Daily Summary for 2026-11-09

daily

Transcript

Introduction to the show: ident: High Energy Physics: Phenomenology. The day in particle phenomenology: models, signatures and constraints.

Kai: It's the ninth of November, twenty twenty-six, and this is the day's research.

Mira: 15 new papers came out today.

Kai: I'm Kai.

Mira: And I'm Mira, here with Lev.

Kai: We'll take the day in one pass, then pull out the papers we're staying with.

Mira: Start us off with Phenomenological investigation of order statistics for multi-jet events.

The day's research: Kai: The work investigates how bottom quark loops and annihilation channels modify the invariant mass distribution of Higgs pairs, dsigma dmHH.

Mira: Understanding these effects near the production threshold gives us clues about new physics in the Standard Model.

Lev: While total cross section level is not significantly affected by bottom quarks, their impact near the threshold is notable.

Kai: When using a renormalised bottom quark mass in the MS scheme with a twenty-five giga-electron volt bin size, the differential cross section increases by about six percent.

Mira: Conversely, if an on-shell mass renormalization is used for the bottom quark, this increase near the threshold rises to approximately twelve percent.

Lev: The contribution from bottom quark annihilation at next-to-leading order QCD is small in terms of total cross section, around zero point zero three femtobarns at a fourteen teraelectron volt centre of mass energy.

Kai: Including this channel at NLO QCD leads to relative corrections up to eighteen percent in the invariant mass distribution compared to leading order gluon fusion.

Mira: Including bottom quark annihilation at leading order results in an enhancement of up to eighteen percent at low mHH with respect to the leading order gluon fusion.

Lev: Although this occurs, the peak region and tail of the mHH distribution remain unaffected by this channel.

Kai: Calculations using four flavours versus five flavours show that the difference between them is less than one percent near the Higgs pair production threshold.

Mira: This difference grows as the Higgs pair invariant mass mHH increases.

Lev: The effect of massive bottom quarks on this distribution depends on whether the top quark mass is renormalised in the OS scheme and the bottom quark mass in the MS scheme.

Kai: In general, these schemes yield almost identical results for massive bottom quarks.

Mira: We also have studies looking at how extensions to the Higgs sector affect top-quark pair production and decay.

Lev: In those studies, NLO plus pseudoscalar predictions show an enhanced peak and a deeper dip compared to leading order plus pseudoscalar calculations.

Kai: Meanwhile, in particle physics beyond the Standard Model, there is ongoing work exploring constraints from electroweak precision data using effective field theory analyses that suggest new physics at the teraelectron volt scale.

Mira: The most significant finding is that a single spectator sector first-order phase transition during inflation can consistently link vacuum bubble nucleation, primordial black hole formation, baryogenesis, and gravitational wave production.

Lev: This connection is crucial because it provides a unified mechanism for generating both the observed matter asymmetry and specific gravitational wave signatures from the early universe dynamics.

Kai: The characteristic masses of these primordial black holes are found to be in the range of O105 to O106 grams for representative viable parameter points.

Mira: Investigating bottom quark effects in Higgs boson pair production: This paper examines how bottom quarks influence Higgs boson pair production near the energy threshold.

Lev: Top-Quark Pair Production and Decay in Extended Higgs Sectors: This study computes cross sections for top-quark pair production when extending the Standard Model with extra heavy scalar and pseudoscalar particles.

Kai: SMEFT Constraints on Top Electroweak Couplings and a Composite Higgs Interpretation: This work uses LHC measurements to constrain new physics couplings in an effective field theory approach and proposes a composite Higgs model.

Mira: Unitary KK and K pi inputs for an amplitude analysis of the D+ to K-K+ pi+ decay: The paper develops a dynamic model to analyze CP violation in charm decays by including final state interactions.

Lev: Constraining tau-sterile neutrino mixing with high-energy astrophysical neutrinos: This research uses astrophysical neutrino data to set limits on how much a sterile neutrino might mix with the tau neutrino flavor.

Kai: Joint Bayesian Inference for Extraction of Charmonium Long Distance Matrix Elements Using PYTHIA: This paper uses a joint Bayesian method to extract long-distance matrix elements for charmonium states from LHC data.

Mira: Complete set of bounded-from-below conditions for the A 4-symmetric three-Higgs-doublet model: The authors derive necessary and sufficient conditions for the scalar potential of an A4-invariant three Higgs doublet model to be bounded from below.

Lev: Signatures of Neutrinophilic Pion Dark Matter: This paper explores how pion dark matter coupled to neutrinos can reproduce neutrino anomalies through semi-annihilation processes.

Kai: Light Primordial Black Holes from a First-Order Phase Transition near the End of Inflation: This work connects light primordial black hole formation during inflation to baryogenesis and gravitational wave signatures.

Mira: Probing CP Violation in the Complex 2HDMS with a 95 GeV Scalar at Future e+e- Colliders: The study investigates how future electron-positron colliders can probe CP violation in an extended two-Higgs-doublet model.

Lev: Optimal-Transport Contrastive Learning for Centrality Determination in Heavy-Ion Collisions: This paper proposes using optimal transport and contrastive learning to determine the centrality of heavy-ion collisions.

Kai: Power corrections to heavy-quark spin correlations in small x deep-inelastic scattering: The calculation shows that power corrections restore target dependence of the azimuthally averaged transverse heavy-quark spin state in small x DIS.

Mira: Phenomenological investigation of order statistics for multi-jet events: This study uses order statistics on jet rapidity distributions to compare predictions from different high-energy radiation generators.

Lev: The Colour Glass Condensate: This paper provides a comprehensive introduction to the Colour Glass Condensate effective theory for hadronic interactions at high parton densities.

Kai: Investigating bottom quark effects in Higgs boson pair production: This paper examines how bottom quarks influence Higgs boson pair production near the energy threshold.

Lucky paper: 2611.07838: Kai: So we're diving into this paper now, which is titled Phenomenological investigation of order statistics for multi-jet events. It looks like it’s a deep dive into using rank-ordered jet rapidity distributions to look at multijet radiation in proton-proton collisions at twenty three teraelectron volt center of mass energy.

Mira: Right, so the core idea here is moving beyond just looking at single inclusive jet spectra and instead using something that captures the organization of jets within an entire event.

Lev: The theoretical framework assumes independent and identically distributed rapidities as an idealization when setting up this order statistics approach, which leads to a binomial upper-tail probability formula for the cumulative distribution of the k-th order statistic.

Kai: That sounds like it sets up a new way to get more exclusive information about multijet radiation compared to just the inclusive single-jet rapidity spectrum, as mentioned in the paper.

Mira: The study then compares predictions from two different generators: the BFKL based generator called BFKLex and the collinear parton shower generator PYTHIA eight point three zero six, using total variation distance and Wasserstein distance to quantify those differences.

Lev: The results show that these differences are still visible relative to the uncertainty bands in several well-populated regions when you look at the shape distances calculated for jet one, jet two, and jet three at a multiplicity of n equals three.

Kai: When you look at the baseline distances in Table one, for jet one with n equals three, total variation distance is zero point one two one and Wasserstein distance is zero point four six seven.

Mira: If we require a minimum rapidity separation between the extremal jets to be at least three units of rapidity, that is denoted as delta YMF MB greater than or equal to three, this requirement modifies the discriminating power in a rank-dependent way.

Lev: Requiring that minimum separation actually reduces both the total variation distance and the Wasserstein distance for both the extremal and all three ranks when n equals three.

Kai: That makes sense, so that gap requirement selectively enhances the separation by focusing on specific ranks, especially identifying the second-most-extreme ranks as particularly useful observables.

Mira: The paper suggests that among all observables defined across different jet multiplicities, SMB and SMF provide greater separation between the two generator predictions.

Lev: Table five confirms this hierarchy, showing that while MB and MF have relatively small shape differences between the generators, SMB and SMF offer a larger gap in their predictions.

Kai: The study also assesses sensitivity to PDF variations for PYTHIA eight point three zero six and scale variations for BFKLex, combining those using quadrature to get the total asymmetric uncertainty.

Mira: And they found that the separation between those two predictions depends on both the rapidity rank and the event selection criteria used in the analysis.

Lev: Specifically, requiring a rapidity gap of at least three units, delta YMF MB greater than or equal to three, enhances that separation selectively by identifying those second-most-extreme ranks as particularly useful observables for this selection.

Kai: It's interesting that they also checked the infrared cutoff variations using lambda IR equals one giga-electron volt and found that the perturbative scale and PDF set sensitivity is substantially greater than what you see from the infrared cutoff.

Mira: So even when you check those low energy cutoffs, it seems like the uncertainties tied to pdf input and alpha s treatment are driving most of the variation in separation.

Lev: The second-extremal ranks, especially when using that three unit separation requirement, are more sensitive to variations involving mu R than other ranks.

Kai: Experimentally, they say you only need reconstructed jet rapidities and event-by-event ordering to use these observables, but they flag that assessing the achievable precision requires a dedicated study of things like trigger efficiencies and detector response.

Mira: So it's a solid phenomenological foundation for measuring these rank distributions, motivating their inclusion in future analyses of multijet production experiments.

Lucky paper: 2611.08388: Kai: We're looking at Top-Quark Pair Production and Decay in Extended Higgs Sectors today, which is the paper by Banfi et al., and it deals with how we can search for new physics using the top quark pair.

Mira: It seems like this work tackles a big question about the Standard Model extensions, specifically regarding that first-order electroweak phase transition that we haven't fully understood yet.

Lev: The two-Higgs-doublet model is the minimal extension, adding an extra complex SU2L doublet which introduces five new physical particles: one additional scalar, one pseudoscalar, and a pair of charged Higgs bosons.

Kai: These heavy neutral Higgs bosons have masses above the top quark pair production threshold because they interact strongly with the top quark due to that large top-Yukawa coupling in the Standard Model.

Mira: The computational framework they use involves calculating cross sections at next-to-leading order, including leptonic decays of the top quark with full spin correlations.

Lev: They implement this in a modified version of the hvq library within POWHEGBOX V2, which is interfaced with OpenLoops two to handle the calculations.

Kai: The core part of their analysis looks at the total squared amplitude for top-quark pair production where it could result in a Standard Model Higgs, or one of those extra heavy scalar or pseudoscalar particles.

Mira: They focus on the signal structure analysis by looking at top-pair production followed by leptonic W-decays, pp to tt bar plus X, and how that interferes with the QCD background.

Lev: Specifically, the interference between the pseudoscalar signal and the QCD background has both positive and negative contributions that accentuate a peak and a dip compared to leading order predictions.

Kai: That means even though the inclusive K-factor isn't huge, this interference structure is what makes their tool useful for experimental searches.

Mira: Including those full spin correlations in the top-quark decays lets them study differential cross sections as functions of spin sensitive observables, which helps probe these higher-order effects on the shapes of those cross sections.

Lev: The paper also notes that this method enables researchers to see the nuanced peak and dip interference structure more clearly than just looking at inclusive results.

Kai: So, when you look at their numerical results, they show that NLO plus pseudoscalar predictions have an enhanced peak and a deeper dip compared to leading order plus pseudoscalar.

Mira: That suggests that higher-order corrections really do accentuate the interference structure even when the overall K-factor is relatively small.

Lev: The computational framework is quite involved because it needs to correctly handle those NLO QCD corrections while simultaneously tracking the spin correlations in the top quark decays.

Kai: What are you seeing on the ground regarding what these results change for someone who just listens to this?

Mira: For an experimentalist, this tool gives them a way to look beyond simple inclusive measurements and check for that specific interference pattern which might point toward new physics in those extended Higgs sectors.

Lev: It’s important because it allows for the study of differential cross sections as functions of spin sensitive observables, not just the total rate.

Kai: Meng, from a production standpoint, how much does implementing this full spin correlation analysis add to the computational load for running models at scale?

Meng: From a production perspective, calculating those full spin correlations adds significant overhead because you have to track every possible polarization state in the decay chain.

Mira: That makes sense when you think about deploying this on real traffic; you want something that runs efficiently without needing a supercomputer just for the correlation matrix.

Lev: The authors note that while they implement these effects, their tool enables the study of higher-order effects on shapes of differential cross sections, especially those nuanced peak and dip interference structures.

Kai: So, the main limitation you're pointing out is that this method struggles when realistic kinematic cuts are applied because distinct features between central and peripheral collisions get lost.

Lev: That is a real caveat; the unsupervised method using optimal transport successfully clusters events based on particle momentum distributions, but it loses those subtle differences between central and peripheral collisions when you impose realistic cuts.

Mira: So the paper suggests that while this mapping technique is good for general event classification, you still need to be careful when trying to extract very specific features related to collision centrality.

Lev: They are essentially saying that the method works well for clustering events based on Npart, but applying kinematic cuts can smear out those distinct features between central and peripheral collisions.

Kai: It sounds like the power corrections to heavy-quark spin correlations in small x deep-inelastic scattering are another area where model dependence comes into play, right?

Lev: Exactly; they calculate the nonvanishing corrections to heavy quark spin correlations in small x deep inelastic scattering using a model dependent approach, and that part is not entirely fixed.

Mira: If you're building a production model for a collider, you have to be very careful about how much you rely on those specific power corrections when trying to predict the shape of the distribution.

Lev: The paper shows that these corrections are necessary because they are nonvanishing, but it emphasizes that the approach remains model dependent in its final result.

Kai: And what does this mean for us when we look at constraints from future e plus minus colliders probing CP violation in extended Higgs sectors?

Mira: It suggests that while the structure of the interference is clear theoretically, extracting a precise measurement depends heavily on how well you model those underlying spin correlation effects and power corrections.

Lev: The study shows that the tool itself is very useful for experimental searches because it consistently implements those interference effects between scalar/pseudoscalar signals and the QCD background.

Kai: So, to wrap up, this paper gives us a way to calculate these complex interference patterns in top-quark pair production, but you have to be aware of how those underlying spin correlations are modeled.

Lucky paper: 2611.04915: Kai: We're looking at a paper today titled Unitary KK and K pi inputs for an amplitude analysis of the D+ to K-K+ pi+.

Mira: It deals with CP violation in three-body charm decays, specifically D+ to K-K+ pi+.

Lev: The core idea is that since charge parity violation is expected in these three-body decays, we need an amplitude analysis to properly understand the final state interactions.

Kai: So the authors are trying to separate the weak vertex from those final state interactions, which they model using a kernel K for SU(three) meson-meson interactions.

Mira: They factorize the weak vertex into leading order topologies like color allowed and color suppressed diagrams.

Lev: They parametrize these weak vertices using their CKM content multiplied by a polynomial with some free parameters that they fit to data, which is described in equation three of the paper.

Kai: The final state interaction formalism departs from extending resonant chiral perturbation theory to describe K-K and K-pi interactions.

Mira: They define this kernel K as the sum of a contact interaction and resonance exchange, which lets them introduce resonances into the model description.

Lev: As a proof of concept, they fitted a toy dataset from the S-wave analysis of D+ to K-K+ pi+ by CLEO-c using this new model.

Kai: What is interesting here is how they are handling the isospin decomposition when modeling these meson interactions.

Mira: The paper specifically mentions that the description of the final state interaction expansion will be used to introduce resonances and contact interactions in the model.

Lev: They use a toy dataset to validate this approach before applying it to real data, which gives us some initial confidence in the methodology.

Kai: The paper focuses on modeling CP violation in these decays because while CPV is expected, it hasn't been observed yet in charm three-body decays.

Mira: The authors suggest that a full amplitude analysis might be more sensitive to localized CPV effects within intermediate amplitudes, even if the integrated asymmetry looks low.

Lev: They are looking at how the CKM elements affect the different vertices when they try to fit this model against existing data.

Kai: How does this approach change things for someone just listening? It suggests that instead of looking at just one total decay rate, you need to look at the specific pathways of how those particles interact after the decay happens.

Mira: Right. And it points out that they are using a toy dataset for validation first, which is good because it means their method is working on something concrete before they move to more complex real data sets.

Lev: The paper suggests that by parameterizing the weak vertices with CKM content multiplied by a polynomial, they can account for different CKM elements in the vertex terms.

Kai: So, if we look at the numbers, what are some specific values from their toy fitting process?

Mira: They fit their S-wave model against that toy dataset and the resulting parameters allow them to see how much data is needed to constrain the free parameters.

Lev: The study shows that this amplitude analysis framework successfully captures the structure of the decay, even with a relatively small training set for validation.

Kai: I wonder if this method is scalable for analyzing much larger datasets like those from BESIII or future experiments.

Mira: The authors imply that it can be applied to real data, but they also point out that when they apply realistic kinematic cuts, the distinct features between central and peripheral collisions might get lost.

Lev: This means the method is unsupervised in terms of centrality determination because it relies on particle momentum distributions rather than explicit event shape cuts.

Kai: So it's using optimal transport to map heavy-ion collision events onto centrality classes, which is a different kind of mapping than what we see here.

Mira: Yes, that’s the other major piece of work they are doing, Optimal-Transport Contrastive Learning for Centrality Determination in Heavy-Ion Collisions.

Lev: That method uses the Wasserstein distance to fixed landmarks to compare the full final state momentum distributions across events.

Kai: And then it compresses those distances into a compact event embedding using contrastive learning, which is unsupervised clustering based on similarity.

Mira: While this approach successfully clusters events with similar number of participating nucleons into the same centrality class, it struggles when realistic kinematic cuts are applied because features between central and peripheral collisions get lost.

Lev: So the limitation they state is that applying realistic kinematic cuts doesn't always preserve those distinct features they are trying to map out.

Kai: That makes sense from an engineering standpoint; if you cut too aggressively, you might just lose the subtle physics differences you’re trying to isolate.

Mira: Exactly, and it shows that while the unsupervised method is good for initial clustering, it isn't a direct replacement for traditional slicing methods when precise kinematic boundaries are essential.

Lev: We also have work on constraining tau-sterile neutrino mixing using high-energy astrophysical neutrinos which explores how cosmological distances affect flavor composition measurements.

Kai: That shifts the focus from particle decay kinematics to cosmology and astrophysics, looking at how distant signals constrain new particle physics parameters.

Mira: This paper uses astrophysical neutrino data to set limits on the mixing angle of a sterile neutrino with the tau neutrino flavor based on cosmological distances.

Lev: The constraints they put on this mixing parameter are quite tight because they are leveraging high-energy astrophysical events from sources like active galactic nuclei.

Kai: It’s interesting how different areas of particle physics, from charm decay amplitudes to cosmology, seem to be using similar data analysis techniques in different ways.

Mira: That's the beauty of it; the underlying statistical machinery for fitting parameters and extracting signals is often shared across these diverse fields.

Lev: In summary, the paper on Unitary KK and K pi inputs provides a detailed amplitude model that uses final state interactions to better understand CP violation in charm decays by fitting a toy dataset.

Kai: And we've covered how that compares to the machine learning approach for centrality determination in heavy-ion collisions.

Mira: We also touched on the cosmological constraints from astrophysical neutrinos and the flavor physics work involving bottom quark loops and Higgs pair production.

Lev: Those papers show how different theoretical tools, from chiral perturbation theory extensions to Bayesian inference, are all trying to pinpoint specific physical effects in high energy environments.

Lucky paper: 2611.05170: Kai: We're looking at this paper, Constraining tau-sterile neutrino mixing with high-energy astrophysical neutrinos, and the core question is how we can use these high energy neutrinos to test a mixing parameter that terrestrial experiments struggle to constrain.

Mira: It seems like they are focusing on the tau neutrino mixing, denoted as Utau42, because that specific channel is less constrained by current experiments than the electron or muon mixing.

Lev: They argue that TeV to PeV astrophysical neutrinos offer a way to test this mixing parameter independently of terrestrial experiments and supernova data, which holds true for every value of delta m squared forty-one.

Kai: The paper explains that standard neutrino oscillations predict about one third of the neutrinos reaching Earth are tau neutrinos even though the sources make almost none.

Mira: If there is non-standard mixing with a sterile neutrino, this would make those tau neutrinos scarcer, which lowers their fraction at Earth, and this is only detectable because electron and tau signals look similar in neutrino telescopes.

Lev: They fit the flavor composition measured by IceCube with the sources left free, finding that Utau42 is less than zero point eight eight at a ninety percent confidence level.

Kai: Then they project that combining IceCube with planned neutrino telescopes could lower this limit down to zero point two four to zero point two nine by the year two thousand forty and further down to zero point one two to zero point one four by the year two thousand fifty.

Mira: By the year two thousand fifty these projected limits would surpass those of Super-K and MINOS, but they wouldn't reach the strongest terrestrial limits available today.

Lev: A discovery would require Utau42 to be at least zero point one four to zero point two three by the year two thousand fifty for pion-decay sources, or zero point one four for muon-damped ones, which is where it starts pointing toward a smaller delta m squared forty-one or tension with terrestrial data above delta m squared forty-one equal to three times ten to the minus three electron volt squared.

Kai: The paper points out that the best fit from IceCube MESE two thousand twenty-five gives Utau42 as zero point zero nine, which is only a delta chi two of zero point zero three away from standard mixing parameters.

Mira: That means the present data actually agree with standard mixing, but the limit is weak because MESE allows for a wide range of that tau neutrino fraction, ftau plus minus zero point thirty-three plus or minus zero point twenty-five.

Lev: They show that only sterile mixing that removes most of those tau neutrinos is disfavored when Utau42 is set to zero point eight eight, where the fraction drops from zero point three three down to zero point zero six for pion-decay sources.

Kai: The authors emphasize that the main projection for the year two thousand fifty combination could discover sterile mixing with tau neutrino if Utau42 is at least zero point one four to zero point two three for pion-decay sources or zero point one four for muon-damped ones, rejecting standard mixing at three sigma in the median.

Mira: They also mention that flavor measurements constrain the other two active sterile mixings much less than Utau42, and they set no limit on Ue42 because the free source composition absorbs its effect.

Lev: The conclusion is that by two thousand fifty these limits would surpass Super-K and MINOS but not reach the strongest terrestrial limits, and a discovery would require a value of Utau42 greater than or equal to zero point one four to zero point two three for pion-decay sources by then.

Kai: The paper suggests that the first point, which is better separation of electron from tau neutrinos in neutrino telescopes due to the electron tau degeneracy, is key for improving precision on that fraction.

Mira: And the second point they make is better knowledge of the sources themselves, which helps constrain what we are actually looking at.

Lev: The core idea here is that this astrophysical probe provides a test of Utau42 that isn't dependent on terrestrial experiments or supernovae, and it holds true regardless of the value of delta m squared forty-one.

Lucky paper: 2611.07465: Kai: Okay, we're looking at the paper now, Optimal-Transport Contrastive Learning for Centrality Determination in Heavy-Ion Collisions. This one is all about getting high resolution centrality classes for event-by-event analyses, especially for net proton fluctuations aiming at the QCD critical point.

Mira: It uses an unsupervised method that relies on optimal transport to map final particle momentum distributions onto a set of fixed landmark events, and then contrastive learning compresses those distances into a compact event embedding.

Lev: The core mechanic involves calculating the one-Wasserstein distance between every event and these landmarks using the Hungarian algorithm, which gives us these one-Wasserstein distances for each event.

Kai: And then they train a point-cloud network where the objective is that this embedding should reproduce the geometry of the transport feature space, meaning nearest neighbors in that feature space are set as positives to minimize info inception loss.

Mira: They mention training on only the first two thousand events with Npart greater than zero as their training sample, which gives us some context on how they're setting up this contrastive learning objective.

Lev: When they compare their OT-contrastive learning method against conventional methods like Nch slicing, they found that the 'OT all' approach, which uses all particles, performed the best.

Kai: But there’s a catch there because when they apply realistic kinematic cuts to test it, the distinct features in momentum distribution between central and peripheral collisions get lost.

Mira: That loss of distinct features seems like a major bottleneck for applying this method where you need fine resolution for analyzing things like net proton fluctuations.

Lev: The performance is quantified by centrality resolution, which they define as the ratio of the standard deviation to the mean of the Npart distribution in each class, and they show 'OT all' has a better resolution than Nch slicing.

Kai: Conversely, they noted that 'OT charged eta' has the worst centrality resolution compared to conventional methods and even worse than Nch slicing because applying kinematic cuts destroys those distinct features.

Mira: It sounds like the method is powerful when it can use all particles, but its practical application in a real experiment with necessary cuts is limited by this feature loss.

Lev: The authors attribute this degradation to the loss of distinct features between central and peripheral collisions specifically because of applying kinematic cuts to the charged particles only.

Kai: So, if we're aiming for event-by-event analyses, like locating the QCD critical point, what does this mean for someone who only listens to us?

Mira: It means that while this OT-contrastive learning approach can extract some information about collision geometry even when using only charged particles with a realistic eta cut, they still need more work to apply it broadly.

Lev: Future work needs to focus on mitigating that loss of distinct features between collision classes because that seems to be the main roadblock they identified.

Kai: Alright, so we've seen how this unsupervised method works and what its limitations are regarding kinematic cuts. We've heard from Kai, Mira, Lev, and now Lu, Meng, Lalam.

More episodes

← Home