On invariant solutions of linear time-fractional diffusion-wave equations with variable coefficients
Sodbaatar Adiya, Khongorzul Dorjgotov, Bayarmagnai Gombodorj, Hiroyuki Ochiai, Uuganbayar Zunderiya ∗ · 2026
Read On invariant solutions of linear time-fractional diffusion-wave equations with variable coefficients by Sodbaatar Adiya, Khongorzul Dorjgotov, Bayarmagnai Gombodorj, Hiroyuki Ochiai, Uuganbayar Zunderiya ∗ (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-21267.txt.
Introduction
Sodbaatar Adiya and four colleagues find exact invariant solutions for a class of linear time-fractional diffusion-wave equations with variable coefficients, using Lie symmetry analysis. They first pin down the tiny symmetries tied to those coefficients, one by one. Those symmetries then yield solutions in terms of Mittag-Leffler functions, generalized Wright functions, and Fox H-functions. The equations model odd transport and wave motion in many real physical systems, giving the symmetry route a direct payoff in physics.
Catalog record
| CREATOR | Sodbaatar Adiya, Khongorzul Dorjgotov, Bayarmagnai Gombodorj, Hiroyuki Ochiai, Uuganbayar Zunderiya ∗ |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 27968 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.21267
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
On invariant solutions of linear time-fractional diffusion-wave equations with variable coefficients Thanks: ∗ Corresponding Author: U. Zunderiya Sodbaatar Adiya, Khongorzul Dorjgotov, Bayarmagnai Gombodorj, Hiroyuki Ochiai, Uuganbayar Zunderiya ∗ Abstract . We study invariant solutions of a certain class of time-fractional diffusion-wave equations with variable coefficients via Lie symmetry analysis. In physics, the fractional diffusion equation describes transport dynamics that are governed by anomalous diffusion while the fractional wave equation describes oscillations and wave propagation in various physical systems. In order to obtain exact invariant solutions of these equations, we firstly determine infinitesimal symmetries with respect to the variable coefficients of the equations. With the help of these symmetries, we then find solutions in terms of Mittag-Leffler functions, generalized Wright functions and Fox H-functions. 2020 Mathematics Subject Classification 35R11, 35B06, 33E12 1. Introduction Fractional diffusion-wave equations (FDWEs) have accurate descriptions of many physical processes including the modeling of anomalous diffusive systems, wave propagation phenomena and problems in electromagnetic theory. For example, many electromagnetic and acoustic problems in physics can be modeled accurately and efficiently using FDWE [ 21 , 22 ] . The fractional diffusion equation was introduced in physics by studying diffusion processes in media with fractal geometry [ 22 ] , while the fractional wave equation was introduced by investigating the propagation of mechanical diffusive waves in visco-elastic media which exhibit a power-law creep [ 16 ] . Further, time FDWEs were applied to the propagation of stress waves in visco-elastic media relevant to acoustics and seismology [ 17 ] . Recently, FDWEs were used in order to obtain the generalized formula for retarded potentials in electrodynamics with Maxwell equations containing fractional time derivatives [ 26 ] . For more research related to fractional differential equations and their applications in various fields of physics and other sciences, we recall the works [ 8 , 13 , 14 , 20 , 23 , 25 , 27 , 28 , 29 , 30 , 1 , 2 ] and references therein. In this study, we consider a class of linear time FDWEs with variable coefficients of the following form (1.1) \partial^{\alpha}_{t}u(x,t)=a(x)^{2}u(x,t)_{xx}+b(x)u(x,t)_{x},\quad a(x)\neq 0, where time derivative order \alpha is a positive real number, a(x) and b(x) are sufficiently differentiable functions. The Riemann-Liouville fractional differentiation is defined as \displaystyle\partial^{\alpha}_{t}u(x,t):=\left\{\begin{array}[]{ll}\frac{\partial^{n}u}{\partial t^{n}},&\mbox{ for }\alpha=n\in\mathbb{N},\\ \frac{1}{\Gamma(n-\alpha)}\frac{\partial^{n}}{\partial t^{n}}\int_{0}^{t}\frac{u(x,s)}{(t-s)^{\alpha-n+1}}ds,&\mbox{ for }\alpha\in(n-1,n),\text{ }n\in\mathbb{N}.\end{array}\right. By employing Lie symmetry analysis, we investigate symmetries and obtain invariant solutions to Eq. ( 1.1 ). Our results contain previously known solutions as their particular cases in a sense of time derivative order \alpha and the coefficients a(x) and b(x) . For example, symmetries and invariant solutions to Eq. ( 1.1 ) when \alpha=2 and b(x)=0 , are studied in [ 4 ] as well as in [ 15 , 24 , 10 , 11 ] when \alpha=1 . For \alpha>1 , symmetries and invariant solutions to Eq. ( 1.1 ) with constant coefficients i.e. a(x)=\text{const} and b(x)=0 , are obtained in [ 5 , 9 ] . For 0<\alpha<1 , invariant solutions of Eq. ( 1.1 ) with a(x)=ax^{m} , a>0 and b(x)=bx^{m-1} are given in [ 19 ] by using Laplace transformations. In the following sections of this paper, we first obtain the Lie point symmetries of Eq. ( 1.1 ) depending on the coefficient functions. We then explicitly derive invariant solutions corresponding to infinitesimal symmetries. 2. Group classification of time FDWEs We look for invariant solutions of time FDWEs described by Eq. ( 1.1 ) under Lie point symmetry. The infinitesimal and extended infinitesimal generators of Eq. ( 1.1 ) are given as X=\xi(x,t,u)\partial_{x}+\tau(x,t,u)\partial_{t}+\eta(x,t,u)\partial_{u}\quad\text{and} (2.1) X^{*}=X+\mu^{(1)}\partial_{u_{x}}+\mu^{(2)}\partial_{u_{xx}}+\mu^{(\alpha)}\partial_{\partial^{\alpha}_{t}u}, respectively in [ 7 ] , where \displaystyle\mu^{(1)} \displaystyle= \displaystyle\eta_{x}+(\eta_{u}-\xi_{x})u_{x}-\tau_{x}u_{t}-\xi_{u}u_{x}^{2}-\tau_{u}u_{x}u_{t}, \displaystyle\mu^{(2)} \displaystyle= \displaystyle\eta_{xx}+(2\eta_{xu}-\xi_{xx})u_{x}-\tau_{xx}u_{t}+(\eta_{uu}-2\xi_{xu})u_{x}^{2}-2\tau_{xu}u_{x}u_{t}-\xi_{uu}u_{x}^{3} \displaystyle-\tau_{uu}u_{x}^{2}u_{t}+(\eta_{u}-2\xi_{x})u_{xx}-2\tau_{x}u_{xt}-3\xi_{u}u_{xx}u_{x}-\tau_{u}u_{xx}u_{t}-2\tau_{u}u_{xt}u_{x}, \displaystyle\mu^{(\alpha)} \displaystyle= \displaystyle\partial^{\alpha}_{t}\eta+(\eta_{u}-\alpha D_{t}\tau)\partial^{\alpha}_{t}u-u\partial^{\alpha}_{t}\eta_{u} \displaystyle+\sum_{n=1}^{\infty}\left[\binom{\alpha}{n}\partial^{n}_{t}\eta_{u}-\binom{\alpha}{n+1}D_{t}^{n+1}\tau\right]\partial^{\alpha-n}u-\sum_{n=1}^{\infty}\binom{\alpha}{n}D_{t}^{n}\xi\partial^{\alpha-n}u_{x}+\mu, with \mu=\sum_{n=2}^{\infty}\sum_{m=2}^{n}\sum_{k=2}^{m}\sum_{r=0}^{k-1}\binom{\alpha}{n}\binom{n}{m}\binom{k}{r}\frac{1}{k!}\frac{t^{n-\alpha}}{\Gamma(n+1-\alpha)}\left(-u\right)^{r}\frac{\partial^{m}}{\partial t^{m}}\left(u^{k-r}\right)\frac{\partial^{n-m+k}\eta}{\partial t^{n-m}\partial u^{k}}, \binom{\alpha}{n}=\frac{(-1)^{n-1}\alpha\Gamma(n-\alpha)}{\Gamma(1-\alpha)\Gamma(n+1)}\quad\text{and}\quad D_{t}=\frac{\partial}{\partial t}+u_{t}\frac{\partial}{\partial u}+u_{xt}\frac{\partial}{\partial u_{x}}+\cdots. It should be noted that \mu=0 when \mu is linear in u . Moreover, the initial condition becomes (2.2) \tau(x,t,u)|_{t=0}=0 due to the fixed lower limit in the integral in ( 1 ). The infinitesimal symmetries for group-invariant solutions of Eq. ( 1.1 ) should satisfy the determining equation (2.3) \left.X^{*}\left(\frac{\partial^{\alpha}u}{\partial t^{\alpha}}-a(x)^{2}u_{xx}-b(x)u_{x}\right)\right|_{(\ref{1})}=0. By substituting ( 2.1 ) into ( 2.3 ) and equating the coefficients of various linearly independent variables to zero, we obtain the following system of equations: (2.4) \displaystyle\tau_{x}=0,\quad\tau_{u}=0,\quad\xi_{u}=0,\quad\binom{\alpha}{n}D_{t}^{n}\xi=0,~\forall n\in\mathbb{N}, (2.5) \displaystyle a(x)^{2}(2\xi_{x}-\alpha D_{t}\tau)-2a(x)a(x)^{\prime}\xi=0, (2.6) \displaystyle-a(x)^{2}\eta_{uu}=0,\quad\binom{\alpha}{n}\partial^{n}_{t}\eta_{u}-\binom{\alpha}{n+1}D_{t}^{n+1}\tau=0,~\forall n\in\mathbb{N}, (2.7) \displaystyle\partial^{\alpha}_{t}\eta-u\partial^{\alpha}_{t}\eta_{u}-a^{2}\eta_{xx}-b(x)\eta_{x}+\mu=0, (2.8) \displaystyle b(x)\left(\xi_{x}-\alpha D_{t}\tau\right)-a^{2}(2\eta_{xu}-\xi_{xx})-b(x)^{\prime}\xi=0. The equations in ( 2.4 ) and ( 2.5 ) along with the condition ( 2.2 ) yield us (2.9) \tau(t)=\frac{s_{1}}{\alpha}t\quad\mbox{and}\quad\xi(x)=a(x)\left(s_{2}+\frac{s_{1}}{2}\int_{\beta}^{x}\frac{dr}{a(r)}\right), where s_{1},s_{2}\in\mathbb{R} and \beta is a suitable real number. By solving the equations in ( 2.6 ) and ( 2.8 ), we have \mu=0 and (2.10) \eta=\left[-\frac{1}{2}\left(\frac{b(x)}{a(x)}-a(x)^{\prime}\right)\left(s_{2}+\frac{s_{1}}{2}\int_{\beta}^{x}\frac{dr}{a(r)}\right)+s_{3}\right]u+d(x,t), where d(x,t) is a function depending on x and t only. Furthermore, if we substitute ( 2.9 ) and ( 2.10 ) into ( 2.7 ), we see that d(x,t) is an arbitrary solution of Eq. ( 1.1 ) and (2.11) 2\left(c(x)^{\prime}a(x)+\frac{1}{2}c(x)^{2}\right)^{\prime}s_{2}\\ =-\left[\left(c(x)^{\prime}a(x)+\frac{1}{2}c(x)^{2}\right)^{\prime}\int_{\beta}^{x}\frac{dr}{a(r)}+2\left(c(x)^{\prime}a(x)+\frac{1}{2}c(x)^{2}\right)\frac{1}{a}\right]s_{1}, where c(x)=\frac{b(x)}{a(x)}-a(x)^{\prime} . Thus, for the arbitrary functions a(x) and b(x) , Eq. ( 1.1 ) has the following infinitesimal symmetries X_{d(x,t)}=d(x,t)\partial_{u}\quad\mbox{ and }\quad X_{1}=u\partial_{u}, where d(x,t) is an arbitrary solution of Eq. ( 1.1 ). By analyzing the relation ( 2.11 ), all of the infinitesimal symmetries of Eq. ( 1.1 ) that satisfy the determining Eq. ( 2.3 ), are given by the following Table 1 . Table 1. \begin{array}[]{c}b(x)=a(x)c(x)+a(x)a(x)^{\prime},~\omega(x)=\int_{\beta}^{x}\frac{dr}{a(r)}+\lambda_{1},\\ (\text{where }\lambda_{1}\in\mathbb{R}\text{ and }\beta\text{ is a suitable real number})\end{array} Cases c(x) X 1 any sufficiently differentiable X_{d(x,t)}=d(x,t)\partial_{u}, function except cases 2-8 X_{1}=u\partial_{u} 2 \dfrac{\left(\ln\omega(x)+\lambda_{2}\right)}{\omega(x)\left(\ln\omega(x)+\lambda_{2}-2\right)}, X_{d(x,t)}=d(x,t)\partial_{u}, X_{1}=u\partial_{u}, \lambda_{2}\in\mathbb{R} X_{2}=a(x)\omega(x)\partial_{x}+\frac{2}{\alpha}t\partial_{t}-\dfrac{c(x)}{2}u\partial_{u} 3 \dfrac{\left(\lambda_{2}+1\right)\bigl(\lambda_{3}\omega(x)^{\lambda_{2}}-\lambda_{2}+1\bigl)}{\omega(x)\left(\lambda_{3}\omega(x)^{\lambda_{2}}+\lambda_{2}+1\right)}, X_{d(x,t)}=d(x,t)\partial_{u}, X_{1}=u\partial_{u}, \lambda_{2},\lambda_{3}\in\mathbb{R},~\lambda_{2}\neq-1 X_{3}=a(x)\omega(x)\partial_{x}+\frac{2}{\alpha}t\partial_{t}-\dfrac{c(x)}{2}u\partial_{u} 4 \dfrac{\bigl(\lambda_{2}^{2}+1\bigr)}{\omega(x)\left(\lambda_{2}\tan\left(\frac{\lambda_{2}}{2}\ln\omega(x)+\lambda_{3}\right)+1\right)} , X_{d(x,t)}=d(x,t)\partial_{u}, X_{1}=u\partial_{u}, \lambda_{2},\lambda_{3}\in\mathbb{R}, \lambda_{2}\neq 0 X_{4}=a(x)\omega(x)\partial_{x}+\frac{2}{\alpha}t\partial_{t}-\dfrac{c(x)}{2}u\partial_{u} 5 \dfrac{\lambda_{2}\left(\epsilon e^{\lambda_{2}\omega(x)}-1\right)}{\epsilon e^{\lambda_{2}\omega(x)}+1}, X_{d(x,t)}=d(x,t)\partial_{u}, X_{1}=u\partial_{u}, \lambda_{2}\in\mathbb{R},~\lambda_{2}\neq 0,~\epsilon=\pm 1 X_{5}=a(x)\partial_{x}-\dfrac{c(x)}{2}u\partial_{u} 6 \lambda_{2}\tan\left(-\frac{\lambda_{2}}{2}\omega(x)\right), X_{d(x,t)}=d(x,t)\partial_{u}, X_{1}=u\partial_{u}, \lambda_{2}\in\mathbb{R},\lambda_{2}\neq 0 X_{6}=a(x)\partial_{x}-\dfrac{c(x)}{2}u\partial_{u} 7 0 X_{d(x,t)}=d(x,t)\partial_{u}, X_{1}=u\partial_{u}, X_{7}=a(x)\omega(x)\partial_{x}+\frac{2}{\alpha}t\partial_{t}, X_{8}=a(x)\partial_{x} 8 \dfrac{2}{\omega(x)} X_{d(x,t)}=d(x,t)\partial_{u}, X_{1}=u\partial_{u}, X_{7}=a(x)\omega(x)\partial_{x}+\frac{2}{\alpha}t\partial_{t}, X_{9}=a(x)\partial_{x}-\dfrac{c(x)}{2}u\partial_{u} In cases 2-8 in Table 1 , Eq. ( 1.1 ) becomes (2.12) \partial^{\alpha}_{t}u(x,t)=a(x)^{2}u(x,t)_{xx}+\left(a(x)\overline{c}(\omega(x))+a(x)a(x)^{\prime}\right)u(x,t)_{x},\quad a(x)\neq 0, where \overline{c}(\omega(x))=c(x) . By the transformation \omega=\int_{\beta}^{x}\frac{dr}{a(r)}+\lambda_{1} , Eq. ( 2.12 ) is then reduced the equation (2.13) \partial^{\alpha}_{t}u(\omega,t)=u(\omega,t)_{\omega\omega}+\overline{c}(\omega)u(\omega,t)_{\omega}. Hence, we are now able to formulate the following theorem. Theorem 2.1 . All of the infinitesimal symmetries of Eq. ( 2.13 ) are given by the following cases: (1) for any sufficiently differentiable \overline{c}(\omega) , X_{d(\omega,t)}=d(\omega,t)\partial_{u}\quad\mbox{ and }\quad X_{1}=u\partial_{u}, where d(\omega,t) is an arbitrary solution of Eq. ( 2.13 ); (2) for \overline{c}(\omega)=\dfrac{\left(\ln\omega+\lambda_{2}\right)}{\omega\left(\ln\omega+\lambda_{2}-2\right)} , \lambda_{2}\in\mathbb{R} , \displaystyle X_{d(\omega,t)},\quad X_{1}\quad\mbox{ and } \displaystyle X_{2}=\omega\partial_{\omega}+\frac{2}{\alpha}t\partial_{t}-\dfrac{\ln\omega+\lambda_{2}}{2\left(\ln\omega+\lambda_{2}-2\right)}u\partial_{u}; (3) for \overline{c}(\omega)=\dfrac{\left(\lambda_{2}+1\right)\bigl(\lambda_{3}\omega^{\lambda_{2}}-\lambda_{2}+1\bigl)}{\omega\left(\lambda_{3}\omega^{\lambda_{2}}+\lambda_{2}+1\right)} , \lambda_{2},\lambda_{3}\in\mathbb{R} , \lambda_{2}\neq-1 , \displaystyle X_{d(\omega,t)},\quad X_{1}\quad\mbox{ and } \displaystyle X_{3}=\omega\partial_{\omega}+\frac{2}{\alpha}t\partial_{t}-\dfrac{\left(\lambda_{2}+1\right)\left(\lambda_{3}\omega^{\lambda_{2}}-\lambda_{2}+1\right)}{2\left(\lambda_{3}\omega^{\lambda_{2}}+\lambda_{2}+1\right)}u\partial_{u}; (4) for \overline{c}(\omega)=\dfrac{\bigl(\lambda_{2}^{2}+1\bigr)}{\omega\left(\lambda_{2}\tan\left(\frac{\lambda_{2}}{2}\ln\omega+\lambda_{3}\right)+1\right)} , \lambda_{2},\lambda_{3}\in\mathbb{R} , \lambda_{2}\neq 0 , \displaystyle X_{d(\omega,t)},\quad X_{1}\quad\mbox{ and } \displaystyle X_{4}=\omega\partial_{\omega}+\frac{2}{\alpha}t\partial_{t}-\dfrac{\bigl(\lambda_{2}^{2}+1\bigr)}{2\left(\lambda_{2}\tan\left(\frac{\lambda_{2}}{2}\ln\omega+\lambda_{3}\right)+1\right)}u\partial_{u}; (5) for \overline{c}(\omega)=\dfrac{\lambda_{2}\left(\epsilon e^{\lambda_{2}\omega}-1\right)}{\epsilon e^{\lambda_{2}\omega}+1} , \lambda_{2}\in\mathbb{R},~\lambda_{2}\neq 0,~\epsilon=\pm 1, X_{d(\omega,t)},\quad X_{1}\quad\mbox{ and }\quad X_{5}=\partial_{\omega}-\dfrac{\lambda_{2}\left(\epsilon e^{\lambda_{2}\omega}-1\right)}{2\left(\epsilon e^{\lambda_{2}\omega}+1\right)}u\partial_{u}; (6) for \overline{c}(\omega)=\lambda_{2}\tan\left(-\frac{\lambda_{2}}{2}\omega\right) , \lambda_{2}\in\mathbb{R} , \lambda_{2}\neq 0 , X_{d(\omega,t)},\quad X_{1}\quad\mbox{ and }\quad X_{6}=\partial_{\omega}-\frac{\lambda_{2}}{2}\tan\left(-\frac{\lambda_{2}}{2}\omega\right)u\partial_{u}; (7) for \overline{c}(\omega)=0 , X_{d(\omega,t)},\quad X_{1},\quad X_{7}=\omega\partial_{\omega}+\frac{2}{\alpha}t\partial_{t}\quad\mbox{ and }\quad X_{8}=\partial_{\omega}; (8) for \overline{c}(\omega)=\dfrac{2}{\omega} , X_{d(\omega,t)},\quad X_{1},\quad X_{7}\quad\mbox{ and }\quad X_{9}=\partial_{\omega}-\frac{1}{\omega}u\partial_{u}. Table 1 and Theorem 2.1 gives us a complete group classification of Eq. ( 1.1 ). In the next section, we express invariant solutions as solutions of so-called reduced fractional ordinary differential equations. Finally, we obtain exact invariant solutions to Eq. ( 2.13 ) by using similarity transformations. 3. Invariant solutions of time FDWEs In this section, we look for invariant solutions that remain unaltered under