Page 103 - IJOCTA-15-1
P. 103

On a robust stability criterion in the Cattaneo–Hristov diffusion equation

            Definition 2. The unperturbed solution ¯y j ≡ 0 of  according to the expression (9) and (11). This
            (8) is called robustly stable with respect to exter-  function satisfies (2) and (3). From this last ex-
                                     , if for all ε j > 0 there  pression, we observe that
            nal perturbations v j ∈ V δ j
            exists η j (ε j ) > 0 such that under the condition:       ∂y        X
                                                                                  ∞
                    ≤ δ j < η j (ε j ), any other solution of (8)         (x, t) =   β j cos(β j x)y j (t),  (15)
                                                                       ∂x
            ∥v j ∥ V δ j
                            < ε j .                                              j=1
            satisfies ∥y j ∥ Y δ j
                                                              and
                                                                                   ∞
                                                                        ∂y        X
            We show that the unperturbed solution ¯y j of (8)              (x, t) =   sin(β j x) ˙y j (t).  (16)
            is robustly stable with respect to external pertur-         ∂t         j=1
                            according to Definition 2, which
            bations v j ∈ V δ j                               We obtain the following result.
            follows from the inequalities (11) and (13). In-
                                                                             P ∞
                                                                                   δ
            deed, we observe that for all ε j > 0, the inequality  Lemma 1. If  j=1 j is convergent with sum
                    < ε j is obtained after choosing          equal to δ, then the norm of the function y de-
            ∥y j ∥ Y δ j
                                                                                                     ≤ 2Mδ,
                                                              fined in (14) admits the bound: ∥y∥ Y Ω
                                         ε j
                 δ j ≤ η j (ε j ) =  (              )         where
                                         1        κ 2                            2
                               2 max         , 1 +                              R      R        κ 2
                                       2κ 1 β 2   κ 1              M = max       2   ,     , 1 +
                                           j                                   2π κ 1 2πκ 1     κ 1
                                                                                                         (17)
                                    ε j                                     κ 2
                            =          ,                            = 1 +    .
                                      κ 2                                   κ 1
                               2 1 +
                                      κ 1
                                                              Proof. We observe from (14) and (11) that
            where the last equality is obtained from our as-              ∞           ∞               ∞
            sumption made in (4) about the thermal conduc-     |y(x, t)| ≤  X |y j (t)| ≤  X  δ j  <  R  2 X  δ j .
                                                                                                  2
            tivity and elastic conductivity constants. If δ j > 0                        κ 1 β j 2  π κ 1
                                                                         j=1         j=1              j=1
            satisfies this inequality, then every solution y j ob-         P  ∞
                                                                                 δ
                                                              If the series   j=1 j is convergent, then y is
                                          on the right-hand
            tained by choosing v j ∈ V δ j                                                    P ∞
                                                                                                    δ
            side of the linear fractional differential equation  bounded on Ω. We note that if  j=1 j is con-
                                              < ε j . Hence   vergent, then the Weierstrass M-test implies that
            (8) satisfies the inequality ∥y j ∥ Y δ j
                                                              the series on the right-hand side of (14) converges
            the unperturbed solution ¯y j is robustly stable in
            the sense of Definition 2.                        uniformly to y on Ω.
                                                              Using the expression (15) and the inequality (11)
                                                              we obtain
                                                                               ∞
                                                                      ∂y       X
            5. Robust stability criterion                              (x, t) ≤    β j |y j (t)|

                                                                      ∂x
                                                                                 j=1
            In this section we obtain our main result with the                   ∞               ∞
                                                                                 X    δ j    R  X
            help of the robust stability criterion that has been              ≤           <         δ j .
                                  , which will allow us to es-                      κ 1 β j  πκ 1
            obtained for the set Y δ j                                           j=1            j=1
            tablish a robust stability criterion in the set Y Ω            P ∞                          ∂y
                                                              If the series  j=1 j is convergent, then     is
                                                                                 δ
            using Definition 1.                                                            P ∞          ∂x
                                                                                                 δ
                                                              bounded on Ω. If the series    j=1 j is conver-
            Let ε > 0. We define E ε as the family of sequences  gent, then the Weierstrass M-test implies that
            {ε j } j∈N of positive real numbers for which the fol-  the series on the right-hand side of (15) converges
                            P ∞
            lowing equality   j=1 j = ε holds. This set in-   uniformly to  ∂y  on Ω.
                                  ε
                                                                           ∂x
            cludes the sequences {ε p,j } j∈N defined by ε p,j =  From (16) and (12) we finally obtain
            ε(1 − p)p j−1  for each p ∈ (0, 1).
                                                                           ∞                    ∞
                                                                  ∂y       X                  κ 2  X
            We consider a sequence of positive real num-           (x, t) ≤    | ˙y j (t)| ≤ 2 1 +    δ j .

            bers {δ j } j∈N and define the following function      ∂t      j=1                κ 1  j=1
                      P ∞
            v(x, t) =       sin(β j x)v j (t), where v j ∈ V δ j  is
                        j=1                                   If the series  P ∞  δ                     ∂y  is
            arbitrary for each j ∈ N. If {y j } j∈N is the se-               j=1 j is convergent, then  ∂t
                                                                                             ∞
                                                                                           P
                                                              bounded on Ω. If the series    j=1 j is conver-
                                                                                                 δ
            quence of Fourier coefficients y j that are obtained
                                                              gent, then the Weierstrass M-test implies that
            as a solution of (8) when the function v j is con-
                                                              the series on the right-hand side of (15) converges
            sidered on the right-hand side, then the solution              ∂y
            of (1) is the function                            uniformly to  ∂t  on Ω.
                                 ∞                            As a conclusion from the three inequalities, it fol-
                                X
                       y(x, t) =   sin(β j x)y j (t),  (14)   lows that ∥y∥    ≤ 2Mδ. We observe that the
                                                                           Y Ω
                                j=1                           value of M is a consequence of the assumption
                                                            97
   98   99   100   101   102   103   104   105   106   107   108