🔢 mathematics

The Euler system of gas dynamics

This survey examines recent results on the well-posedness and ill-posedness of the Euler system of gas dynamics, focusing on how solutions emerge as limits of consistent approximations and the extent to which the First and Second laws of thermodynamics ensure the uniqueness of physically admissible solutions.

Original authors: Eduard Feireisl

Published 2026-04-01

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Eduard Feireisl

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 trying to predict the weather, or perhaps how a drop of ink spreads in a glass of water. In the world of physics, there is a famous set of rules called the Euler System. Think of these rules as the "constitution" for how ideal gases (like air) move, push, and carry energy.

For a long time, mathematicians believed that if you knew the starting conditions (where the air is, how fast it's moving, how hot it is), these rules would give you one single, perfect prediction of the future. This is what we call a "well-posed" problem.

However, this paper by Eduard Feireisl reveals a shocking truth: The Euler System is broken.

Here is the breakdown of the paper's journey, explained with simple analogies.

1. The Problem: Too Many Answers

In the real world, if you drop a stone in a pond, the ripples happen in a specific way. But mathematically, the Euler equations are so flexible that for many starting situations, they don't just give you one answer; they give you infinite answers.

The Analogy: Imagine you are driving a car. You know your starting point and your destination. The rules of the road say you can drive there. But the math says you could drive there in a straight line, or you could drive in a circle for 10 hours, or drive backwards and then forwards, and all of those paths are technically "correct" according to the basic laws of motion.
The Shock: Recently, mathematicians discovered that for many starting conditions, there are actually infinitely many valid ways the gas could behave. This is called "ill-posedness." It means the math alone cannot tell us what will actually happen in the real world.

2. The Culprit: The Second Law of Thermodynamics

To fix this, we usually bring in a helper: Entropy.

The Analogy: Think of entropy as a measure of "messiness" or "disorder." The Second Law of Thermodynamics is like a strict rule: "Things always get messier over time; they never spontaneously get neater."
The Expectation: We thought that if we added this rule (that the gas must always get messier), it would filter out the weird, impossible solutions and leave us with just the one "real" solution.
The Reality: Even with this rule, the math still allows for infinite solutions. The "messiness" rule isn't strict enough to pick a winner.

3. The New Approach: "Dissipative Measure-Valued" (DMV) Solutions

Since we can't find a single perfect solution, the author suggests we change our perspective. Instead of looking for one specific path, we should look at the average behavior of all possible paths.

The Analogy: Imagine a crowd of people trying to leave a stadium. Instead of tracking every single person (which is impossible and chaotic), we look at the "cloud" of people. We ask: "Where is the crowd most likely to be?"
The Math: The paper introduces DMV solutions. These aren't single points; they are "clouds" of possibilities (mathematically called measures). They represent the state of the gas when it is turbulent, oscillating, and chaotic.
Key Insight: If the gas is calm, this "cloud" shrinks down to a single point (a normal solution). But if the gas is turbulent, the cloud stays wide, representing the chaotic energy of the system.

4. The Selection Process: Picking the "Best" Cloud

Now we have a cloud of infinite possibilities. How do we pick the one that nature actually chooses? The paper explores a few "selection criteria" (rules to pick the winner).

A. The "Maximal Messiness" Rule (DiPerna & Dafermos)

The Idea: Nature loves to create disorder. So, maybe the "real" solution is the one that creates the most entropy (messiness) possible.
The Result: This is a promising idea, but it's hard to prove that such a "most messy" solution actually exists or is unique. It's like trying to find the single most chaotic day in history; it's a hard concept to pin down.

B. The "Energy Minimization" Rule (The Two-Step Selection)

The Idea: The author proposes a two-step filter to narrow down the cloud.
1. Step 1: Pick the solutions that create the most entropy (the messiest ones).
2. Step 2: From those, pick the one that minimizes a specific "distance" from a calm, equilibrium state.
The Result: This creates a unique, mathematically sound "solution semigroup." It's a machine that takes your starting data and spits out one specific prediction. It works, but it feels a bit like we are forcing the math to behave rather than discovering a natural law.

C. The "Turbulence" vs. "Smoothness" Debate

The paper ends with a fascinating philosophical question.

Scenario A (Turbulence): Maybe the "infinite solutions" are real. Maybe the gas does oscillate wildly, and the "cloud" of DMV solutions is the true description of reality. The gas is inherently chaotic.
Scenario B (Smoothness): Maybe the "infinite solutions" are just a mathematical artifact. Maybe in the real world, the gas always picks one smooth path, and the other infinite paths are just mathematical ghosts that don't exist physically.

The Big Takeaway

The paper concludes that we are currently stuck between two worlds:

Mathematically: The Euler equations are broken. They allow for infinite answers, even with the laws of thermodynamics.
Physically: We don't know which answer nature picks.

The author suggests that the only way to solve this might not be through more math, but through experiments or computer simulations. We need to see what happens in the real world (or in high-precision simulations) to see if the gas behaves like a single smooth path or a chaotic, oscillating cloud.

In short: The rules of gas dynamics are too loose to predict the future uniquely. We have to invent new rules (like "pick the messiest path" or "pick the path closest to calm") to make the math work, but we aren't 100% sure if those new rules match the physical universe. The mystery of "which solution is the real one" remains unsolved.

1. Problem Statement

The paper addresses the fundamental issue of well-posedness for the Euler system of gas dynamics, which models the time evolution of an ideal compressible fluid. The system consists of conservation laws for mass, momentum, and energy.

The Core Conflict: While classical (strong) solutions exist locally in time for smooth initial data, they develop singularities (shock waves) in finite time. Consequently, the analysis must shift to weak solutions.
The Non-Uniqueness Crisis: Recent developments using convex integration (De Lellis, Székelyhidi) have demonstrated that even within the class of entropy-admissible weak solutions (those satisfying the Second Law of Thermodynamics), the Euler system is ill-posed. Specifically, for certain "wild" initial data, there exist infinitely many distinct entropy-admissible weak solutions.
The Goal: The paper investigates whether a unique, physically admissible solution can be recovered by:
1. Broadening the solution concept to Dissipative Measure-Valued (DMV) solutions.
2. Identifying specific selection criteria (based on thermodynamics) that isolate a unique solution or a solution semigroup from the set of all possible DMV solutions.

2. Methodology and Framework

A. Dissipative Measure-Valued (DMV) Solutions

The author adopts the framework of DMV solutions, originally proposed by DiPerna and Majda. Instead of seeking a single function $(\varrho, m, S)$ , the solution is defined as a triple consisting of:

Macroscopic variables: $(\varrho, m, S)$ representing the weak limits of density, momentum, and entropy.
Young Measure: A parametrized probability measure $\mathcal{V}_{t,x}$ capturing oscillations in the sequence of approximations.
Concentration Measure: A measure $C$ capturing concentrations (singularities) in the energy/momentum flux.

Key Property: The set of all DMV solutions emanating from a given initial data set is convex and compact in a suitable weak topology. This convexity is crucial for applying selection principles.

B. Consistent Approximations

The paper defines "consistent approximations" as sequences of solutions to perturbed problems (e.g., Navier-Stokes-Fourier with vanishing viscosity or numerical schemes) that satisfy the Euler equations modulo errors vanishing as the perturbation parameter $\epsilon \to 0$ .

The limit of such sequences yields a DMV solution.
If the limit preserves total energy, it is a weak solution. If energy is lost (converted to "turbulent energy"), it is a dissipative solution.

C. Selection Procedures

Since the set of DMV solutions is generally not a singleton, the paper explores various selection processes to identify a unique "physical" solution or a solution semigroup. These rely on the Second Law of Thermodynamics (entropy maximization).

3. Key Contributions and Results

1. Existence of Global DMV Solutions

The paper confirms that for any finite energy initial data, a global-in-time DMV solution exists. This resolves the existence problem in a generalized sense, even when classical or standard weak solutions fail to exist or are non-unique.

2. Analysis of Maximality Criteria

The author examines two historical criteria for selecting unique solutions based on entropy production:

DiPerna's Criterion: Selects solutions that maximize the total entropy at any given time $t$ $t$ .
- Result: While maximal solutions exist, it is an open problem whether they are always weak solutions (DiPerna's conjecture). However, it is proven that maximal solutions become "regular" (energy defect vanishes) as $t \to \infty$ .
Dafermos' Criterion: A local criterion selecting solutions that maximize the rate of entropy production at the first point of non-uniqueness.
- Result: It is proven that any Dafermos-maximal DMV solution is a weak solution. However, the existence of such a maximal solution for arbitrary data remains an open problem.

3. Construction of a Solution Semigroup

The paper proposes and validates methods to construct a solution semigroup (a mapping $U(t, \text{data}) \to \text{solution}$ ) that satisfies:

Semigroup property: $U(t+s) = U(t) \circ U(s)$ .
Measurability: The mapping is Borel measurable with respect to initial data.
Asymptotic stability: Turbulent energy (energy defect) vanishes as $t \to \infty$ .

Two specific selection procedures are detailed:

Two-Step Selection (Convexity-based):
1. Maximize the total entropy functional $F_S = \int e^{-t} \int S \, dx$ .
2. If the set is not a singleton, minimize the energy defect (turbulent energy) functional $F_E$ .
- Outcome: This yields a unique solution semigroup.
Single-Step Selection (Thermodynamic Distance):
- Minimize a cost functional based on the Bregman divergence (distance) between the DMV solution and a background equilibrium state. The functional is $F = \int e^{-t} \int (E - \vartheta S) \, dx$ .
- Outcome: Due to the strict convexity of the energy functional, this single step uniquely identifies an admissible solution.

4. Absolute Entropy Maximizers

The paper introduces the concept of an "absolute entropy maximizer"—a solution that minimizes the thermodynamic distance for all decay rates $\lambda$ in the cost functional.

Theorem: If such a solution exists, it is a weak solution (energy defect is zero for all $t > 0$ ).
Open Problem: Characterizing the initial data for which such maximizers exist.

4. Significance and Implications

Bridging Math and Physics: The paper provides a rigorous mathematical framework to handle the non-uniqueness of the Euler equations, which is a major obstacle in fluid dynamics. It suggests that "non-uniqueness" might be an artifact of looking at the wrong class of functions, and that the "physical" solution is a specific selection from the DMV set.
Turbulence Interpretation: The DMV framework interprets the Euler system as the inertial range of turbulence. The "turbulent energy" (energy defect) represents the energy cascade to small scales. The selection criteria effectively determine how this energy is dissipated.
Numerical Relevance: The paper connects theoretical DMV solutions to consistent approximations (numerical schemes). It raises the critical question of whether the set of limits of numerical approximations is convex and if it coincides with the set of all DMV solutions.
Open Problems: The survey highlights several deep open problems, including:
- Does a Dafermos-maximal solution always exist?
- Is the set of limits of consistent approximations convex?
- Can the non-uniqueness of limits be attributed to "random" errors in computation (Monte Carlo analogy)?

Conclusion

Eduard Feireisl's survey argues that while the Euler system is ill-posed in the classical weak sense, it can be rendered well-posed by adopting the DMV solution framework and applying thermodynamic selection principles (specifically entropy maximization or distance minimization to equilibrium). This approach yields a unique solution semigroup that is consistent with the Second Law of Thermodynamics and asymptotically recovers energy conservation, offering a promising path forward for both mathematical theory and numerical simulation of compressible flows.