Abstract
This paper discusses conditions under which the solution of linear system with minimal Schatten-p norm, 0 < p ⩽ 1, is also the lowest-rank solution of this linear system. To study this problem, an important tool is the restricted isometry constant (RIC). Some papers provided the upper bounds of RIC to guarantee that the nuclear-norm minimization stably recovers a low-rank matrix. For example, Fazel improved the upper bounds to δ A4r < 0.558 and δ A3r < 0.4721, respectively. Recently, the upper bounds of RIC can be improved to δ A2r < 0.307. In fact, by using some methods, the upper bounds of RIC can be improved to δ A2r < 0.4931 and δ A2r < 0.309. In this paper, we focus on the lower bounds of RIC, we show that there exists linear maps A with δ A2r > 1/√2 or δ A r > 1/3 for which nuclear norm recovery fail on some matrix with rank at most r. These results indicate that there is only a little limited room for improving the upper bounds for δ A2r and δ A r . Furthermore, we also discuss the upper bound of restricted isometry constant associated with linear maps A for Schatten p (0 < p < 1) quasi norm minimization problem.
Similar content being viewed by others
References
Argyriou A, Micchelli C, Pontil M. Convex multi-task feature learning. Machine Learning, 2008, 73: 243–272
Beck C, Andrea R. Computational study and comparisons of LFT reducibility methods. In: Proceedings of the American Control Conference. Michigan: American Automatic Control Council, 1998, 1013–1017
Cai J F, Candès E J, Shen Z W. A singular value thresholding algorithm for matrix completion. SIAM J Optim, 2010, 20: 1956–1982
Cai T T, Wang L, Xu G W. Shifting inequality and recovery of sparse signals. IEEE Trans Signal Process, 2010, 58: 1300–1308
Cai T T, Wang L, Xu G W. New bounds for restricted isometry constants. IEEE Trans Inform Theory, 2010, 56: 4388–4394
Candès E J. Compressive sampling. In: Proceedings of International Congress of Mathematicians, vol. 3. Madrid: European Mathematical Society Publishing House, 2006, 1433–1452
Candès E J. The restricted isometry property and its implications for compressed sensing. C R Acad Sci Paris Ser, 2008, 346: 589–592
Candès E J, Plan Y. Tight oracle bounds for low-rank recovery from a minimal number of random measurements. IEEE Trans Inform Theory, 2009, 57: 2342–2359
Candès E J, Recht B. Exact matrix completion via convex optimization. Found Comput Math, 2008, 9: 717–772
Candès E J, Tao T. Decoding by linear programming. IEEE Trans Inform Theory, 2005, 51: 4203–4215
Candès E J, Tao T. The power of convex relaxation: Near-optimal matrix completion. IEEE Trans Inform Theory, 2009, 56: 2053–2080
Chartrand R, Staneva V. Restricted isometry properties and nonconvex compressive sensing. Inverse Problems, 2008, 24: 1–14
Davies M E, Gribonval R. Restricted isometry constants where l p sparse recovery can fail for 0 < p ⩽ 1. IEEE Trans Inform Theory, 2009, 55: 2203–2214
Donoho D. Compressed sensing. IEEE Trans Inform Theory, 2006, 52: 1289–1306
Fazel M, Hindi H, Boyd B. A rank minimization heuristic with application to minimum order system approximation. In: Proceedings of the American Control Conference, vol. 6. Washington: IEEE, 2001, 4734–4739
Foucart S, Lai M J. Sparsest solutions of underdetermined linear systems via lq-minimization for 0 < q ⩽ 1. Appl Comput Harmon Anal, 2009, 26: 395–407
Horn R A, Johnson C R. Topics in Matrix Analysis. New York: Cambridge University Press, 1991
Lin Z C, Chen M M, Wu L Q. The augmented lagrange multiplier method for exact recovery of corrupted low-rank matrices. Technical Report UILU-ENG-09-2215, UIUC, 2009
Ma S, Goldfarb D, Chen L. Fixed point and bregman iterative methods for matrix rank minimization. Math Program, 2011, 128: 321–353
Meka R, Jain P, Dhillon I S. Guaranteed rank minimization via singular value projection. Adv Neural Information Processing Systems, 2011, 23: 937–945
Mirsky L. A trace inequality of John von Neumann. Monatsh Math, 1975, 79: 303–306
Mo Q, Li S. New bounds on the restricted isometry constant δ 2r . Appl Comput Harmon Anal, 2011, 31: 460–468
Mohan K, Fazel M. New restricted isometry results for noisy low-rank recovery. In: IEEE International Symposium on Information Theory Proceedings. Seattle: IEEE, 2010, 1573–1577
Oymak S, Hassibi B. New null space results and recovery thresholds for matrix rank minization. Math Program, in press
Oymak S, Mohan K, Fazel M, et al. A simplified approach to recovery conditions for low rank matrices. In: IEEE International Symposium on Information Theory Proceedings. Petersburg: Information Theory Society, 2011, 2318–2322
Recht B, Fazel M, Parrilo P. Guaranteed minimum-rank solutions of linear matrix equations via nuclear norm minimization. SIAM Rev, 2009, 14: 1–46
Rennie J D, Srebro N. Fast maximum margin matrix factorization for collaborative prediction. In: Proceedings of the International Conference of Machine Learning. New York: Association Computing Machinery, 2005, 713–719
Shen Y, Li S. Restricted p-isometry property and its application for nonconvex compressive sensing. Adv Comput Math, 2012, 37: 441–452
Weinberger K Q, Saul L K. Unsupervised learning of image manifolds by semidefinite programming. Int J Computer Vision, 2006, 70: 77–90
Yuan M, Ekici A, Lu Z, et al. Dimension reduction and coefficient estimation in multivariate linear regression. J Roy Statist Soc Ser B, 2007, 69: 329–346
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Wang, H., Li, S. The bounds of restricted isometry constants for low rank matrices recovery. Sci. China Math. 56, 1117–1127 (2013). https://doi.org/10.1007/s11425-013-4624-y
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11425-013-4624-y
Keywords
- restricted isometry constants
- low-rank matrix recovery
- Schatten-p norm
- nuclear norm
- compressed sensing
- convex optimization