A new relaxed CQ algorithm for solving split feasibility problems in Hilbert spaces and its applications

Aviv Gibali; Dang Thi Mai; Nguyen The Vinh

doi:10.3934/jimo.2018080

Article Contents

2019, Volume 15, Issue 2: 963-984. Doi: 10.3934/jimo.2018080

This issue Previous Article Predicting 72-hour reattendance in emergency departments using discriminant analysis via mixed integer programming with electronic medical records Next Article Predicting non-life insurer's insolvency using non-kernel fuzzy quadratic surface support vector machines

A new relaxed CQ algorithm for solving split feasibility problems in Hilbert spaces and its applications

1.
Department of Mathematics, ORT Braude College, 2161002 Karmiel, Israel
2.
The Center for Mathematics and Scientific Computation, University of Haifa, Mt. Carmel, 3498838 Haifa, Israel
3.
Department of Mathematics, University of Transport and Communications, 3 Cau Giay Street, Hanoi, Vietnam

* Corresponding author: avivg@braude.ac.il
* Corresponding author: avivg@braude.ac.il

Received: October 31, 2017

Revised: December 31, 2017

Early access: June 2018

Published: January 2019

The first author work is supported by the EU FP7 IRSES program STREVCOMS, grant no. PIRSES-GA-2013-612669. The research of the third author was partially supported by University of Transport and Communications (UTC) [grant number T2018-CB-002] and Vietnam Institute for Advanced Study in Mathematics (VIASM).

Abstract Full Text(HTML) Figure(6) / Table(2) Related Papers Cited by

Abstract

Inspired by the works of López et al. [21] and the recent paper of Dang et al. [15], we devise a new inertial relaxation of the CQ algorithm for solving Split Feasibility Problems (SFP) in real Hilbert spaces. Under mild and standard conditions we establish weak convergence of the proposed algorithm. We also propose a Mann-type variant which converges strongly. The performances and comparisons with some existing methods are presented through numerical examples in Compressed Sensing and Sparse Binary Tomography by solving the LASSO problem.

Keywords:

Mathematics Subject Classification: Primary: 65K05, 65K10; Secondary: 49J52.

Citation:

Full Text(HTML)

Figure 1. Numerical results for m = 2¹⁰; n = 2¹²; k = 20

Download: Full-size image PowerPoint slide

Figure 2. Numerical results for m = 2¹⁰; n = 2¹²; k = 40

Download: Full-size image PowerPoint slide

Figure 3. Numerical results for m = 2¹²; n = 2¹³; k = 50

Download: Full-size image PowerPoint slide

Figure 4. Parallelbeam geometry set-up: a set of parallel rays is shot through the object from different directions. These are typically coined as one projection. Two projections are illustrated above. (Left) Illustration of a single projection corresponding to a measurement along one ray. The image domain $\Omega$ is tiled into pixels or mathematically Haar-basis functions. Hence, a single projection corresponds to the line integral over a piecewise constant function

Download: Full-size image PowerPoint slide

Figure 5. Vessel test image from [4] (left). We illustrate how such a $32\times 32$ image is sampled (right) along $45$ parallel and equidistant lines that are all perpendicular to $\theta: = (\cos(30^\circ),\sin(30^\circ))^T$

Download: Full-size image PowerPoint slide

Figure 6. Reconstructing a binary test image $u\in \mathbb{R}^{64\times 64}$ from [4] representing a vascular system containing larger and smaller vessels. The results are for the recover $u$ from a $15$ (limited number of) tomographic projections

Download: Full-size image PowerPoint slide

Table 1. Numerical results obtained by Algorithm 1 compared with Lopez et al. algorithm [21] ((8)-(10)) and Zhou and Wang [39,Algorithm 3.1]

K and m, n	Methods	$\epsilon $=10^-6
K and m, n	Methods	Iter	fk (10)	CPU time (sec.)	$\frac{\\|x^*-x^k\\|}{\\|x^0-x^{k+1}\\|}$
K = 10	Algorithm 1	3310	2.23e - 5	3.3438	0.0033
m = 2¹⁰	Lopez et al. [21]	3491	3.23e - 5	3.8125	0.0045
n = 2¹²	Zhou and Wang [39]	3991	2.79e - 4	4.7438	0.0012
K = 20	Algorithm 1	7180	9.6601e - 13	5.08281	3.380e - 5
m = 2¹⁰	Lopez et al. [21]	8901	7.6601e - 13	5.28281	3.271e - 5
n = 2¹²	Zhou and Wang [39]	8180	8.4375e - 12	6.478	3.260e - 5
K = 50	Algorithm 1	8780	5.7301e - 10	12.3312	4.276e - 4
m = 2¹²	Lopez et al. [21]	9901	6.6601e - 9	11.2761	4.238e - 4
n = 2¹³	Zhou and Wang [39]	9180	7.251e - 10	11.453	3.457e - 4

| Show Table

DownLoad: CSV

Table 2. Numerical results obtained by Algorithm 1 compared with Lopez et al. algorithm [21] ((8)-(9)) and Zhou and Wang [39,Algorithm 3.1]

K and m, n	Methods	$ \epsilon$ = 10^-6
K and m, n	Methods	Iter	fk (10)	CPU time (sec.)	$\frac{\\|x^*-x^k\\|}{\\|x^0-x^{k+1}\\|}$
K = 916	Algorithm 1	87	37.6590	0.6406	65.7008
m = 1365	Lopez et al. [21]	100	49.3198	0.3541	38.2665
n = 4096	Zhou and Wang [39]	100	48.5597	0.6565	68.3321

| Show Table

DownLoad: CSV

Related Papers

Cited by

References

[1]	F. Alvarez and H. Attouch, An inertial proximal method for maximal monotone operators via discretization of a nonlinear oscillator with damping, Set-Valued Anal., 9 (2001), 3-11. doi: 10.1023/A:1011253113155.
[2]	Q. H. Ansari and A. Rehan, Split feasibility and fixed point problems, In: Q. H. Ansari (ed.), Nonlinear Anal. Approx. The., Optim. Appl., Springer, (2014), 281-322.
[3]	J. P. Aubin, Optima and Equilibria: An Introduction to Nonlinear Analysis, Springer, 1993. doi: 10.1007/978-3-662-02959-6.
[4]	K. J. Batenburg, A network flow algorithm for reconstructing binary images from discrete X-rays, J. Math. Imaging Vis., 27 (2007), 175-191. doi: 10.1007/s10851-006-9798-2.
[5]	D. Butnariu and A. N. Iusem, Totally Convex Functions for Fixed Points Computation and Infinite Dimensional Optimization, Kluwer Academic Publishers, Dordrecht, The Netherlands, 2000. doi: 10.1007/978-94-011-4066-9.
[6]	H. H. Bauschke and J. M. Borwein, On projection algorithms for solving convex feasibility problems, SIAM Rev., 38 (1996), 367-426. doi: 10.1137/S0036144593251710.
[7]	C. Byrne, Iterative oblique projection onto convex sets and the split feasibility problem, Inverse Problems, 18 (2002), 441-453. doi: 10.1088/0266-5611/18/2/310.
[8]	C. Byrne, A unified treatment of some iterative algorithms in signal processing and image reconstruction, Inverse Problems, 20 (2004), 103-120. doi: 10.1088/0266-5611/20/1/006.
[9]	L. C. Ceng, Q. H. Ansari and J. C. Yao, An extragradient method for solving split feasibility and fixed point problems, Comput. Math. Appl., 64 (2012), 633-642. doi: 10.1016/j.camwa.2011.12.074.
[10]	Y. Censor and T. Elfving, A multiprojection algorithm using Bregman projection in a product space, Numer. Algorithms, 8 (1994), 221-239. doi: 10.1007/BF02142692.
[11]	Y. Censor, T. Bortfeld, B. Martin and A. Trofimov, A unified approach for inversion problems in intensity modulated radiation therapy, Phys. Med. Biol., 51 (2003), 2353-2365. doi: 10.1088/0031-9155/51/10/001.
[12]	Y. Censor, T. Elfving, N. Kopf and T. Bortfeld, The multiple-sets split feasibility problem and its applications for inverse problems, Inverse Problems, 21 (2005), 2071-2084. doi: 10.1088/0266-5611/21/6/017.
[13]	S. Chen, D. Donoho and M. Saunders, Atomic decomposition by basis pursuit, SIAM J. Comput., 20 (1998), 33-61. doi: 10.1137/S1064827596304010.
[14]	Y. Dang and Y. Gao, The strong convergence of a KM-CQ-like algorithm for a split feasibility problem, Inverse Problems, 27 (2011), 015007, 9 pp. doi: 10.1088/0266-5611/27/1/015007.
[15]	Y. Dang, J. Sun and H. K. Xu, Inertial accelerated algorithms for solving a split feasibility problem, J. Indus. Manage. Optim., 13 (2017), 1383-1394. doi: 10.3934/jimo.2016078.
[16]	A. Gibali, L.-W. Liu and Y.-C. Tang, Note on the modified relaxation CQ algorithm for the split feasibility problem, Optim. Lett., (2017), 1-14. doi: 10.1007/s11590-017-1148-3.
[17]	A. Gibali and S. Petra, DC-Programming versus l₀-Superiorization for Discrete Tomography, To appear in Analele Ştiinţifice ale Universitatii Ovidius Constanţa, Seria Mathematica, 2017.
[18]	K. Goebel and S. Reich, Uniform Convexity, Hyperbolic Geometry, and Nonexpansive Mappings, Marcel Dekker, New York and Basel, 1984.
[19]	P. C. Hansen and M. Saxild-Hansen, AIR tools - a MATLAB package of algebraic iterative reconstruction methods, J. Comput. Appl. Math., 236 (2012), 2167-2178. doi: 10.1016/j.cam.2011.09.039.
[20]	M. Li and Y. Yao, Strong convergence of an iterative algorithm for λ-strictly pseudocontractive mappings in Hilbert spaces, Analele Ştiinţifice ale Universitatii Ovidius Constanţa, Seria Mathematica, 18 (2010), 219-228.
[21]	G. López, V. Martín-Márquez, F. Wang and H. K. Xu, Solving the split feasibility problem without prior knowledge of matrix norms, Inverse Problems, 28 (2012), 085004, 18 pp. doi: 10.1088/0266-5611/28/8/085004.
[22]	Y. Lou and M. Yan, Fast l₁ - l₂ Minimization via a proximal operator, arXiv: 1609.09530.
[23]	P. E. Maingé, Approximation methods for common fixed points of nonexpansive mappings in Hilbert spaces, J. Math. Anal. Appl., 325 (2007), 469-479. doi: 10.1016/j.jmaa.2005.12.066.
[24]	P. E. Maingé, Convergence theorems for inertial KM-type algorithms, J. Comput. Appl. Math., 219 (2008), 223-236. doi: 10.1016/j.cam.2007.07.021.
[25]	P. E. Maingé, Strong convergence of projected subgradientmethods for nonsmooth and nonstrictly convex minimization, Set-Valued Anal., 16 (2008), 899-912. doi: 10.1007/s11228-008-0102-z.
[26]	A. Moudafi and A. Gibali, l₁-l₂ Regularization of split feasibility problems, Numer. Algorithms, (2017), 1-19. doi: 10.1007/s11075-017-0398-6.
[27]	T. L. N. Nguyen and Y. Shin, Deterministic sensing matrices in compressive sensing: A survey, Sci. World J., 2013 (2013), Article ID 192795, 6 pages. doi: 10.1155/2013/192795.
[28]	Z. Opial, Weak convergence of the sequence of successive approximations for nonexpansive mappings, Bull. Am. Math. Soc., 73 (1967), 591-597. doi: 10.1090/S0002-9904-1967-11761-0.
[29]	Y. Shehu and O. S. Iyiola, Accelerated hybrid viscosity and steepest-descent method for proximal split feasibility problems, Optimization, 67 (2018), 475-492. doi: 10.1080/02331934.2017.1405955.
[30]	Y. Shehu and O. S. Iyiola, Convergence analysis for the proximal split feasibility problem using an inertial extrapolation term method, J. Fixed Point Theory Appl., 19 (2017), 2483-2510. doi: 10.1007/s11784-017-0435-z.
[31]	S. Suantai, N. Pholasa and P. Cholamjiak, The modified inertial relaxed CQ algorithm for solving the split feasibility problems, J. Ind. Manag. Optim., (2017). doi: 10.3934/jimo.2018023.
[32]	R. Tibshirani, Regression shrinkage and selection Via the lasso, J. Royal Stat. Soc., 58 (1996), 267-288.
[33]	F. Wang and H.-K. Xu, Cyclic algorithms for split feasibility problems in Hilbert spaces, Nonlinear Anal., 74 (2011), 4105-4111. doi: 10.1016/j.na.2011.03.044.
[34]	F. Wang, Polyak's gradient method for split feasibility problem constrained by level sets, Numer. Algorithms, 77 (2018), 925-938. doi: 10.1007/s11075-017-0347-4.
[35]	H.-K. Xu, Iterative methods for the split feasibility problem in infinite-dimensional Hilbert spaces, Inverse Problems, 26 (2010), 105018, 17 pp. doi: 10.1088/0266-5611/26/10/105018.
[36]	H.-K. Xu, Iterative algorithms for nonlinear operators, J. Lond. Math. Soc., 66 (2002), 240-256. doi: 10.1112/S0024610702003332.
[37]	Z. Xu, X. Chang, F. Xu and H. Zhang, l_1-2 Regularization: A thresholding representation theory and a fast solver, IEEE Trans. Neural Netw. Learn. Syst., 23 (2012), 1013-1027.
[38]	Q. Yang, The relaxed CQ algorithm for solving the split feasibility problem, Inverse Problems, 20 (2004), 1261-1266. doi: 10.1088/0266-5611/20/4/014.
[39]	H. Zhou and P. Wang, Adaptively relaxed algorithms for solving the split feasibility problem with a new step size, J. Inequal. Appl., 2014 (20214), 22pp. doi: 10.1186/1029-242X-2014-448.