🔢 mathematics

Barriers, Barenblatt solutions and regularity of soda can domains for the heat equation and nonlinear $p$ -parabolic equations

This paper investigates the regularity of the origin as a boundary point for "soda can" domains with nonconvex time sections in the context of the heat equation and nonlinear $p$ -parabolic equations, providing a complete characterization for cases where $p=2$ or $p<2n/(n+1)$ .

Original authors: Anders Björn, Jana Björn

Published 2026-05-01

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Anders Björn, Jana Björn

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 movie of a physical process, like heat spreading through a metal or a drop of ink diffusing in water. In mathematics, we often ask a specific question about the very end of this movie: As the scene fades to black (time $t=0$ ), does the temperature or concentration at the exact center point settle down smoothly to match the boundary conditions, or does it jump, glitch, or behave erratically?

This paper, written by Anders and Jana Björn, investigates this question for a very specific, oddly shaped "room" where these physical processes happen. They call these rooms "Soda Can Domains."

Here is a breakdown of their findings using everyday analogies.

1. The Shape: The "Soda Can"

Usually, mathematicians study heat flow in simple boxes or cylinders. But these authors are interested in a shape that looks like the inside bottom of a soda can.

The Shape: Imagine a cylinder where the floor isn't flat. Instead, the floor curves upward in the middle, like the concave bottom of a soda can.
The Math: As time moves forward toward zero, the "room" gets smaller and smaller. The walls curve inward so fast that the space eventually pinches off to a single point at the center (the origin) at the very last moment.
The Question: If you have a continuous temperature on the walls of this shrinking room, does the temperature at the very center (where the room disappears) settle down to a predictable value, or does it go haywire?

2. The Rules of the Game: The "Equation"

The paper studies two main types of physical laws governing how things move:

The Heat Equation ( $p=2$ ): This is the standard, linear way heat spreads (like a hot cup of coffee cooling down).
The $p$ -Parabolic Equation ( $p \neq 2$ ): This is a "nonlinear" version. Think of it as a more complex fluid or a substance that diffuses differently depending on how concentrated it is.
- If $p > 2$ , the substance is "stubborn" (degenerate diffusion).
- If $1 < p < 2$ , the substance is "fast" and spreads out aggressively (singular diffusion).

3. The Main Discovery: When Does the Center "Behave"?

The authors found that whether the center point behaves nicely (is "regular") depends entirely on how fast the soda can pinches shut compared to the type of diffusion happening inside.

They measured the "pinching speed" with a number called $l$ .

Fast Pinching (Large $l$ ): The walls curve in very sharply.
Slow Pinching (Small $l$ ): The walls curve in gently.

Here is what they discovered:

A. The "Standard" Heat Equation ( $p=2$ )

The Rule: If the room pinches shut slowly enough (specifically, if $l \ge 2$ ), the temperature at the center settles down perfectly. The boundary data is respected.
The Exception: If the room pinches shut too fast (if $l < 2$ ), the center point becomes "irregular." The temperature at the center might jump or fail to match the boundary, even if the boundary is smooth.
Special Case ( $n=2$ ): In a 2D world (like a flat sheet), the center is always regular, no matter how fast the room pinches, unless the room is just a hole in the middle of a sheet (a punctured cylinder).

B. The "Nonlinear" Equations ( $p \neq 2$ )

This is where it gets tricky. The rules change based on whether the substance is "stubborn" ( $p>2$ ) or "fast" ( $p<2$ ).

For "Fast" Diffusion ( $1 < p < 2$ ):
- If the room pinches very fast ( $l > p$ ), the center is regular.
- If the room pinches slowly ( $l$ is small), the center is often irregular. The authors found specific ranges where the center point "breaks" and refuses to settle, even though the walls are smooth.
- Analogy: If the substance spreads too fast and the room closes too slowly, the center gets "confused" and doesn't know what temperature to be.
For "Stubborn" Diffusion ( $p > 2$ ):
- If the room pinches fast ( $l > p$ ), the center is regular.
- If the room pinches moderately ( $p/(p-1) \le l \le p$ ), the center is regular only if the temperature changes on the walls aren't too steep. If the walls change temperature too sharply near the center, the center point might glitch.
- If the room pinches very slowly ( $l < p/(p-1)$ ), the authors don't know for sure yet. They leave this as an "Open Problem."

4. How They Solved It: The "Barrier" Method

To prove these results, the authors used a mathematical tool called a Barrier.

The Metaphor: Imagine you are trying to prove that a ball rolling down a hill will stop at the bottom. You build a "fence" (a barrier) around the ball. If you can show that the ball cannot escape the fence, and the fence forces the ball to stop at the bottom, you've proven it will stop.
In the Paper: They constructed special mathematical functions (barriers) that act like fences.
- If they could build a "family" of these fences that get tighter and tighter as they approach the center, they proved the center is Regular (the solution behaves).
- If they could show that no such fence could exist (or that a specific "bad" function could sneak through), they proved the center is Irregular.

They used two types of fences:

Power-type fences: Simple curves based on powers of distance.
Barenblatt fences: More complex shapes derived from known solutions to the diffusion equation, which act like perfect, custom-made molds for the problem.

5. Real-World Interpretations (From the Paper)

The authors suggest two ways to visualize this:

The Dissolving Substance: Imagine a ball of dye dissolving in water. The ball shrinks over time until it vanishes at $t=0$ . If the ball shrinks in the shape of a soda can, the math tells us whether the concentration of dye at the very center (where the ball vanished) will be predictable or chaotic.
The Moving Heater: Imagine a hot object (like a magma stream) moving through a layer of water and eventually leaving the layer at $t=0$ . If the cross-section of the object shrinks like a soda can, the math predicts whether the water temperature right where the object left will be smooth or if it will have a sudden jump.

Summary

This paper is a map of "regularity." It tells us exactly how fast a shrinking, curved room must close for the physics inside to remain calm and predictable at the very last moment.

Too slow? Chaos (Irregularity).
Just right? Calm (Regularity).
Too fast? Also Calm (Regularity).

The authors have filled in many gaps in this map, particularly for the tricky "nonlinear" cases, but they admit there are still a few uncharted territories (the "Open Problems") where the answer remains a mystery.

1. Problem Statement

The paper investigates the boundary regularity of the origin $(0,0)$ for the Dirichlet problem associated with the $p$ -parabolic equation:
$\partial_t u - \Delta_p u = 0, \quad \text{where } \Delta_p u = \text{div}(|\nabla u|^{p-2}\nabla u),$
in specific non-cylindrical domains known as "soda can domains".

These domains are defined as:
$\Theta_{l,\theta} := \{(x, t) \in \mathbb{R}^{n+1} : 0 < -t < \theta|x|^l < \theta\},$
where $l, \theta > 0$ . Geometrically, for $l > 1$ , these domains resemble the concave interior of the bottom of a soda can. The time sections are non-convex annuli with shrinking inner radii that converge to a punctured ball as $t \to 0$ .

The central question is: Under what conditions on the exponent $l$ , the dimension $n$ , and the parameter $p$ is the origin $(0,0)$ a regular boundary point?
A point is regular if the solution to the Dirichlet problem attains the continuous boundary data continuously at that point.

2. Methodology

The authors employ a combination of classical potential theory techniques adapted to the nonlinear parabolic setting:

Barrier Construction: The primary tool is the construction of barriers and barrier families. A barrier is a positive $p$ -superparabolic function vanishing at the boundary point but bounded away from zero elsewhere. The existence of a family of such barriers is equivalent to regularity (Theorem 3.3).
Scaling Arguments: The authors utilize scaling lemmas (Lemma 3.6) to show that for $p \neq 2$ , regularity is independent of the scaling parameter $\theta$ . This allows them to normalize domains for analysis.
Two Types of Barriers:
1. Power-type barriers: Functions of the form $v(x,t) \sim |x|^\alpha + t$ . These are used to establish regularity for specific ranges of $l$ and $p$ .
2. Barenblatt-type barriers: Utilizing the fundamental Barenblatt solution (a self-similar solution to the $p$ -parabolic equation) to construct barriers. This approach is particularly effective for the range $p/(p-1) \le l \le p$ when $p > 2$ .
Wiener Criterion (Heat Equation): For the linear case ( $p=2$ ), the authors apply the Wiener criterion for the heat equation (using thermal capacity) to derive necessary and sufficient conditions for regularity.
Comparison Principles: The proofs rely heavily on the parabolic comparison principle and the Perron method for constructing solutions.

3. Key Contributions and Results

The paper provides a nearly complete characterization of regularity for soda can domains, distinguishing between the linear case ( $p=2$ ) and the nonlinear cases ( $p \neq 2$ ).

A. The Nonlinear Case ( $p \neq 2$ )

Regularity for Large $l$ :
- If $1 < p \le 2$ and $l > p$ , the origin is regular (Proposition 1.2).
- If $p > 2$ and $l > p$ , the origin is regular (Theorem 1.1).
- If $p > n$ (regardless of $l$ ), the origin is regular (Proposition 1.3).
Irregularity for Small $l$ (when $1 < p < 2 \le n$ ):
- The origin is irregular if $l$ is sufficiently small. Specifically, if $p < \frac{2n}{n+1}$ , irregularity holds for $l \le p$ . If $p \ge \frac{2n}{n+1}$ , irregularity holds for $l < \frac{(n-p)(2-p)}{p-1}$ (Proposition 1.4).
Partial Regularity for Intermediate $l$ ( $p > 2$ ):
- For the range $p/(p-1) \le l \le p$ , the authors prove partial regularity. The origin is regular for boundary data $f$ that are "small" near the origin, satisfying a specific growth condition:
  $|f(x,t) - f(0,0)| \le C \min\{|x|, \delta\}^{p/(p-1)}.$
- This result is achieved via two distinct proofs: one using power-type barriers and another using Barenblatt solutions (Proposition 1.6).
- Open Problem: It remains unknown whether the origin is fully regular for large boundary data in the range $p/(p-1) \le l \le p$ when $p > 2$ .

B. The Linear Case ( $p = 2$ , Heat Equation)

The authors provide a complete characterization for $n \ge 2$ :

Dimension $n=2$ : The origin is regular for all $l > 0$ . However, it is irregular for the punctured cylinder (a limiting case).
Dimension $n \ge 3$ : The origin is regular if and only if $l \ge 2$ $l \geq 2$ .
- This contrasts with the nonlinear case where the threshold depends on $p$ .
- The proof utilizes the Wiener criterion involving thermal capacity of cylinders.

C. Comparison with Previous Work

The paper contrasts these results with Petrovskiĭ's criterion (1935) for convex time sections, where regularity depends on logarithmic corrections.
It highlights that for $p \neq 2$ , regularity is independent of radial scaling in $x$ or $t$ , a property not shared by the heat equation ( $p=2$ ).
It corrects a claim in Kilpeläinen–Lindqvist regarding the sufficiency of an exterior ball touching from the past for all $p$ , showing that a single barrier is insufficient for $1 < p < 2$ .

4. Significance and Implications

Geometric Insight: The study of "soda can" domains (non-convex, power-law shrinking) fills a gap between convex domains (Petrovskiĭ) and cylindrical domains. It reveals that the geometry of the time-section (annular vs. disk) significantly alters regularity conditions, especially for $p \neq 2$ .
Methodological Advancement: The successful application of Barenblatt solutions to construct barriers for $p$ -parabolic equations in non-cylindrical domains demonstrates a powerful technique for handling nonlinear degeneracy. The fact that both power-type and Barenblatt-type barriers yield the same sharp bounds for partial regularity suggests a deep structural connection.
Physical Interpretation:
- Diffusion: The results model the concentration of a substance diffusing from a shrinking source (e.g., a dissolving particle). Regularity implies the concentration approaches the boundary value continuously as the source vanishes; irregularity implies a potential jump or discontinuity.
- Heat Transfer: In the context of a heater/cooler moving through a medium, the results determine whether the temperature of the medium near the separation point reflects the final temperature of the body.
Open Problems: The paper identifies critical gaps, particularly regarding the full regularity for $p > 2$ in the intermediate range $p/(p-1) \le l \le p$ and the precise capacity formulas for parabolic cylinders in $n=2$ .

Summary Table of Regularity Conditions (Simplified)

Case	Condition for Regularity of $(0,0)$ in $\Theta_{l,\theta}$
$p > n$	Regular for all $l > 0$ .
$p = 2, n = 2$	Regular for all $l > 0$ .
$p = 2, n \ge 3$	Regular $\iff l \ge 2$ .
$1 < p < 2$	Regular if $l > p$ ; Irregular if $l$ is small (depends on $n$ ).
$p > 2$	Regular if $l > p$ . Partial regularity (small data) if $p/(p-1) \le l \le p$ .

This work significantly advances the understanding of boundary behavior for nonlinear parabolic equations in time-varying domains, providing sharp criteria and novel barrier constructions.