [J.-L. Guermond, 2004; J.-L. Guermond and B. Popov, 2009] Hamilton-Jacobi, transport equation CS techniques for PDEs 2010 [S. Jokar, V. Mehrmann, M. Pfetsch, and H. Yserentant, 2010] Recursive mesh refinement based on CS (Poisson equation) 2011 [A. Doostan and H. Owhadi, 2011; K. Sargsyan et al., 2014; H. Rauhut and C. Schwab, 2014; J.-L. Bouchot et al., 2015] Application of CS to parametric PDEs and Uncertainty Quantification 2015 [S. B., S. Micheletti, S. Perotto, 2015; S. B., F. Nobile, S. Micheletti, S. Perotto, 2016] CORSING for ADR problems 13 CORSING (COmpRessed SolvING) Assembly phase ¬ Choose two sets of N independent elements of U and V : trials ! f 1; : : : ; N g; f'1;:::;'N g tests; choose m N tests f'τ1 ;:::;'τm g: ® build A 2 Cm×N and f 2 Cm as [A]ij := a( j;' τi )[f]i := F('τi ): Recovery phase N Find a compressed solution um to Au = f, via sparse recovery. 14 CORSING (COmpRessed SolvING) Assembly phase ¬ Choose two sets of N independent elements of U and V : trials ! f 1; : : : ; N g; f'1;:::;'N g tests; choose m N tests f'τ1 ;:::;'τm g: ® build A 2 Cm×N and f 2 Cm as [A]ij := a( j;' τi )[f]i := F('τi ): Recovery phase N Find a compressed solution um to Au = f, via sparse recovery. 15 Classical case: square matrices When dealing with Petrov-Galerkin discretizations, one usually ends up with a big square matrix. 1 2 3 4 5 6 7 ####### 2 3 2 3 2 3 '1 ! × × × × × × × u1 F('1) '2 ! 6 × × × × × × × 7 6u27 6F('2)7 6 7 6 7 6 7 '3 ! 6 × × × × × × × 7 6u37 6F('3)7 6 7 6 7 6 7 '4 ! 6 × × × × × × × 7 6u47 = 6F('4)7 6 7 6 7 6 7 '5 ! 6 × × × × × × × 7 6u57 6F('5)7 6 7 6 7 6 7 '6 ! 4 × × × × × × × 5 4u65 4F('6)5 '7 ! × × × × × × × u7 F('7) | {z } a( j ;'i) 16 “Compressing” the discretization We would like to use only m random tests instead of N, with m N... 1 2 3 4 5 6 7 ####### 2 3 2 3 2 3 '1 ! ××××××× u1 F('1) '2 ! 6 ××××××× 7 6u27 6F('2)7 6 7 6 7 6 7 '3 ! 6 ××××××× 7 6u37 6F('3)7 6 7 6 7 6 7 '4 ! 6 ××××××× 7 6u47 = 6F('4)7 6 7 6 7 6 7 '5 ! 6 ××××××× 7 6u57 6F('5)7 6 7 6 7 6 7 '6 ! 4 ××××××× 5 4u65 4F('6)5 '7 ! ××××××× u7 F('7) | {z } a( j ;'i) 17 Sparse recovery ...in order to obtain a reduced discretization. 1 2 3 4 5 6 7 ####### 2 3 ' ! ××××××× u1 F(' ) 2 = 2 ' ! ××××××× 6u27 F(' ) 5 6 7 5 | {z } 6u37 a( j ;'i) 6 7 6u47 6 7 6u57 6 7 4u65 u7 The solution is then computed using sparse recovery techniques. 18 One heuristic criterion commonly used in CS is to choose one basis sparse in space, and the other in frequency. Hierarchical hat functions Sine functions [O. Zienkiewicz et al., 1982] H 0,0 0.5 0.5 0.4 0.4 H H 0.3 S 1,0 1,1 2 S 0.2 1 0.3 S H H H H 5 2,0 2,1 2,2 2,3 0.1 S 3 0.2 0 S 4 0.1 −0.1 −0.2 0 0 0.2 0.4 0.6 0.8 1 −0.3 0 0.2 0.4 0.6 0.8 1 H S We name the corresponding strategies CORSING HS and SH. How to choose f jg and f'ig? 19 H 0,0 0.5 0.5 0.4 0.4 H H 0.3 S 1,0 1,1 2 S 0.2 1 0.3 S H H H H 5 2,0 2,1 2,2 2,3 0.1 S 3 0.2 0 S 4 0.1 −0.1 −0.2 0 0 0.2 0.4 0.6 0.8 1 −0.3 0 0.2 0.4 0.6 0.8 1 How to choose f jg and f'ig? One heuristic criterion commonly used in CS is to choose one basis sparse in space, and the other in frequency.