🔢 mathematics

Self-similar solutions to the time-fractional Porous-Medium Equation

This paper establishes the existence of optimal-range self-similar solutions with constant finite mass to the time-fractional Porous-Medium Equation for all spatial dimensions $d \ge 1$ , distinguishing between compactly supported solutions in the slow-diffusion regime and heavy-tailed solutions in the sub-critical fast-diffusion regime while also analyzing the linear limit.

Original authors: David Gómez-Castro, Łukasz Płociniczak, Juan Luis Vázquez

Published 2026-04-13

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: David Gómez-Castro, Łukasz Płociniczak, Juan Luis Vázquez

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are watching a drop of ink spread out in a glass of water. In the classic world of physics (the "classical" case), this ink spreads smoothly and instantly, filling the whole glass over time, following a predictable bell-curve shape. This is the standard Porous Medium Equation, used to model things like water soaking into soil or gas flowing through rock.

Now, imagine a twist: Time itself behaves differently. Instead of flowing smoothly like a river, time in this new world is "fractional." It's like the ink is spreading through a thick, sticky honey where the past constantly whispers to the present, slowing things down and making the history of the drop matter more than it usually does. This is the Time-Fractional Porous Medium Equation.

The paper you provided is a mathematical detective story. The authors (David, Lukasz, and Juan) are trying to find the "perfect shapes" (called self-similar solutions) that this ink drop takes as it spreads in this sticky, time-warped world. They want to know: Does the drop spread out? Does it stop at a certain edge? Does it leave a long, thin tail behind?

Here is the breakdown of their discovery, using simple analogies:

1. The Two Main Characters: Slow vs. Fast Diffusion

The behavior of the ink depends heavily on how "thick" the fluid is (represented by a number called $m$ ).

The "Slow" Drip ( $m > 1$ ): The Compact Blob
Imagine pushing a thick blob of dough through a sieve. It moves slowly, and it has a very sharp, clean edge. It doesn't leak out infinitely; it stops at a distinct boundary.
- The Discovery: The authors proved that even with the weird "fractional time," if the fluid is thick enough, the solution still forms a compact blob. It has a sharp edge (a "free boundary") and doesn't stretch out forever. It's like a snowball that melts but keeps its round shape until it's gone.
- The Limit: If the fluid gets infinitely thick (mathematically, $m \to \infty$ ), the blob turns into a perfect, flat-topped cylinder (a "mesa"), just like in the classic world.
The "Fast" Drip ( $m < 1$ ): The Heavy Tail
Now imagine a very thin, watery fluid. In the classic world, this spreads out so fast it creates a "tail" that stretches infinitely far, getting thinner and thinner but never quite reaching zero.
- The Discovery: In this fractional time world, the authors found that these thin fluids also spread out with heavy tails. However, the shape of the tail is different. Instead of the exponential decay (like a bell curve) seen in normal time, these tails decay like a simple power law (a straight line on a log-log graph). It's like a long, fading echo that never quite disappears.

2. The "Magic" Threshold

There is a critical point (called $m_c$ ) that acts like a cliff edge.

If the fluid is too thin (below this threshold), the math breaks down in a way that implies the drop has infinite mass (it's too spread out to measure).
The authors proved that as long as you are above this cliff (even just a tiny bit), you get a well-behaved solution with a finite amount of ink (mass) that is conserved over time. This range is "optimal," meaning you can't go any lower without the math collapsing.

3. The "Memory" Effect

One of the most fascinating parts of this paper is how the "fractional time" remembers the past.

In normal physics, once a drop of ink starts spreading, it forgets exactly where it started after a moment.
In this fractional world, the system has memory. The authors showed that the "force" driving the spread ( $\Delta u^m$ ) always "remembers" the initial point where the ink was dropped (the Dirac delta). It's as if the ink drop is constantly looking back at its starting point, pulling itself together slightly differently than it would in normal time.

4. The Bridge Between Worlds

The authors didn't just solve the problem for the weird fractional time; they built a bridge to the normal world.

They showed that if you slowly turn the "fractional time knob" back to normal (making $\alpha$ go from 0.9 to 1.0), their strange, new solutions smoothly transform into the famous, well-known Barenblatt solutions (the standard shapes we already know).
They also checked what happens when the fluid gets super-thick or super-thin, and the math holds up, matching our intuition from the classical world.

5. The "Very Singular" Solution

In the fast-diffusion range, there is a special, extreme solution called the Very Singular Solution (VSS).

Imagine a drop of ink that starts as an infinitely dense point and instantly explodes outward.
The authors found that all the other "normal" solutions (with finite mass) form a sort of fan around this extreme explosion. They are like ripples that are slightly less chaotic than the main shockwave, all hugging the shape of this singular solution as they spread out.

Summary: Why Does This Matter?

This paper is a roadmap for the future.

For Scientists: It gives them the exact formulas and shapes to use when modeling real-world phenomena where time doesn't flow normally—like how pollutants move through complex, layered soil, or how heat moves through materials with "memory" (like certain polymers or biological tissues).
For Mathematicians: They proved that these solutions not only exist but are unique (there's only one right answer for each scenario) and behave predictably. They even provided a numerical recipe (an algorithm) for computers to draw these shapes.

In a nutshell: The authors took a complex, time-warped physics problem, found the perfect shapes the solutions take, proved they are stable and unique, and showed how they connect back to the familiar world we live in. They turned a chaotic, sticky mess of equations into a clear, organized picture of how things spread when time itself is a bit "fractional."

1. Problem Statement

The paper investigates the existence and properties of non-negative self-similar solutions with finite mass to the time-fractional Porous-Medium Equation (PME):
$\partial_t^\alpha u = \Delta(u^m)$
where:

$t > 0, x \in \mathbb{R}^d$ ( $d \ge 1$ ).
$\partial_t^\alpha$ is the Caputo fractional derivative of order $\alpha \in (0, 1)$ .
$m > 0$ is the diffusion exponent.
The initial condition is a Dirac delta mass: $u(0, \cdot) = M\delta_0$ .

The primary goal is to establish the existence of self-similar profiles $u(t, x) = t^{-a}U(t^{-b}|x|)$ that conserve the total mass $M$ for all $t > 0$ , and to characterize their behavior across different regimes of $m$ (slow diffusion, linear, and fast diffusion).

2. Methodology

The authors employ a combination of scaling analysis, integral equation formulation, and fixed-point arguments.

Scaling Analysis: By assuming the self-similar form, the authors derive the scaling exponents $a$ and $b$ :
$a = \frac{\alpha d}{2 + d(m-1)}, \quad b = \frac{\alpha}{2 + d(m-1)}$
These exponents are positive if and only if $m > m_c = (1 - 2/d)^+$ .
Integral Equation Formulation: A key contribution is the reduction of the partial differential equation (PDE) to a non-linear integral equation for the profile $U(z)$ (where $z = t^{-b}|x|$ ):
$U(z)^m = \int_z^\infty K(z, \rho) U(\rho) d\rho$
The kernel $K(z, \rho)$ is explicitly derived and depends on the fractional order $\alpha$ and dimension $d$ . The authors significantly simplify the expression of this kernel compared to previous literature.
Existence Proofs:
- Slow Diffusion ( $m > 1$ ): The authors use the method of sub- and super-solutions. They construct a reference profile $V_{\alpha, m}$ (compactly supported) and prove the existence of a solution via monotone iteration (Picard iterations).
- Fast Diffusion ( $m_c < m < 1$ ): They construct solutions that form an "asymptotic fan" around a Very Singular Solution (VSS) of infinite mass. They use sub- and super-solutions based on the VSS profile $U^*(z) \sim z^{-2/(1-m)}$ .
- Linear Case ( $m=1$ ): The solution is recovered via the Fourier transform, yielding a profile related to the Mittag-Leffler function.
Weak Formulation: The paper rigorously defines the notion of a "very weak solution" for the time-fractional equation, handling the singularity at the origin and the initial Dirac mass using integration by parts and properties of the Riemann-Liouville derivative.

3. Key Contributions and Results

A. Optimal Range of Existence

The paper proves the existence of self-similar profiles for all spatial dimensions $d \ge 1$ and exponents $m > m_c = (d-2)^+/d$ . This range is shown to be optimal; for $m \le m_c$ , finite-mass self-similar solutions do not exist (they either have infinite mass or extinguish in finite time).

B. Classification of Solutions by Regime

The behavior of the profile $U(z)$ changes drastically depending on $m$ :

Slow Diffusion ( $m > 1$ ):
- Compact Support: Solutions are compactly supported, i.e., $U(z) = 0$ for $z \ge R$ .
- Free Boundary Behavior: Near the free boundary $z \to R^-$ , the solution behaves like a power law:
  $U(z) \sim (R-z)^{\frac{2-\alpha}{m-1}}$
  The paper provides the exact sharp constant for this asymptotic behavior.
- Mesa Limit ( $m \to \infty$ ): As $m \to \infty$ , the solution converges to a characteristic function (a "mesa" shape), independent of $\alpha$ .
Linear Diffusion ( $m = 1$ ):
- The solution is the fundamental solution of the time-fractional heat equation.
- The profile decays exponentially (Mittag-Leffler type) rather than algebraically.
- The authors prove uniqueness of the weak solution in this case.
Mildly Fast Diffusion ( $m_c < m < 1$ ):
- Heavy Tails: Solutions are strictly positive everywhere ( $z \in (0, \infty)$ ) and decay algebraically.
- Asymptotic Fan: Finite-mass solutions form a family converging to the Very Singular Solution (VSS) $U^*(z) = c^* z^{-2/(1-m)}$ as mass $M \to \infty$ .
- Tail Expansion: The authors derive the precise asymptotic expansion of the tail:
  $U(z) \sim U^*(z) \left( 1 - C z^{-\gamma^*} \right)$
  where $\gamma^*$ is determined by a specific spectral equation involving the kernel.

C. Continuity and Limits

Continuity in Mass: For a fixed mass $M$ , the solution $U_{\alpha, m, M}$ is continuous with respect to $m$ even as $m$ crosses the critical value $1$.
Classical Limit ( $\alpha \to 1$ ): As $\alpha \to 1^-$ , the fractional profiles converge to the classical Barenblatt profiles (for $m \neq 1$ ) and the Gaussian (for $m=1$ ).
Singularity at Origin: The paper characterizes the behavior of $U^m$ near $z=0$ . For $d \ge 2$ , $U^m$ behaves like the Newtonian potential (singular), while for $d=1$ , it is bounded. This reflects the "memory" of the initial Dirac mass in the fractional time derivative.

D. Numerical Scheme

The authors propose a numerical scheme based on Picard iterations for the integral equation. They provide a discretization of the kernel and demonstrate convergence for the $m > 1$ case, illustrating the transition from compact support to heavy tails as parameters vary.

4. Significance

Completeness: This work provides the first complete existence theory for self-similar solutions of the time-fractional PME across all relevant dimensions and exponents ( $m > m_c$ ).
Unification: It unifies the understanding of slow diffusion, linear diffusion, and fast diffusion in the fractional time setting, showing how the fractional order $\alpha$ modifies the classical Barenblatt profiles (e.g., changing the decay rate at the free boundary from $(1-z)^{1/(m-1)}$ to $(1-z)^{(2-\alpha)/(m-1)}$ ).
Memory Effects: The results highlight how the Caputo derivative introduces a "memory" effect, causing the solution to retain the singularity of the initial Dirac mass at $x=0$ for all time, unlike the classical case where the singularity smooths out immediately for $m>1$ .
Foundation for Future Work: The paper establishes the necessary tools (integral equation, weak formulation, asymptotic analysis) to tackle open problems such as the stability of these profiles as global attractors for general initial data and the treatment of the "very singular" range ( $0 < m < m_c$ ).

In summary, the paper rigorously extends the classical theory of the Porous-Medium Equation to the time-fractional setting, providing explicit characterizations of the solution profiles, their asymptotic behaviors, and their dependence on the fractional order and diffusion exponent.