Bridging Filtering and Point-Splitting Approaches for Variable-Density Flows

arXiv:2601.01810 · physics.flu-dyn · Submitted 2026-01-05 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Physics. The day in general physics: instruments, methods and results across the field.

Kai: We're looking at the paper: "Bridging Filtering and Point-Splitting Approaches for Variable-Density Flows".

Mira: Here's the gist of the paper. Bridging Filtering and Point-Splitting Approaches for Variable-Density Flows develops a general empirical framework to determine multipoint correlation functions corresponding to any filtered energy definition in compressible turbulence The Favre filtered energy corresponds to an infinite series of multipoint correlation functions achieved by expanding the Favre velocity as a power series in local density fluctuations Filtering and Point-Splitting Approaches The paper begins by recalling the filtering and point-splitting approaches in uniform-density flows, where the large-scale energy is defined either through filtering or two-point correlation functions like the KHM relation The two approaches are related by noting that R(r) = R ub(k) / 2 e ik·rd 3k/2 by the WienerKhinchin theorem Angular averaging is shown to be qualitatively the same as low-pass filtering in dimensions d > 1.

Kai: Angular Averaging as Filtering The angular averaging of a field can be interpreted as a low-pass filtering operation where the Fourier transform of the radial delta function gives the equivalent filter This filter, Gδ K(k), is non-negative, normalized, and has its support on the shell of radius r centered at x.

Mira: The angular-averaged multipoint correlation functions are shown to approximate the moments of the filtered fields by defining Cp(r) = hχ(χ delta K) p−1 i.

Kai: Point-Splitting Analogue to the Favre Filtered Budget The Favre velocity ueK and the Favre filtered energy E(K) are defined, and they can be formally expanded as a polynomial series under the assumption that the density field is bounded from above The leading order term of this expansion corresponds to the two-point momentum-momentum correlator which has a regularized KHM relation in the infinite Reynolds number limit.

Paper summary: Kai: So this paper is called "Bridging Filtering and Point-Splitting Approaches for Variable-Density Flows". It's about finding a general way to figure out the multipoint correlation functions for any filtered energy definition in compressible turbulence.

Mira: It claims they develop an empirical framework to determine these correlation functions, specifically connecting the Favre filtered energy to an infinite series of multipoint correlation functions. The big idea is that this helps us understand energy transfers in flows where density isn't constant.

Lev: What matters here for real hardware is that this connects a theoretical definition, the Favre filtered energy, to something we can actually calculate using these correlation functions. It sets up a way to study how energy moves across different scales.

Kai: The paper starts by looking at the old ways of doing things in uniform-density flows where you used either filtering or two-point correlation functions like the Kármán–Howarth–Monin relation. They point out that these approaches are related because angular averaging is like a low-pass filter in dimensions greater than one.

Mira: That link is key because it suggests that what we do with angular averaging can actually be interpreted as a low-pass filtering operation, and the Fourier transform of the radial delta function gives us the equivalent filter.

Lev: So they're saying that if you average over all angles, you're effectively applying a specific type of filter to see how energy behaves at different scales. That sounds like something we need to consider when we design our numerical simulations for these complex flows.

Kai: They then define the angular-averaged multipoint correlation functions as approximating the moments of the filtered fields by defining some quantity, C p(r), which involves a filter delta K.

Mira: And they show that this quantity is essentially the product of one bare field and some other term involving that filter, which means it's a way to approximate those moments.

Lev: It sounds like they are providing an analogue for what we do in uniform flows, but now extended to variable density where the definition of filtered energy itself gets complicated. Where does this lead us practically?

Paper summary: Kai: They move on to defining the Favre velocity, u e, and the Favre filtered energy, E(K), and they expand these in powers of local density fluctuations, assuming the density is bounded from above.

Mira: The leading order term of that expansion corresponds to a two-point momentum-momentum correlator which has a regularized Kármán–Howarth–Monin relation in the infinite Reynolds number limit.

Lev: So they're saying that for very high Reynolds numbers, the basic energy transfer is still described by something we already know from simpler flows, even with density variations included. But what about the higher-order terms?

Kai: The buoyancy contribution to this filtered energy budget is expressed as a series expansion in terms of psi, and it involves two, three, and four-point correlation functions for terms up to second order in density fluctuations.

Mira: Similarly, the pressure contribution is expanded into a series where higher-order terms are needed if you want better agreement with the actual physics. That means getting a good picture requires more than just the first few terms of the expansion.

Lev: When we think about running this on real hardware, those higher-order correlation functions mean we're looking at needing more measurements and calculating more complex interactions between different scales. The overhead there is definitely going up.

Kai: The main result they get is that the Favre filtered energy can be expressed using a series of multipoint correlation functions, which comes from expanding the density-weighted velocity field in powers of local density fluctuations.

Mira: And this leads them to an approximation where C p is approximately equal to M p, which is just the p-th moment of the filtered fields evaluated with some arbitrary positive filter.

Lev: So, they're proposing that we can use this approximation, C p about M p, as a point-splitting analogue for terms in the Favre filtered energy budget. That seems like a solid starting point for studying energy transfers in these variable-density flows.

Kai: The whole point of this paper is to show this unified way of looking at inter-scale energy transfers in generic variable-density flows by writing the Favre filtered contributions as an infinite series of multipoint correlation functions.

Paper summary: Mira: They are trying to establish that the approximation C p about M p is the most appropriate choice for studying energy transfers in variable-density flows. They expect this series expansion to converge based on flow parameters like the density ratio.

Lev: If that series converges, it means we have a systematic way to capture these energy transfers without needing an infinite number of terms, which would be impossible on real hardware anyway. It gives us a concrete way to assess the error in our approximations.

Kai: So, title is "Bridging Filtering and Point-Splitting Approaches for Variable-Density Flows", and they are Narula and Perlekar. It really shows how we can generalize old techniques from uniform density flows to handle the complexity of variable density turbulence.

Mira: It suggests that whatever energy definition you pick in a compressible flow, you can relate it back to a set of multipoint correlation functions. That's a big theoretical step because it provides structure where there was previously just some ambiguity about which correlation functions apply.

Lev: For me, the implication is that if we want to model buoyancy-driven flows accurately on real hardware, we should focus on constructing these multipoint correlation functions based on this framework instead of trying to define the filtered energy budget directly in a way that requires too many terms.

Kai: So, in short, it provides a unified description for how energy moves across scales when density varies. It shows that the Favre filtered contributions can be written as an infinite series of multipoint correlation functions, and they propose using C p approximated by M p as our best tool for studying energy transfers.

Mira: The future work is expected to focus on validating this series expansion across a wider range of physical parameters, particularly looking at how the convergence depends on things like the density ratio in these flows.

Lev: If it converges well, that means we have a robust theoretical handle on the dynamics of energy transfer in these complex environments. It gives us confidence that our simulations are capturing the right physics when we use this method.

Conclusion: Kai: So we're wrapping up this look at "Bridging Filtering and Point-Splitting Approaches for Variable-Density Flows."

Mira: Basically, they take the old ways of looking at turbulence in uniform density flows—using filtering or two-point correlations—and they build a bridge to handle flows where the density changes all over.

Lev: They show how you can figure out those multipoint correlation functions that describe the energy transfer just by expanding the velocity field into powers of those local density fluctuations.

Kai: It suggests that whatever way you define your filtered energy in these compressible flows, it can actually be written as a long series of these multipoint correlation functions.

Mira: And they propose this specific approximation, where one term is about the p-th moment of the field with some filter, which they call C sub p.

Lev: That means we can use that C sub p equals M sub p approximation as a way to study energy transfers in variable density flows rather than needing an infinite number of terms.

Kai: It’s a unified way to look at inter-scale energy transfers across different densities.

Mira: The authors expect this series expansion to converge based on things like the density ratio in the flow.

Lev: If it converges, that gives us a systematic way to capture what’s happening without needing an impossible number of measurements for real hardware.

Kai: It shows how we can use these correlation functions as analogues for point-splitting terms in the Favre filtered energy budget.

Mira: The main implication is that we have a concrete tool for studying how energy moves across scales when density isn't constant.

Lev: Now, if this approximation holds up under real-world conditions, it points us toward what kind of measurements we actually need to make for buoyancy-driven flows.

Hridey Narula, Prasad Perlekar

Tata Institute of Fundamental Research

physics.flu-dyn

Submitted: 2026-01-05

Updated: 2026-07-26

Comments: revised, 8 page, 5 figures

Journal ref: Hridey Narula and Prasad Perlekar 2026 EPL 156 13001

DOI: 10.1209/0295-5075/ae8ea9

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 75/100

The gist: Bridging Filtering and Point-Splitting Approaches for Variable-Density Flows develops a general empirical framework to determine multipoint correlation functions corresponding to any filtered energy…

Key concepts

Favre Filtered Energy
This is the total energy of the flow after applying a specific filtering technique called Favre filtering. It is represented by an infinite sum of different multipoint correlation functions, which helps describe how energy moves across different scales in variable-density turbulence.
Filtering and Point-Splitting Approaches
These are two established methods used to define large-scale energy in uniform flows. The paper connects them by showing their relationship through angular averaging, suggesting that angular averaging acts like a low-pass filter in higher dimensions.
Angular Averaging as Filtering
A specific type of filtering is achieved through angular averaging of a field. This process uses a non-negative, normalized filter function whose support is confined to a specific radius around the point of interest, approximating moments of the filtered fields.

Terminology

Summary

Bridging Filtering and Point-Splitting Approaches for Variable-Density Flows develops a general empirical framework to determine multipoint correlation functions corresponding to any filtered energy definition in compressible turbulence The Favre filtered energy corresponds to an infinite series of multipoint correlation functions achieved by expanding the Favre velocity as a power series in local density fluctuations

Filtering and Point-Splitting Approaches

The paper begins by recalling the filtering and point-splitting approaches in uniform-density flows, where the large-scale energy is defined either through filtering or two-point correlation functions like the KHM relation The two approaches are related by noting that R(r) = R ub(k) / 2 e ik·rd 3k/2 by the WienerKhinchin theorem Angular averaging is shown to be qualitatively the same as low-pass filtering in dimensions d > 1.

Angular Averaging as Filtering

The angular averaging of a field can be interpreted as a low-pass filtering operation where the Fourier transform of the radial delta function gives the equivalent filter This filter, Gδ K(k), is non-negative, normalized, and has its support on the shell of radius r centered at x. The angular-averaged multipoint correlation functions are shown to approximate the moments of the filtered fields by defining Cp(r) = hχ(χ delta K) p−1 i. p. 1

Point-Splitting Analogue to the Favre Filtered Budget

The Favre velocity ueK and the Favre filtered energy E(K) are defined, and they can be formally expanded as a polynomial series under the assumption that the density field is bounded from above The leading order term of this expansion corresponds to the two-point momentum-momentum correlator which has a regularized KHM relation in the infinite Reynolds number limit.

Buoyancy and Pressure Contributions

The Favre filtered energy budget can be expressed as a series of correlation functions by expanding the density-weighted velocity field in powers of local density fluctuations The buoyancy contribution F g(K) is expressed as a series expansion in ψ using eq. (12), involving two, three and four-point correlation functions for terms up to second order in density fluctuations. p. 4 Similarly, the pressure contribution P(K) is expanded into a series where higher-order terms are required for better agreement. p. 5

Conclusion

The study provides a unified way of looking at inter-scale energy transfers in generic variable-density flows by showing that the Favre filtered contributions can be written as the sum of an infinite series of multipoint correlation functions The approximation Cp ≈ Mp is obtained, where Mp is the p-th moment of the filtered fields evaluated with an arbitrary positive filter. This allows for the identification of point-splitting analogues for terms in the Favre filtered energy budget, which is considered the most appropriate choice to study energy transfers in variable-density flows. p. 5 The series expansion is expected to converge based on flow parameters like the density ratio. p. 6

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improved AI system can determine multipoint correlation functions corresponding to any filtered energy definition by expanding Favre velocity as a power series in local density fluctuations. This allows for a general empirical framework applicable to variable-density flows.

  2. Improved AI system can approximate the Favre filtered budget using an infinite series of multipoint correlation functions, specifically showing that Cp approximately equals Mp with Mp being the p-th moment of the filtered field evaluated with an arbitrary positive filter.

  3. Improved AI system can rigorously test energy transfer mechanisms by evaluating buoyancy and pressure contributions in the Favre budget using a series expansion up to second order in density fluctuations, confirming agreement between correlation functions and filtered fields for specific flow types like bubbly flows.

Sources

Related papers