A Kac system interacting with two heat reservoirs: the shearing case

Federico Bonetto and Matthew Powell · 2026

Read A Kac system interacting with two heat reservoirs: the shearing case by Federico Bonetto and Matthew Powell (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-2606-30529.txt.

Introduction

Federico Bonetto and Matthew Powell study a simplified gas model brushing against two heat reservoirs at once. In the Kac model, particles swap energy through random collisions, mimicking a real gas in a tractable way. Here a system of particles in three dimensions collides with two much larger reservoirs. Each reservoir starts as a Maxwellian distribution, the bell-shaped spread of particle velocities, but centred on a different average velocity, giving a shearing setup.

The pair prove that, over a time window that grows with the reservoir size, this finite setup behaves like one with two infinite reservoirs, described as shearing dynamic Maxwellian thermostats. In plain terms, the huge reservoirs act as steady heat baths whose flow pattern keeps sliding. When the two reservoirs share the same temperature offset and average speed, their approximation holds uniformly for all times, with no growing error.

Catalog record

CREATORFederico Bonetto and Matthew Powell
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE77 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2606.30529

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

A Kac system interacting with two heat reservoirs: the shearing case Federico Bonetto and Matthew Powell Abstract L^{2} , [ 4 ] for convergence in L^{1} and [ 5 ] for the GTW metric d_{2} . It also helped clarify the conditions for the emergence of macroscopic behaviors like propagation of chaos and the validity of the Boltzmann-Kac equation, see e.g. [ 6 , 7 , 8 ] . Finally, the ideas behind the Kac evolution are partially present in the DSMC scheme for numerical simulations, see [ 9 ] . In a series of papers (see [ 10 , 11 , 12 , 13 ] ) we have tried to extend the reach of the Kac model by adding interactions both with reservoirs (still modeled via random collisions) and Maxwellian thermostats (seen as infinite reservoirs), and by looking at particles moving in three spacial dimensions, instead of one as in the original model. In particular, in [ 13 ] we looked at a three dimensional system with M particles interacting with two large heat reservoirs with N particles each. The reservoirs are initially in equilibrium at temperatures T_{+} and T_{-} , and we assumed that the centers of mass of both the system and the reservoirs are initially at rest. We proved that the system + 2 reservoirs, when N>>M , that is the reservoirs are much larger than the system, is well approximated by a system of M particles interacting with 2 Maxwellian thermostats. This means that we can formally take the limit for N\to\infty and replace the collisions of the particle in the system with particles in the reservoirs with collisions with virtual particles randomly selected from Maxwellian distributions with temperature T_{+} or T_{-} . Contrary to the case of only one reservoir, see [ 11 ] , our results in [ 13 ] were only valid for a time shorter than \sqrt{N}/M . Observe that, by taking the limit N\to\infty , we eliminate the influence of the system on the reservoirs. This leads to the notion of a thermostat with fixed temperature which was analyzed in [ 11 ] . On the other hand, for any fixed N , over a time of the order of N/M the temperatures of the reservoir and system evolve. That is, no matter how large (but finite) the reservoirs are, there will be temperature evolution. Thus the approximation with thermostats at fixed temperature can only be valid for a relatively short time. In this paper, we continue this line of study by considering a generalization of the model considered in [ 13 ] in which the centers of mass of the two reservoirs initially have non-zero velocities \vec{p}_{+} and \vec{p}_{-} , in addition to having different temperatures. If these velocities are different, they will create a shear in the system. On a macroscopic level, the energy stored in the center of mass motion of the reservoirs is converted into heat so that, in this case, the final temperature of the reservoirs is higher than the average of the initial temperatures. It is remarkable that one can write closed evolution equations for the momentum of the centers of mass and the temperatures of the system and reservoirs, see ( 9 ) and ( 10 ) below. As discussed in Subsection 2.1 , two time scales emerge. On a short time scale of order 1 the system reaches a quasi stationary state while on a longer time scale of order N/M the temperatures and center of mass velocities of the two reservoirs tend to become equal and the all system + 2 reservoirs converges to equilibrium. We can thus attempt to take the limit N\to\infty in a more careful way. To create our effective evolution we assume that the particles in the system interact with 2 infinite reservoirs whose center of mass velocities and temperatures evolve in the same way as the corresponding quantities for the reservoirs with finite N . This give rise to the notion of dynamic thermostat as introduced in Subsection 1.3 below. More precisely, we compare the evolution of the system + 2 reservoir with that of a system interacting with 2 Maxwellian thermostat with center of mass velocities and temperatures, which evolves in time. We believe that the notion of dynamic thermostat is the major conceptual novelty of this paper and that thermostats of this type will be instrumental in extending our results beyond the time range obtained here. Our main result is contained in Theorem 1 below. Similarly to [ 13 ] , we obtain that the system + 2 reservoirs is well approximated by the system + 2 dynamic thermostats for times shorter than \sqrt{N}/M . On a technical level, as in our previous works, we use the GTW d_{2} metric (see [ 14 ] ) to measure the distance between the reservoir and thermostat evolution. Observe that the d_{2} metric requires the the distributions being compared have the same first moment. However, as already noticed, the first moment of the distributions of the system and/or the reservoirs evolve in time so that if the first moment of a thermostat remained fixed the d_{2} -distance between the reservoir evolution and a fixed thermostat evolution would diverge immediately. This makes the introduction of the dynamic thermostat necessary also at a technical level, at least as far as the center of mass velocities are involved. To prove our main theorem we follow the strategy of [ 13 ] based on a Duhamel expansion and on the functional inequality introduced in [ 11 , Lemma 3.3] . After this, we are left with estimating the evolution of the thermostated system. Due to the presence of the dynamic thermostats, this last step is considerable more involved than the corresponding estimates in [ 13 ] . In this sense, Subsection 3.2 is the novel technical achievement of this paper. It also makes clear that the main obstruction to a result uniform in time is the variation of the state of the reservoirs/thermostats over the long time scale. We also observe that if we suppose that the two reservoirs have equal temperatures ( T_{+}=T_{-} ) and initial velocities ( \vec{p}_{+}=\vec{p}_{-} ), then the system is essentially equivalent to a system interacting with one shearing reservoir. In this case there is no long time scale and both the shear and the temperature of the reservoir change very little along the evolution. Thus we are again able to obtain estimates uniform in time, see Corollary 2 . We note that the system involving only one thermostat can be analyzed using a suitable L^{2} norm, see [ 10 , 15 ] . Although the analysis is simpler, the L^{2} norm, as discussed for example in [ 10 ] , is not well suited for systems with many degrees of freedom. Moreover, it is unclear how to extend these L^{2} results to the case of a system interacting with more than one reservoir, even for a short time. In Subsections 1.1 and 1.2 we introduce the Kac evolution for the system and the reservoirs. These are included for the sake of completeness but can be skipped by the reader familiar with this family of models.

Read the complete text