 Research
 Open access
 Published:
A preconditioning method of the CQ algorithm for solving an extended split feasibility problem
Journal of Inequalities and Applications volume 2014, Article number: 163 (2014)
Abstract
In virtue of preconditioning technology, we propose a preconditioning CQ algorithm for an extend split feasibility problem (ESFP). Comparing with the others, the proposed algorithm can get faster convergence without considering to adjust the stepsize. The convergence is also established under mild conditions. Several extensions of the preconditioning CQ algorithm are presented. Moreover, we present an approximate variable preconditioner which does not compute the matrix inverse. Finally, some numerical experiments show the better behaviors of the proposed methods.
MSC:47J10, 47J20, 65B05.
1 Introduction
The problem to find x\in C with Ax\in Q, if such x exists, was called the split feasibility problem (SFP) by Censor and Elfving [1], where C\in {R}^{N} and Q\in {R}^{M} are nonempty closed convex sets, and A is an M by N matrix. This problem plays an important role in the study of signal processing, image reconstruction, and so on [2, 3]. Censor and Elfving’s algorithm in [1], as well as others obtained later [4, 5] involve matrix inverses at each step. Byrne [6] presented a method called the CQ algorithm for solving the SFP that does not involve matrix inverses.
The CQ algorithm Let {x}^{0} be arbitrary. For k=0,1,\dots , calculate
where \gamma \in (0,2/L) and L denotes the largest eigenvalue of the matrix {A}^{T}A, I is the identical matrix. {P}_{C} and {P}_{Q} are the orthogonal projections onto C and Q, respectively.
In recent years, how to modify the CQ algorithm so that it can easily be implemented and converge faster is the hot topic. The typical modifications are as follows: Yang [7] presented a relaxed CQ algorithm for solving the SFP, then the orthogonal projections onto halfspaces {C}_{k} and {Q}_{k} can be executed exactly. Qu and Xiu [8] proposed the modified relaxed algorithm which does not need to compute the largest eigenvalue of the matrix {A}^{T}A and can get an adaptive stepsize by adopting an Armijolike search. The paper [9] extended the algorithm in [10] and proposed a relaxed inexact projection method for the SFP. Xu [11] extended the problem into infinitedimensional Hilbert spaces, and modified the CQ algorithm with Mann’s iteration. In [12], López et al. presented a variable stepsize, and improved the algorithm with a Halperntype iteration.
However, using preconditioning technology to accelerate the CQ algorithm not only has not been taken into account, but also one will obtain a special effect. In this paper, we consider to modify the CQ algorithm from the views of fixed point and variational inequality. Combining with the appropriate preconditioner, the SFP can be transformed into an extended split feasibility problem (ESFP). Naturally, a preconditioning CQ algorithm for solving the ESFP can also solve the SFP indirectly.
The rest of the paper is organized as follows. In Section 2, we review some concepts and existing results. In Section 3, we propose a preconditioning CQ algorithm for solving ESFP and establish its convergence. Several extensions are presented in Section 4. In Section 5, we discuses the methods how to estimate the approximate inverse preconditioner. In Section 6, we report some computational results with the proposed algorithm and methods. Finally, Section 7 gives some concluding remarks.
2 Preliminaries
Our argument mainly depends on monotone operators, nonexpansive mappings, and the metric projections.
Definition 2.1 [13]
Let T be a mapping from a set C\subset {R}^{N} into itself. Then

(i)
T is said to be monotone on C, if
\u3008TxTy,xy\u3009\ge 0,\phantom{\rule{1em}{0ex}}\text{for all}x,y\in C, 
(ii)
a mapping T:C\to C is nonexpansive if
\parallel TxTy\parallel \le \parallel xy\parallel ,\phantom{\rule{1em}{0ex}}\text{for all}x,y\in C.
We denote by Fix(T) the set of fixed points of T; that is, Fix(T)=\{x\in C:Tx=x\}. Note that Fix(T) is always closed and convex (but maybe empty).
The metric projection from {R}^{N} onto C is the mapping {P}_{C}:{R}^{N}\to C, which assigns to each point x\in C the unique point {P}_{C}x\in C satisfying the property
where \parallel \cdot \parallel is the 2norm.
The following properties of projections are useful and pertinent to our purpose.
Lemma 2.1 [13]
Given x\in {R}^{N}
Consequently, {P}_{C} is nonexpansive and monotone, and I{P}_{C} is also nonexpansive, then
Lemma 2.2 For \mathrm{\forall}x,y\in {R}^{N}
Lemma 2.3 [14]
Let U=I\gamma {A}^{T}(I{P}_{Q})A, where \gamma \in (0,2/L).

(i)
U is an averaged operator; i.e. there exist some \beta \in (0,1) and a nonexpansive operator V, U=(1\beta )U+\beta V.

(ii)
Fix(U)={A}^{1}(Q), then Fix({P}_{C}U)=C\cap {A}^{1}(Q).
Proposition 2.1 [15]
For every k\ge 0, let {x}^{k}\in {R}^{N}, {C}_{k} and {Q}_{k} be defined as in [7]. Then for any x\in {R}^{N} and y\in {R}^{M} we have
and
3 The preconditioning CQ algorithm
Stand [16] and Piana and Bertero [17] have applied the preconditioning matrix technologies to improve the Landweber and projected Landweber algorithms. The analyses deal with the operators A and {A}^{\ast}A, which are based on the singular value decomposition and a more general spectrum, respectively. We can also extend the technologies to improve the CQ algorithm.
As the SFP is to find a point {x}^{\ast}\in C, with A{x}^{\ast}\in Q. Firstly, we set \mathrm{\Omega}=C\cap {A}^{1}(Q), {A}^{1}(Q)=\{{x}^{\ast}\in {R}^{N}A{x}^{\ast}\in Q\}. From Lemma 2.3, (1) can be depicted from the view of fixed point,
Assume that \mathrm{\Omega}\ne \mathrm{\varnothing}, i.e. the SFP has a nonempty solution set, and {x}^{\ast} is the solution of SFP. Thus, we have
so
Then let D:C\to C be a N\times N symmetrical positive definite matrix, and (AD){x}^{\ast}\in Q. Referring to (9), we can deduce that
or
Then {U}_{D} also has the same properties of U in Lemma 2.3, and we can obtain
Now we present a new algorithm, which is named a preconditioning CQ algorithm (PCQ).
Algorithm 3.1 Let D:C\to C be a N\times N symmetrical positive definite matrix, {x}^{0}\in C be arbitrary. For k=0,1,\dots , calculate
where \gamma \in (0,2/L), L={\parallel D{A}^{T}\parallel}^{2}.
Algorithm 3.1 is to solve an extended SFP (ESFP), which can be represented as follows.
Definition 3.1 Let C and Q be nonempty closed convex sets in {R}^{N} and {R}^{M}, respectively, and A is an M by N matrix, D is an N by N symmetrical positive definite matrix, the ESFP is to find x\in C with ADx\in Q. We denote the solution set of ESFP by G.
Remark 3.1 If we set \tilde{x}=Dx\in D(C)=\tilde{C}, then A\tilde{x}\in Q, the problem in Definition 3.1 is transformed into SFP.
Remark 3.2 If we set D is an unit matrix, then to find x\in C with Ax\in Q, the problem in Definition 3.1 is transformed into SFP.
From Remark 3.1 we know that the SFP is to minimize the equation
Substituting \tilde{x}=Dx into (13), its gradient operator is
While C={R}^{N} and C\cap {(AD)}^{1}(Q)\ne \mathrm{\varnothing}, we also have
where {x}^{\ast}\in {R}^{N} is the solution set of the extended SFP. We can obtain the following variational inequality:
Therefore, we have the next constrained leastsquares problem:
The following immediately follows.
Theorem 3.1 Assume G\ne \mathrm{\varnothing}, then {x}^{\ast}\in G, if and only if {x}^{\ast}=argmin\{f(x)x\in C\}, if and only if \u3008\mathrm{\nabla}f({x}^{\ast}),x{x}^{\ast}\u3009\ge 0, \mathrm{\forall}x\in C, where
As {U}_{D}=I\gamma D{A}^{T}(I{P}_{Q})AD, from (12) we have
In order to establish the convergence of Algorithm 3.1, we need the following theorem.
Theorem 3.2 Assume that G=C\cap {(AD)}^{1}(Q)\ne \mathrm{\varnothing}, then

(i)
Fix({U}_{D})={(AD)}^{1}(Q)=\{x\in {R}^{N}ADx\in Q\};

(ii)
Fix({P}_{C}{U}_{D})=G.
Proof As {D}^{T}=D, {(AD)}^{T}=D{A}^{T}, we have {U}_{D}=I\gamma {(AD)}^{T}(I{P}_{Q})(AD).
Firstly, we prove {(AD)}^{1}(Q)\subset Fix({U}_{D}). For \mathrm{\forall}x\in {(AD)}^{1}(Q), then x\in {R}^{N} and ADx\in Q, we have {P}_{Q}ADx=ADx. So
Therefore, x\in Fix({U}_{D}).
Secondly, we prove Fix({U}_{D})\subset {(AD)}^{1}(Q).
As G=C\cap {(AD)}^{1}(Q)\ne \mathrm{\varnothing}, we choose z\in G, then z\in C and z\in {(AD)}^{1}(Q), so (AD)z\in Q.
For \mathrm{\forall}x\in Fix({U}_{D}), we have {A}^{T}(I{P}_{Q})ADx=0. From the properties of a projection, we can deduce
therefore,
then (I{P}_{Q})ADx=0, and ADx={P}_{Q}(AD)x\in Q.
We obtain x\in {(AD)}^{1}(Q), thus (i) is proved.
We can also deduce that Fix({P}_{C}{U}_{D})=Fix({P}_{C})\cap Fix({U}_{D})\triangleq G=C\cap {(AD)}^{1}(Q)\ne \mathrm{\varnothing}. □
Theorem 3.3 Assume G\ne \mathrm{\varnothing}, 0<\gamma <2/L, L={\parallel D{A}^{T}\parallel}^{2}, the sequence \{{x}^{k}\} is generated by (15), there exists {lim}_{k\to \mathrm{\infty}}{x}^{k}\to {x}^{\ast}\in G.
Proof Firstly, we show that if \gamma =2/L, the operator
is nonexpansive.
For \mathrm{\forall}x,y\in C, from (4) and (6) we have
therefore,
Next, we can easily obtain 0<\gamma L/2<1, and we set
From (16) we deduce that
and V is nonexpansive, hence, while 0<\gamma <2/L, {U}_{D} is an averaged nonexpansive operator.
Finally, we choose \mathrm{\forall}p\in G, where p\in C and p={U}_{D}p. We have p={P}_{C}{U}_{D}p={P}_{C}p. From (7), we have
which implies that \{{\parallel {x}^{k}p\parallel}^{2}\} is monotonically decreasing and hence {lim}_{k\to \mathrm{\infty}}{\parallel p{x}^{k}\parallel}^{2}=d\ge 0. Specially, \{{x}^{k}\} is bounded.
From (19) we can deduce that
as {\parallel {x}^{k}p\parallel}^{2}{\parallel {x}^{k+1}p\parallel}^{2}\to 0, we have {x}^{k}V{x}^{k}\to 0.
Let {x}^{\ast} be an arbitrary cluster point of the sequence \{{x}^{k}\}. Then there exists a subsequence \{{x}^{{k}_{j}}\}\subset \{{x}^{k}\}, then {x}^{{k}_{j}}\to {x}^{\ast} (j\to \mathrm{\infty}). As \{{x}^{k}\}\subset C, {x}^{\ast}\in C, and {x}^{\ast}={P}_{C}{x}^{\ast}. Because V is nonexpansive and continuous, then V{x}^{{k}_{j}}\to V{x}^{\ast} (j\to \mathrm{\infty}).
As \parallel {x}^{\ast}V{x}^{\ast}\parallel \le \parallel {x}^{\ast}{x}^{{k}_{j}}\parallel +\parallel {x}^{{k}_{j}}V{x}^{{k}_{j}}\parallel +\parallel V{x}^{{k}_{j}}V{x}^{\ast}\parallel \to 0, we have {x}^{\ast}=V{x}^{\ast}, then {U}_{D}{x}^{\ast}=(1\beta ){x}^{\ast}+\beta V{x}^{\ast}={x}^{\ast}. Therefore, {x}^{\ast}={P}_{C}{x}^{\ast}={P}_{C}{U}_{D}{x}^{\ast}, from Theorem 3.2, we have {x}^{\ast}\in G. However, {lim}_{n\to \mathrm{\infty}}\parallel {x}^{k}{x}^{\ast}\parallel =d\ge 0 exists, and there exists a subsequence \{{x}^{{k}_{j}}\} of \{{x}^{k}\} s.t. {x}^{{k}_{j}}\to {x}^{\ast} (j\to \mathrm{\infty}), therefore, there must be {x}^{k}\to {x}^{\ast} (j\to \mathrm{\infty}). □
4 Several extensions of the preconditioning CQ algorithm
In virtue of kinds of CQlike algorithms for solving the SFP, we can also deduce the following meaningful results for solving the ESFP without proof.
According to the relaxed CQ algorithm [7], we firstly obtain the relaxed projection method.
Algorithm 4.1 Let D:{C}_{k}\to {C}_{k} be a N\times N symmetrical positive definite matrix, {x}^{0} be arbitrary. For k=0,1,\dots , calculate
where \gamma \in (0,2/L), L={\parallel D{A}^{T}\parallel}^{2}.
Theorem 4.1 Let \{{x}^{k}\} be a sequence generated by the relaxed preconditioning CQ algorithm. Then \{{x}^{k}\} converges to a solution of ESFP.
Next, from the papers [8] and [14], define \mathrm{\nabla}{f}_{k}:{R}^{N}\to {R}^{N} by
and we can obtain an adaptive algorithm with strong convergence.
Algorithm 4.2 Let D:{C}_{k}\to {C}_{k} be a N\times N symmetrical positive definite matrix, given constants \lambda >0, l\in (0,1), \mu \in (0,1). Let {x}^{0} be arbitrary, for k=0,1,\dots , let
where {\rho}_{k}=\lambda {l}^{{m}_{k}} and {m}_{k} is the smallest nonnegative integer m such that
Set
where \{{\alpha}_{k}\} is a real sequence in (0,1) that satisfies conditions (C_{1}) {lim}_{k\to \mathrm{\infty}}{\alpha}_{k}=0; and (C_{2}) {\sum}_{k=1}^{\mathrm{\infty}}{\alpha}_{k}=\mathrm{\infty}.
Lemma 4.1 For all k=0,1,\dots , \mathrm{\nabla}{f}_{k} is Lipschitz continuous on {R}^{N} with constant L and cocoercive on {R}^{N} with modulus 1/L, where L is the largest eigenvalue of the matrix {A}^{T}A. Therefore, the Armijolike search rule (22) is well defined.
Lemma 4.2 For all k=0,1,\dots , \frac{\mu l}{L}<{\rho}_{k}\le \gamma.
Theorem 4.2 Let \{{x}^{k}\} be a sequence generated by Algorithm 4.2. If the solution set of the SFP is nonempty, then \{{x}^{k}\} converges strongly to a solution of the ESFP.
As there exists an Armijolike search step in Algorithm 4.2, the complexity of the implementation will be increased. Next, we propose a new variable stepsize to improve Algorithm 3.1.
Algorithm 4.3 Let {D}^{k}:C\to C be a variable N\times N symmetrical positive definite matrix, k=0,1,\dots . For \mathrm{\forall}{x}^{0}\in C, calculate
where {\gamma}^{k}\in (0,2/(L\ast {M}_{D})), L is the largest eigenvalue of {A}^{T}A, {M}_{D} is the minimum value of all the largest eigenvalues of {D}^{k}, for k=0,1,\dots . Specially, set {\gamma}^{k}=1/(L\ast {M}_{D}).
5 Approximating a variable preconditioner
In the above algorithms, the preconditioner D is continuous, positive definite and bounded so that it has a continuous inverse [17]. According to the preconditioning CQ algorithm, we set D commutes with the operator {A}^{T}A. Therefore, we set a matrix function F with positive value, and its dimension should be consistent with dim({A}^{T}A). Moreover, F should have a positive lower bound in order to satisfy the existing inverse. Strand [16] also assumes that F is a polynomial or a rational function. Thus, the operator D is given by
Assume {({A}^{T}A)}^{1} is existed, the product D{A}^{T}AD is without restrictions in (10), we can deduced that sometimes D should be chose close to {({A}^{T}A)}^{1/2} as more as possible.
The best condition is to calculate {({A}^{T}A)}^{1} exactly, but it is always hard with the signal and image reconstruction problems. If F be a polynomial function, Stand [16] have provided an example with seventhorder, and Neumann’s series approach can also express {({A}^{T}A)}^{1}. However, the polynomial method needs to calculate the high order matrix multiplication, therefore, it can not be implemented easily.
If we choose F be a rational function, a simple example, closely related to the Tikhonov regularization method has been used in [17]. According to the example, the approximate inverse preconditioner D can be given by
where Re denotes the real part, α is a positive real parameter, and good choices of α may be much smaller than the values provided by the methods used for estimating optimum values of the Tikhonov regularization parameter.
As (26) involves the matrix inverse, we next propose a diagonal format of D that does not calculate matrix inverses. Furthermore, the choice of D is related to the convergence properties of the algorithm. If D is evolutive following the iterations, the convergence rate of algorithm will also be accelerated.
From (10), we can deduce that
As D should be chosen closely to {({A}^{T}A)}^{1/2}, we assume λ is the approximate eigenvalue matrix of {A}^{T}A, and then we set
Therefore, we can obtain the approximate variable preconditioners with respect to {({A}^{T}A)}^{1} on the (k+1)th iteration:
Then we can also get the approximate variable preconditioners with respect to {({A}^{T}A)}^{1/2} on the (k+1)th iteration:
Otherwise, a variable stepsize in Algorithm 4.3 can be estimated. Set {L}_{{D}^{k}} is the largest eigenvalue of {D}^{k}, k=0,1,\dots , on the k th iteration, we have {M}_{{D}^{k}}=min\{{L}_{{D}^{n}}n=0,1,\dots ,k\}. Then set {l}_{{\overline{D}}^{k}}, the minimum eigenvalue of {\overline{D}}^{k}, k=0,1,\dots , on the k th iteration, we have {L}_{k}=max\{1/{l}_{{\overline{D}}^{n}}n=0,1,\dots ,k\}. Therefore, a variable stepsize with respect to Algorithm 4.3 can be approximated by
6 Numerical results
In this section, we present some numerical results for the proposed method. The following three examples are taken from the test problems in [15]. For Examples 6.2 and 6.3, we should first transform into the ESFP. The stopping criterion is \parallel {x}^{k+1}{x}^{k}\parallel <\epsilon, and we took \epsilon ={10}^{10}, \gamma =1/{\parallel D{A}^{T}\parallel}^{2}, \alpha =0.1. The projections are computed by Proposition 2.1.
Algorithm 4.1 was implemented in the Matlab R2011b (Windows version) programming environment. The codes were ran on a PC with 1.98 GB memory and Intel(R) Pentium(R) dualcore CPU G630 running at 2.69 GHz. The iteration numbers and the computational time for the methods in Section 5 with different starting points are given in Tables 1, 2, and 3, all CPU times reported are in seconds.
Example 6.1 (A convex feasibility problem, CFP)
Let C=\{x\in {R}^{3}{x}_{2}^{2}{x}_{3}^{2}4\le 0\}, Q=\{y\in {R}^{3}{y}_{3}1{y}_{1}^{2}\le 0\}. Find some point x in C\cap Q.
Example 6.2 (A split feasibility problem)
Let A=\left(\begin{array}{ccc}2& 1& 3\\ 4& 2& 5\\ 2& 0& 2\end{array}\right), C=\{x\in {R}^{3}{x}_{1}+{x}_{2}^{2}+2{x}_{3}\le 0\}, Q=\{y\in {R}^{3}{y}_{1}^{2}+{y}_{2}{y}_{3}\le 0\}. Find x\in C with Ax\in Q.
Example 6.3 (A split feasibility problem)
Let A=\left(\begin{array}{ccc}1& 1& 0\\ 0& 1& 1\\ 2& 0& 1\end{array}\right), C=\{x\in {R}^{3}{x}_{1}+{x}_{2}^{2}+2{x}_{3}\le 0\}, Q=\{y\in {R}^{3}{y}_{1}^{2}+{y}_{2}{y}_{3}\le 0\}. Find x\in C with Ax\in Q.
Compared with the results in [15], we can obtain:

(1)
For the CFP, as A=I, the relaxed preconditioning method played a role inconspicuously.

(2)
For the SFP, the results are better than the ones in [15], and when A is not sparse the effect is obvious.
7 Conclusions
In this paper, by adopting the preconditioning techniques, a modified CQ algorithm is named preconditioning CQ algorithm, and its extensions for solving the ESFP have been presented. The approximate methods for how to estimate the preconditioner D are also discussed; the approximate diagonal preconditioner method does not need to compute the matrix inverses and the largest eigenvalue of the matrix {A}^{T}A. Thus, the algorithm can be implemented easily. Moreover, the corresponding convergence property has been established in the feasible case of ESFP. The numerical results showed that the proposed algorithms and methods are effective to solve some problems.
References
Censor Y, Elfving T: A multiprojection algorithm using Bregman projections in a product space. Numer. Algorithms 1994, 8: 221239. 10.1007/BF02142692
Bauschke HH, Borwein JM: On projection algorithms for solving convex feasibility problems. SIAM Rev. 1996, 38: 367426. 10.1137/S0036144593251710
Byrne CL: A unified treatment of some iterative algorithms in signal processing and image reconstruction. Inverse Probl. 2004, 20: 103120. 10.1088/02665611/20/1/006
Byrne CL: Iterative projection onto convex sets using multiple Bregman distances. Inverse Probl. 1999, 15: 12951313. 10.1088/02665611/15/5/313
Byrne CL: BregmanLegendre multidistances projection algorithm for convex feasibility and optimization. In Inherently Parallel Algorithms in Feasibility and Optimization and Their Applications. Edited by: Butnariu D, Censor Y, Reich S. Elsevier, Amsterdam; 2001:87100.
Byrne CL: Iterative oblique projection onto convex sets and the split feasibility problem. Inverse Probl. 2002, 18: 441453. 10.1088/02665611/18/2/310
Yang QZ: The relaxed CQ algorithm solving the split feasibility problem. Inverse Probl. 2004, 20: 12611266. 10.1088/02665611/20/4/014
Qu B, Xiu N: A note on the CQ algorithm for the split feasibility problem. Inverse Probl. 2005, 21: 16551665. 10.1088/02665611/21/5/009
Wang Z, Yang Q: The relaxed inexact projection methods for the split feasibility problem. Appl. Math. Comput. 2011, 217: 53475359. 10.1016/j.amc.2010.11.058
Yang Q, Zhao J: The projectiontype methods for solving the split feasibility problem. Math. Numer. Sin. 2006, 28: 121132. (in Chinese)
Xu HK: Iterative methods for the split feasibility problem in infinitedimensional Hilbert spaces. Inverse Probl. 2010, 26: 117.
López G, MartínMárquez V, Wang F, Xu HK: Solving the split feasibility problem without prior knowledge of matrix norms. Inverse Probl. 2012, 28: 118.
Zarantonello EH: Projections on convex sets in Hilbert space and spectral theory. In Contributions to Nonlinear Functional Analysis. Edited by: Zarantonello EH. Academic Press, New York; 1971.
Wang F, Xu HK: Approximating curve and strong convergence of the CQ algorithm for the split feasibility problem. J. Inequal. Appl. 2010., 2010: Article ID 102085
Abdellah B, Muhammad AN: On descentprojection method for solving the split feasibility problems. J. Glob. Optim. 2012, 54: 627639. 10.1007/s1089801197822
Strand ON: Theory and methods related to the singularfunction expansion and Landweber’s iteration for integral equations of the first kind. SIAM J. Numer. Anal. 1974, 11: 798825. 10.1137/0711066
Piana M, Bertero M: Projected Landweber method and preconditioning. Inverse Probl. 1997, 13: 441463. 10.1088/02665611/13/2/016
Acknowledgements
The authors would like to thank the associate editor and the referees for their comments and suggestions. This research was supported by the National Natural Science Foundation of China (11071053).
Author information
Authors and Affiliations
Corresponding author
Additional information
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
Both authors take equal roles in deriving results and writing of this paper. Both authors read and approved the final manuscript.
Rights and permissions
Open Access This article is distributed under the terms of the Creative Commons Attribution 2.0 International License (https://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
About this article
Cite this article
Wang, P., Zhou, H. A preconditioning method of the CQ algorithm for solving an extended split feasibility problem. J Inequal Appl 2014, 163 (2014). https://doi.org/10.1186/1029242X2014163
Received:
Accepted:
Published:
DOI: https://doi.org/10.1186/1029242X2014163