Existence-uniqueness result for a nonlinear \(n\)-term fractional equation (Q1014686)

The author studies the existence of a unique solution for the nonlinear fractional-order (Caputo definition) differential equation \[ b_oD^{\beta_o}u(t)+ \sum_{i=1}^{m-1}b_iD^{\beta_i}u(t)+ \sum_{i=m}^{n-1}b_iD^{\alpha_i}u(t)+b_nD^{\alpha_n}u(t)= f(t,u(t)),\quad t\in (0,T]\tag{1} \] \[ u(0)= f(0),\quad u_t(0)= g(0) \] with different values of \(\beta_i \in (0,1),~\alpha_i \in (1,2)\), \(\beta_i,~\alpha_i \in (0,1)\) and \(\beta_i,~\alpha_i \in (1,2)\) respectively and the function \(f\) is continuous and satisfies a kind of Lipschitz condition. But the results (Lemmas 4-8 and Theorems 1-3) are not correct because there is no (under the stated assumptions) equivalence between the integral equation \[ u(t)=C~ G(t) + \int_{0}^t~G(t-\tau) f(\tau, u(\tau)) d\tau \] and equation (1) as the author claims. This is since the derivatives, in case when \(\alpha_n \in (1,2)\), \[ \frac{d^2}{dt^2} \int_{0}^t~G(t-\tau) f(\tau, u(\tau)) d\tau\text{ and } D^{\alpha_n} u(t) = I^{2-\alpha_n}\frac{d^2}{dt^2} \int_{0}^t~G(t-\tau) f(\tau, u(\tau)) d\tau \] do not exist. And, in case when \(\alpha_n \in (0,1)\), the derivatives \[ \frac{d}{dt} \int_{0}^t~G(t-\tau) f(\tau, u(\tau)) d\tau\text{ and } D^{\alpha_n} u(t) = I^{1-\alpha_n}\frac{d}{dt} \int_{0}^t~G(t-\tau) f(\tau, u(\tau)) d\tau \] do not exist. The results can be true if the fractional order derivatives are in the Riemann-Liouville definition and the initial values will be \[ u(0)= 0,\quad u_t(0)= 0. \]