Oscillation and nonoscillation for certain class of first and second order generalized \(\alpha\)-difference equations (Q2914841)

In this review we use the following standard notations: \([x]\) and \(\{x\}\) denotes integral and fractional parts of \(x\in{\mathbb R}\), respectively. So \(x=[x]+\{x\}\). To distinguish the authors' numbering of formulas from the reviewer's numbering, we place the authors' numberings on the right part of the page and those of the reviewer on the left.NEWLINENEWLINELet \(\ell>0\). Let us consider the set \([a,+\infty).\) The projections NEWLINE\[NEWLINE\pi_{a,\ell,\rho}(x)=\rho=\rho_{a,\ell}(x)=\ell\{(x-a)/\ell\}: [a,+\infty)\to[a,\ell)NEWLINE\]NEWLINE and NEWLINE\[NEWLINE\pi_{a,\ell,\nu}(x)=\nu=\nu_{a,\ell}(x)=[(x-a)/\ell]:[a,+\infty)\to{\mathbb N}_0=[0,+\infty)\cap {\mathbb Z}NEWLINE\]NEWLINE represent the set \([a,+\infty)\) as the direct product \([0,+\ell)\times{\mathbb N}_0\). Therefore, if a function \(v(x)\) is defined on \([a,+\infty)\), then \(v(x)\) can be uniquely represented as a function \(v^\vee_{a,\ell}(\rho,\nu)\) of two independent variables \(\rho=\rho_{a,\ell}(x)=\ell\{(x-a)/\ell\}\in[0,\ell)\) and \(\nu=\nu_{a,\ell}(x)=[(x-a)/\ell]\in{\mathbb N}_0\) as follows: \(v^\vee_{a,\ell}(\rho,\nu)=v(a+\rho+\ell\nu)\).NEWLINENEWLINEThe authors consider the operator \(\Delta_\ell\) on the space of all the real valued functions \(v(x)\) defined on \([a,+\infty)\), which acts as follows: NEWLINE\[NEWLINE(\Delta_\ell v)(x)=v(x+\ell)-v(x).NEWLINE\]NEWLINE This operator \(\Delta_\ell\) is connected with the standard operator \(\Delta\): NEWLINE\[NEWLINE(\Delta\phi)(\rho,\nu)=\phi(\rho,\nu+1)-\phi(\rho,\nu)NEWLINE\]NEWLINE for any function \(\phi(\rho,\nu)\) defined on \([0,\ell)\times{\mathbb N}_0\). More precisely, this connection can be described as follows: NEWLINE\[NEWLINE(\Delta_\ell v)(x)=(\Delta v^\vee_{a,\ell})(\rho,\nu)= v^\vee_{a,\ell}(\rho,\nu+1)-v^\vee_{a,\ell}(\rho,\nu).NEWLINE\]NEWLINE Let \(\alpha>0\). The authors consider also the operator \(\Delta_{\alpha,\ell}\) defined on the space of all real valued functions \(v(x)\) on \([a,+\infty)\) acting as NEWLINE\[NEWLINE(\Delta_{\alpha,\ell} v)(x)=v(x+\ell)-\alpha v(x).NEWLINE\]NEWLINENEWLINENEWLINEReviewer's remark 1. The authors denote this operator \(\Delta_{\alpha,\ell}\) as \(\Delta_{\alpha(\ell)}\). In my opinion, such a notation is not very good, because it produces the illusion that \(\alpha\) depends of \(\ell\). But \(\alpha\) and \(\ell\) are independent.NEWLINENEWLINEThis operator \(\Delta_{\alpha,\ell}\) is also connected with the standard operator \(\Delta\). To describe this connection it is necessary to introduce the functions NEWLINE\[NEWLINEw(x)=v^\wedge_{a,\alpha,\ell}(x)= \alpha^{-\nu_{a,\ell}(x)}v(x)=\alpha^{-[(x-a)/\ell]}v(x).NEWLINE\]NEWLINE Then NEWLINE\[NEWLINE(\Delta_\ell w){x}=\alpha^{-[(x-a)/\ell]-1}(\Delta_{\alpha,\ell}v)(x).NEWLINE\]NEWLINE Clearly, NEWLINE\[NEWLINE\begin{multlined} w^\vee_{a,\ell}(\rho,\nu)=w(a+\rho+\ell\nu)=\alpha^{-[(\rho+\ell\nu)/\ell]}v(a+\rho+\ell\nu) \underset{\rho\in[0,\ell)} =\alpha^{-\nu}v(a+\rho+\ell\nu),\end{multlined}NEWLINE\]NEWLINE and NEWLINE\[NEWLINE(\Delta w^\vee_{a,\ell})(\rho,\nu) =\alpha^{-(\nu+1)} (v(a+\rho+\ell(\nu+1))-\alpha v(a+\rho+\ell\nu))=\alpha^{-(\nu+1)}\Delta_{\alpha,\ell}v(x).NEWLINE\]NEWLINE The authors consider the difference equations NEWLINE\[NEWLINE \Delta_{\alpha,\ell}u(x)+\delta\sum_{i=1}^m f_i(x)F_i(u(g_i(x))),\, x\in[0,+\infty),\,\delta=\pm1,\eqno{(0.1)} NEWLINE\]NEWLINE and NEWLINE\[NEWLINE \Delta^2_{\alpha,\ell}u(x)=f(x,u(x),\Delta_{\beta,\ell}u(x)),\eqno{(0.2)} NEWLINE\]NEWLINE where \(x\in[0,+\infty),\) \(\alpha\), \(\beta\) and \(\ell\) are positive real fixed constants, \(f_i\) and \(F_i\) are defined on \(\mathbb R\) for each \(i\in\{1,\dots ,m\}\), \(g_i(x)\geq0\) and \(f\) is defined on \([0,+\infty)\times{\mathbb R}^2\).NEWLINENEWLINEFurther, the authors suppose that the following conditions are fulfilled: {\parindent=6mm \begin{itemize}\item[(i)] \(f_i(x)\geq0\) for all \(x\in[0,+\infty),\,i\in\{1,\dots ,m\} \) \item[(ii)] \(\lim_{x\to+\infty}g_i(x)=+\infty\), and \(uF_i(u)>0,\) if \(u\neq0\), for all \(i\in\{1,\dots ,m\}\). NEWLINENEWLINE\end{itemize}} The authors denote \({\mathbb N}_{\ell}(\rho)=\rho+\ell{\mathbb N}_{0},\) where \(\rho\in[0,\ell)\).NEWLINENEWLINEDefinition 2.6. The solution \(u(k)\) of (0.1) or (0.2) is called oscillatory by \(\ell\) steps if for any \(x_1\in[a,+\infty)\) there exists \(x_2\in[x_1,+\infty)\) such that \(u(x_2)u(x_2+\ell)\leq0\). The difference equation itself is called oscillatory by \(\ell\) steps if all its solutions are oscillatory by \(\ell\) steps. If the solution \(u(x)\) is not oscillatory by \(\ell\) steps then it is said to be nonoscillatory by \(\ell\) steps (i.e. there exists \(x_1\in[0,+\infty)\) such that \(u(x)u(x+\ell)>0\) for all \(x\in [x_1,+\infty)\)).NEWLINENEWLINEReviewer's remark 2. If \(f_i=0\), \(F_i(0)=0\) for each \(i\in\{1,\,\dots,\,m\}\) and \(f=0\), then \(u(x)=0\) is the solution of the equations (0.1) and (0.2). According to Definition 2.6 this solution is oscillatory by \(\ell\) steps. Let \(\ell=\sqrt{2}\). Then the expression ``oscillatory by \(\sqrt{2}\)-steps'' does not sound very natural. Instead I would suggest to use the expressions \(\ell\)-oscillatory and \(\ell\)-nonoscillatory. Furthermore, the authors do not use these notions in the article but instead the notions ``oscillatory solution'', ``oscillatory equation'', ``nonoscillatory solution'' and ``nonoscillatory equation''.NEWLINENEWLINEReviewer's remark 3. Let \(u(x)\) be a solution of the difference equation (0.1) with \(\ell=1\), and let \(w(x)=\alpha^{-x}u(x)\). Then we come to the equations NEWLINE\[NEWLINE\Delta_{\alpha,1}u(x)=u(x+1)-\alpha u(x) =\alpha^{x+1}\Delta w(x)= \sum_{i=1}^m f_i(x)F_i(\alpha^{g_i(x)}w(g_i(x)))NEWLINE\]NEWLINE and NEWLINE\[NEWLINE\Delta w(x)=\sum_{i=1}^m f^\vee_i(x)F(\alpha^{g_i(x)}w(g_i(x))) \tag{1}NEWLINE\]NEWLINE with \(f^\vee_i(x)=\alpha^{-x-1}f_i(x)\).NEWLINENEWLINESo, in the case \(\ell=1\), one can study the difference equation (1) instead of the difference equation (0.1).NEWLINENEWLINEFirst, the authors present the following theorems about the behavior of the solutions of the equations (0.1) in the case \(\delta=1\).NEWLINENEWLINETheorem 3.1. Let \(\alpha\geq1\) and let there exist an index \(i\in\{1,\,\dots,\,m\}\) such that \(| F_i(u)|\) is bounded away from zero if \(| u|\) is bounded away from zero and for some \(x_2\in [0,\,\infty)\) the equality NEWLINE\[NEWLINE \sum_{r=0}^\infty\alpha^{-[(x_2+\rho)/\ell]-r-1}f_i(x_2+\rho+r\ell)= +\infty \eqno{(3.1)} NEWLINE\]NEWLINE holds for all \(\rho\in[0,\ell)\). Then every solution \(u(x)\) of (0.1) is either oscillatory or \(u(x)=o(\alpha^{[x/\ell]})\).NEWLINENEWLINETheorem 3.3. If \(\alpha\in(0,1)\) and the conditions of Theorem 3.1 hold then every solution of (0.1) is oscillatory or \(u(x)=o(1/x)\).NEWLINENEWLINETheorem 3.5. Let \(\alpha\geq1\) and let there exist an index \(i\in\{1,\,\dots,\,m\}\) such that \(F_i(u)\) is nonincreasing on \({\mathbb R}\{0\}\) and NEWLINE\[NEWLINE \sum_{r=0}^\infty\alpha^{-[(x_2+\rho)/\ell]-r-1} F_i(c\alpha^{[(x_2+\rho)/\ell]+r+1})=\pm\infty \eqno{(3.3)} NEWLINE\]NEWLINE holds for all \(\rho\in[0,\ell)\). Then every solution \(u(x)\) of (0,1) is oscillatory.NEWLINENEWLINETheorem 3.7. If \(0<\alpha<1\) and for some index \(i\in\{1,\,\dots,\,m\}\) (3.1) holds and NEWLINE\[NEWLINE\liminf_{u\to 0+} F_i(u)=a>0, ~ \limsup_{u\to 0-} F_i(u)=b<0,\eqno{(3.4)}NEWLINE\]NEWLINE then every solution of the equation (0.1) is oscillatory.NEWLINENEWLINETheorem 3.9. Let there exist an index \(i\in\{1,\,\dots,\,m\}\) and a positive constant \(L\) such that \(| F_i(u)|\geq L| u|\) for \(u\in{\mathbb R}\), the set \(R_i=\{x\in[0,+\infty):g_i(x)\leq x\}\) is an infinite set, NEWLINE\[NEWLINELf_i(x)\alpha^{[(g_i(x)-x-\ell)/\ell]+1}\geq1NEWLINE\]NEWLINE for all \(x\in R_i\). Then the difference equation (0.1) is oscillatory.NEWLINENEWLINETheorem 3.11. Let \(\alpha\geq1\), there exists an index \(i\in\{1,\,\dots,\,m\}\) such that \(F_i(u)\) is nondecreasing NEWLINE\[NEWLINE\int_0^\beta\frac{dt}{F_i(t)}<+\infty\,\,\text{and} \,\,-\int_{-\beta}^0\frac{dt}{F_i(t)}<+\infty ,\eqno{(3.4)}NEWLINE\]NEWLINE NEWLINE\[NEWLINE\sum_{r=0}^\infty \chi_i(\rho+\nu\ell) \alpha^{-[(\rho+r\ell)/\ell]-1}f_i(\rho+r\ell)= +\inftyNEWLINE\]NEWLINE for all \(\rho\in[0,\ell)\), where \(\chi_i(x)\) denotes characteristic function of the set \(R_i\) defined in Theorem 3.9. Then the difference equation is oscillatory.NEWLINENEWLINETheorem 3.13. Let \(\rho=x-[x/\ell]\ell\) for \(x\in[x_1,\infty)\). If \(0<\alpha\leq1\) and there exists \(i\in\{1,\,\dots,\,m\}\) such that \(F_i(u)\) is nonincreasing on \({\mathbb R}\diagdown\{0\}\), NEWLINE\[NEWLINE\lim_{x\to+\infty} \sum_{r=0}^{(x-\rho)/\ell}\alpha^{[(x-\rho)/\ell]+1}f(x-\rho-rl)=+\infty, NEWLINE\]NEWLINE then every bounded solution of (0.1) is oscillatory.NEWLINENEWLINEFurther, the authors present the following theorems related to the case \(\delta=-1\).NEWLINENEWLINETheorem 3.15. Let \(0<\alpha\leq1\) and let \(F_i(u)\) is nondecreasing on \({\mathbb R}\diagdown\{0\}\). If \(\rho=x-[x/\ell]\ell\) for \(x\in[x_2,\infty)\) for \(x_2\in[0,+\infty)\) and for any \(c\neq0\) NEWLINE\[NEWLINE\lim_{x\to+\infty} \sum_{r=0}^{(x-\rho)/\ell}\alpha^{[(x-x_2-\rho-r\ell)/\ell]+1} f(x_2+\rho+r\ell)F_i(c\alpha^{[(x-x_2-\rho-r\ell)/\ell]+1}=\pm\infty \eqno{(3.8)}NEWLINE\]NEWLINE then every solution of the (0.1) is oscillatory or \(| u(x)|\to+\infty\) as \(x\to+\infty\).NEWLINENEWLINETheorem 3.17. Let \(R^\vee_i=\{x\in[0,+\infty)::g_(x)>x+\ell\}\). Suppose that \(\alpha\geq1\) and there exists \(i\in\{1,\,\dots,\,m\}\) such that \(F_i(u)\) is nondecreasing on \({\mathbb R}\diagdown\{0\}\). NEWLINE\[NEWLINE\int_\beta^{+\infty}\frac{dt}{F_i(t)}<+\infty\,\,\text{and} \,\,-\int_{-\infty}^{-\beta}\frac{dt}{F_i(t)}<+\infty \quad\text{for any}\quad b>0, \eqno{(3.10)}NEWLINE\]NEWLINE NEWLINE\[NEWLINE\sum_{r=0}^\infty \chi^\vee_i(x+r\ell) \alpha^{-[(\rho+r\ell)/\ell]-1}f_i(x+r\ell)= +\infty,NEWLINE\]NEWLINE where \(\chi_i^\vee\) is the characteristic function of \(R_i^\vee\). Then all the solutions of (0.1) are oscillatory.NEWLINENEWLINEReviewer's remark 4: Let me offer a counterexample to ``Theorem 3.1''. Let NEWLINE\[NEWLINE\alpha=\ell=m=1,~F_1(u)=u,~f_1(x)=1,~g_1(x)=x+3/2.NEWLINE\]NEWLINE Clearly, \(g_1(x)>0\) for all the \(x\in[0,+\infty)\) and \(\lim_{x\to+\infty}g_1(x)=+\infty\). Clearly, the polynomial \(z^3+z^2-1\) has the single zero \(\lambda\in(0,1)\). Let \(u(x)=u_1(x)=1/(1-\{2x\})\lambda^{[2x]}\). Clearly, \(u(x)>0\) for all the \(x\in[0,+\infty)\). Clearly, NEWLINE\[NEWLINEu_1(x+1)-u_1(x)=-u_1(x+3/2)=-f_1(x)F_1(u(g_1(x))). \tag{2}NEWLINE\]NEWLINE Clearly, the equality NEWLINE\[NEWLINE\sum_{r=0}^\infty1^{-[(x_2+\rho)/\ell]-r-1}f_i(x_2+\rho+r\ell)= +\infty \eqno{(3.1)}NEWLINE\]NEWLINE holds for all \(\rho\in[0,\ell)\). So, \(u(x)\) is a nonoscillatory solution of the equation (2) and \(u_1(x)\) is not \(o(1^{x})=o(1)\), because the equality \(u_1(x)=o(1)\) implies the equality \(u_1(x)=O(1)\), but the constructed \(u_1(x)\) is unbounded on \([n,n+1)\) for any \(n\in{\mathbb N}_0\).NEWLINENEWLINEWe can offer a second counterexample to ``Theorem 3.3'' now. Let \(\alpha\in(0,1)\) and \(x_2\), \(u_1(x)\), \(g_1(x)\), and \(F_1(u)\) are defined as in the first counterexample. Let, finally, \(f_1(x)=1/\sqrt{\alpha}\) and \(u(x)=u_2(x)=\alpha^xu_1(x)\). NEWLINE\[NEWLINE\Delta_{\alpha,1}u_2(x)=\alpha^{x+1}\Delta u_1(x)=NEWLINE\]NEWLINE NEWLINE\[NEWLINE-\alpha^{x+1} u_1(x+3/2)=-\alpha^{x+1}\alpha^{-x-3/2}u_2(x+3/2)=NEWLINE\]NEWLINE NEWLINE\[NEWLINE-f_1(x)u_2(x+3/2)=-f_1(x)F_1(u_2(g_1(x))),NEWLINE\]NEWLINE and the equality NEWLINE\[NEWLINE \sum_{r=0}^\infty\alpha^{-[(x_2+\rho)/\ell]-r-1}f_i(x_2+\rho+r\ell)= +\infty \eqno{(3.1)} NEWLINE\]NEWLINE holds. So \(u_2(x)\) is a nonoscillatory solution of the equation (0.1) and \(u_2(x)\) is not \(o(1/x)\) (not even \(O(\exp(\exp(x))\).NEWLINENEWLINEWe note also that conditions of the Theorems 3.5 and 3.13 are unclear, because \(F(u_1)< 0<F(u_2),\) if \(u_1<0<u_2\).NEWLINENEWLINELet me offer a third counterexample to ``Theorem 3.7''. Let \(F_1(u)=u-1\), if \(u>1\) and \(F_1(u)=1\), if \(u\in(0,1)\). Let further be \(F_1(u)=-F_1(-u)\), if \(u<0\), and \(F(0)=0\). Then NEWLINE\[NEWLINE\lim_{u\to+0}F_1(u)=1,\,\lim_{u\to-0}F_1(u)=-1.NEWLINE\]NEWLINE Let further be \(x_2=0\), \(\alpha=\ell=m=1\), \(f_1(x)=1\), \(g_1(x)=x+3/2\). Let, finally, be \(u=u_3(x)=1+u_1(x)\) with \(u_1(x)\) defined as in the first counterexample. Then we have NEWLINE\[NEWLINE\Delta u_3(x)=-u_1(x+3/2)=-(u_3(x+3/2)-1)=-f_1(x)F_1(u_3(g_1(x))),NEWLINE\]NEWLINE the conditions (3.1) and (3.4) are fulfilled, but \(u_3(x)\) is a nonoscillatory solution of the equation (0.1).NEWLINENEWLINELet me offer a counterexample to ``Theorem 3.9''.NEWLINENEWLINELet \(v(x)=(1/2)^{[2x]}\). Clearly, NEWLINE\[NEWLINE\Delta v(x)=-(3/4)v(x)=-(3/8)v(x-1/2)=-6v(x+3/2),NEWLINE\]NEWLINE and \(v(x)\in{\mathbb Q}\) for any \(x\in{\mathbb R}\). Let further be \(u(x)=\sqrt{10}v(x)\) if \(\{x\}\geq1/2\), and \(u(x)=v(x)\) if \(\{x\}<1/2\). Let further be \(F_1(u)=(3/8)\sqrt{10}u\), if \(u\in{\mathbb Q}\) and \(F(u)=(6/\sqrt{10})u\), if \(u\not\in{\mathbb Q}\). Let further be \(L=1\), \(g_1(x)=x-1/2\), if \(\{x\}\geq1/2\), and \(g_1(x)=x+3/2\), if \(\{x\}<1/2\). Let finally be \(f_1(x)=1\). We prove now that NEWLINE\[NEWLINE\Delta u(x)=f_1(x)F_1(g_1(x)).NEWLINE\]NEWLINE If \(\{x\}\geq1/2\), then \(\{x+1\}\geq1/2\), and NEWLINE\[NEWLINE\Delta u(x)=\sqrt{10}\Delta v(x)=-(3/8)\sqrt{10}v(x-1/2).NEWLINE\]NEWLINE Since \(\{x\}\geq1/2\) it follows that \(\{x-1/2\}<1/2\), and therefore \(v(x-1/2)=u(x-1/2)\in{\mathbb Q}\). Hence NEWLINE\[NEWLINE-(3/8)\sqrt{10}v(x-1/2)=-F_1(u(x-1/2))=-F_1(u(g_1(x))).NEWLINE\]NEWLINE If \(\{x\}<1/2\), then \(\{x+1\}<1/2\), NEWLINE\[NEWLINE\Delta u(x)=\Delta v(x)=-6v(x+3/2).NEWLINE\]NEWLINE Since \(\{x\}<1/2\) it follows that \(\{x+3/2\}>1/2\), and therefore \(u(x+3/2)=\sqrt{10}v(x+3/2)\not\in{\mathbb Q}\). Therefore NEWLINE\[NEWLINE\Delta u(x)=\Delta v(x)=-(6/\sqrt{10})u(x+3/2)=-F_1(u(g_1(x))).NEWLINE\]NEWLINE Clearly, \(| F(u)|\geq | u|=L| u|\) for any \(u\in{\mathbb R}\). Clearly, NEWLINE\[NEWLINER_1=\bigcup_{n=0}^{+\infty} [n+1/2,n+1)NEWLINE\]NEWLINE is an infinite set, and \(Lf_1(x)=1\geq1\).NEWLINENEWLINELet me to discuss some examples given by the authors.NEWLINENEWLINEFirst, we consider their Example 3.2, which they deduce from Theorem 3.1.NEWLINENEWLINEExample 3.2. Each solution of the equation NEWLINE\[NEWLINE\Delta_{\alpha,\ell}u(x)+2\alpha^{[(2x+2\ell)/\ell]}(-1)^{[x/\ell]} = 0,NEWLINE\]NEWLINE where \(\alpha\geq1\), is oscillatory.NEWLINENEWLINELet us consider this equation for \(\alpha=\ell=1\). We have NEWLINE\[NEWLINE\Delta_{1,1}u(x)+2(-1)^{[x]}=0.NEWLINE\]NEWLINE Then NEWLINE\[NEWLINEu(x)=u(\{x\})+\sum_{\kappa=0}^{[x]-1} 2(-1)^{\kappa+1}.NEWLINE\]NEWLINE Hence, if \(\gamma>2\) is fixed and \(u(x)\geq\gamma\) for \(x\in[0,1)\), then \(u(x)>\gamma-2\) for all \(x\geq0\). Therefore, \(u(x)\) is neither oscillatory nor \(o(1)\).NEWLINENEWLINEIn Example 3.4 to Theorem 3.3, the authors consider the following difference equation NEWLINE\[NEWLINE\Delta_{\alpha,\ell}u(x)=-(x(\alpha+1)+\alpha\ell) \alpha^{[(2x+2\ell)/\ell]}(-1)^{[x/\ell]},NEWLINE\]NEWLINE where \(\alpha\in(0,1)\) and assert that all solutions of this equation are oscillatory. Let me consider this equation for \(\ell=1\), \(\alpha\in(0,1)\). We rewrite this equation in the form NEWLINE\[NEWLINEu(x+1)=\alpha u(x)-\alpha^{2x+2}((\alpha+1)x+\alpha).NEWLINE\]NEWLINE Let \(z(x)=u(x)\alpha^{-x}\). Then we have NEWLINE\[NEWLINEz(x+1)=z(x)-\alpha^{x+1}((\alpha+1)x+\alpha).NEWLINE\]NEWLINE Hence NEWLINE\[NEWLINEz(x)=z(\{x\}-\sum_{n=0}^{[x]-1}\alpha^{\{x\}+1+n} ((\alpha+1)(\{x\}+n)+\alpha)).NEWLINE\]NEWLINE Since NEWLINE\[NEWLINE\big| \sum_{n=0}^{[x]-1}\alpha^{\{x\}+1+n} ((\alpha+1)(\{x\}+n)+\alpha)\big|\leq\sum_{n=0}^{+\infty}\alpha^{1+n} ((\alpha+1)(1+n)+\alpha)=\gamma\in(0,+\infty),NEWLINE\]NEWLINE it follows that, if \(z(x)>\gamma\) for all \(x\in[0,1)\), then \(z(x)>0\) for all the \(x\in[0,+\infty)\).NEWLINENEWLINEIn Example 3.6 to Theorem 3.5 the authors assert that all the solutions of the equation \(\Delta_{\alpha,\ell}u(x) + 2\alpha(-\alpha)^{[x/\ell]} = 0\), where \(\alpha\geq1\), are oscillatory. Let \(\ell=1\), \(z(x)=\alpha^{-[x]}u(x)\). Then NEWLINE\[NEWLINE\Delta_{1,1}z(x)=-2\alpha^2(-1)^{[x]},NEWLINE\]NEWLINE and NEWLINE\[NEWLINEz(x)=z(\{x\})-2\alpha^2\sum_{k=0}^{[x]-1}(-1)^k.NEWLINE\]NEWLINE If the inequality \(z(x)>2\alpha^2\) holds for all \(x\in[0,1)\), then \(z(x)>0\) for all the \(x\geq0\).NEWLINENEWLINEIn Theorems 3.19--3.23, the authors study the case, when the equation is nonoscillatory.NEWLINENEWLINETheorem 3.19. Let \(\alpha>0\) and \(=\{0,+\infty\}\times\{(u,v)\in{\mathbb R}^2: v+(\beta-\alpha)u=0\}\). Further, let {\parindent=6mm \begin{itemize}\item[(i)] \(f(x,u,v)=0\), if \((x,u,v)\in S\), \item[(ii)] \(f(x,u,v)=[v+(\beta-\alpha)u]+\alpha[v+(\beta-\alpha)u]^2>0\) if \((x,u,v)\in[0,+\infty)\times{\mathbb R}^2\diagdown S\). NEWLINENEWLINE\end{itemize}} Then the equation (0.2) is nonoscillatory.NEWLINENEWLINELet me present a part of the proof given by the authors.NEWLINENEWLINEWe cite: ``However NEWLINE\[NEWLINE\Delta^2_{\alpha(\ell)}(k_1)=\Delta_{\alpha(\ell)}(k_1+\ell)- \Delta_{\alpha(\ell)}(k_1)=u(k_1+2\ell)-\alpha u(k_1)=0\text{''}.NEWLINE\]NEWLINE It is well known that NEWLINE\[NEWLINE\Delta^2_{\alpha,\ell}u(x)= u(x+2\ell)-2\alpha u(x+\ell)+\alpha^2u(x).NEWLINE\]NEWLINE Theorem 3.21. Let \(\alpha>0\) and \(T=\{0,+\infty\}\times\{(u,v)\in{\mathbb R}^2: v+\beta u=0\}\). Further, let NEWLINE\[NEWLINEf(x,u,v)[v+\beta u]+\alpha(v+\beta u)[v+(\beta-\alpha)u]<0NEWLINE\]NEWLINE if \((x,u,v))\in {\mathbb R}^2\diagdown T\) then difference equation (0.2) is nonoscillatory.NEWLINENEWLINETheorem 3.23. If \(\alpha=\beta=1\) and \( f(x,u,v)(u+v)>0\) if NEWLINE\[NEWLINE(x,u,v)\in[0,+\infty)\times{\mathbb R}^2.NEWLINE\]NEWLINE Then the equation (0.2) is nonoscillatory.NEWLINENEWLINEIn Theorems 3.25--3.27 the authors suppose that \(\alpha<0\), but at the beginning of the article (see the text directly below (0.1) and (0.2)), the authors declare that ``\(\alpha,\,\beta\) and \(\ell\) are positive constants''.NEWLINENEWLINETheorem 3.25. Let \(\alpha<0\) and \(S=\{0,+\infty\}\times\{(u,v)\in{\mathbb R}^2: v+(\beta-\alpha)u=0\}\). Further, let {\parindent=6mm \begin{itemize}\item[(i)] \(f(x,u,v)=0\), if \((x,u,v)\in S\), \item[(ii)] \(f(x,u,v)=[v+(\beta-\alpha)u]+\alpha[v+(\beta-\alpha)u]^2<0\), if \((x,u,v)\in[0,+\infty)\times{\mathbb R}^2\diagdown S\).NEWLINENEWLINE\end{itemize}} Then the equation (0.2) is oscillatory.NEWLINENEWLINETheorem 3.27. Let \(\alpha<0\) and \(T=\{0,+\infty\}\times\{(u,v)\in{\mathbb R}^2: v+\beta u=0\}\). Further, let NEWLINE\[NEWLINEf(x,u,v)[v+\beta u]+\alpha(v+\beta u)[v+(\beta-\alpha)u]<0NEWLINE\]NEWLINE if \((x,u,v)\in {\mathbb R}^2\diagdown T\), then difference equation (0.2) is oscillatory.NEWLINENEWLINEConclusion. The article contains several mistakes.