I'd like to know the existence of the general solution to the following fractional differential equation $$D_{0+}^{\alpha} y(t)=0 \text{,}\label{1} \tag{1}$$
where $\alpha \in (1,2)$ and
$$ D_{0+}^{\alpha} y(t)=\frac{1}{\Gamma(2-\alpha)} \left( \frac{d}{dt} \right)^2 \int_0^t \frac{y(s)}{(t-s)^{\alpha-1}} \, ds \text{.} $$
I know that $t^{\alpha-1}$ and $t^{\alpha-2}$ are solutions to problem \eqref{1}.
My question is whether $C_1t^{\alpha-1} + C_2 t^{\alpha-2}$ is the general solution to problem \eqref{1}.
Note that $D_{0+}^{\alpha}$ has second order derivative.
My attempt:
Let $Y(t)=\int_0^t \frac{y(s)}{(t-s)^{\alpha-1}}$ for fixed $y$. Then
$y$ is a solution to problem \eqref{1} if and only if $Y$ is a solution to problem $Y''(t)=0$.
We also know that
$D_1+D_2 t$ : a general solution to problem $Y''(t)=0$.
$y(t)=t^{\alpha-1} \implies Y(t)=Ct$ and $y(t)=t^{\alpha-2} \implies Y(t)=C$ for a constant $C=\int_0^1 (1-v)^{1-\alpha}dv. $
I think I can prove it using them, but I'm not sure.
I would be grateful if you could give any comment for my question.
Suppose $y: \mathbf R^+\rightarrow \mathbf R$ is such that $\int_0^\infty |f(t)e^{-at}|dt<\infty$ for some $a\ge0$. You can apply the Laplace transform $\mathcal L$ on $Y$ which is defined by OP in his question since $$Y(t) = y*t^{1-\alpha}$$ where $*$ stands for the convolution operator, and $Y$ is a power function thus converges under the Laplace transform. We will repeatedly use the property that $$\mathcal L\{t^q\}(s)=\frac{\Gamma(q+1)}{s^{q+1}},\quad \forall t>0,\,\text{Re}(s)>0,\, \text{Re}(q)>-1.$$ and the transformation is invertible. Apply the Laplace transform on $Y=D_1+D_2t$. $$\mathcal L\{Y\} = \mathcal L\{y\}\mathcal L\{t^{1-\alpha}\}=\mathcal L\{y\}\frac{\Gamma(2-\alpha)}{s^{2-\alpha}}=\frac{D_1}{s}+\frac{D_2}{s^2},$$ then $$\mathcal L\{y\}(s)=\frac{a_1}{s^{\alpha-1}}+\frac{a_2}{s^\alpha}.$$ Take the inverse Laplace transform $$y(t)=b_1t^{\alpha-2}+b_2t^{\alpha-1}.$$ $a$'s and $b$'s above are constants.