Penguin contributions to CP phases in B_d,s decays to charmonium

Worth your time, with one caveat about the headline number. The real result is the OPE: the up-quark penguin matrix element is written as a sum of local four-quark operator matrix elements with IR-safe Wilson coefficients, up to Λ_QCD/m_ψ. The IR analysis is careful — soft divergences cancel between diagrams, collinear ones vanish by a Ward identity, spectator and long-distance penguins are power-suppressed. That is a standard diagrammatic argument for this subfield, not a theorem, but it is done properly. The correction to Boos–Mannel–Reuter is real: they dropped the LO penguin operator piece, which is why their estimate came out small. The extension to Cabibbo-unsuppressed b → ccd modes gives a genuinely new handle, and the B_d → J/ψ π⁰ prediction favoring Belle over BaBar is a falsifiable claim. The caveat: the 0.68° is not first-principles. Eq. (12) gives |P/T| ≈ 0.78|v8|, and Δφ_d ≈ 1.5°|v8|, so the quoted bound is essentially |v8| < 1/3 dressed with uncertainties. The data constrain only v8 − a8 ∈ [0.07, 0.19]; with |a8| ≤ 1/3 allowed, |v8| could be ~0.5 and the bound ~1°. The body is honest about the ranges, but the abstract and conclusions present 0.68° without the qualification. That is the one thing I would push on in review — either justify |v8| < 1/3 more tightly or say in the abstract that the number is conditional on 1/N_c. Two smaller things: the 'PROOF OF FACTORIZATION' heading oversells a power-counting argument (minor, the content is fine), and the 'higher-order terms in our OPE' uncertainty folded into the numbers is not quantified anywhere (minor). Who it is for: anyone quoting a penguin-pollution uncertainty on 2β or 2β_s, and anyone thinking about b → ccd control channels. The framework is reproducible from the formulas and ranges given. Peer review: yes, send it out. The OPE is the contribution; the numerical bound needs the qualification above, which is a revision issue, not a rejection issue.

The numerical headline rests almost entirely on one modelling input. Eq. (12) gives P/T = 0.01 − 0.02 v0 − (0.71 + 0.33i) v8, so |P/T| ≈ 0.78 |v8| and, via tan(Δφ) ≈ 2ε sinγ Re(P/T), Δφ_d ≈ 1.5° × |v8|. The quoted 0.68° is therefore essentially the statement |v8| < 1/3 dressed with uncertainties. The paper's own branching-ratio analysis (Eq. (8) and the sentence 'one finds the correct branching ratio for 0.07 ≤ |v8−a8| ≤ 0.19 if v0 is set to 1') constrains only the combination v8 − a8, not v8 itself. The additional cap |v8| < 1/3 is a 1/N_c estimate, not a data-driven bound. Since the paper simultaneously allows |a8| ≤ 1/3, the data are consistent with v8 = (v8 − a8) + a8 ≈ 0.19 + 0.33 ≈ 0.52, which would push max |Δφ_d| to roughly 1°. The body is transparent about the ranges used ('We vary v8 and v0 in their allowed ranges |v8|< 1/3 and |v0| = 1 ± 0.15 with the constraint that |a8| ≤ 1/3 must be obeyed'), but the abstract and conclusions present 0.68° as if it were a first-principles or data-driven bound. The same lever controls the B_s → J/ψ φ numbers and the Cabibbo-unsuppressed b → ccd limits in Tabs. I and II. A secondary, smaller gap: the 'higher-order terms in our OPE' uncertainty folded into the quoted numbers is not quantified anywhere, so the size of that systematic cannot be checked. The OPE/factorization argument itself (soft and collinear cancellation, spectator and long-distance power counting) is a standard diagrammatic power-counting argument and is not where the fragility lies; the acknowledged assumptions (perturbative rescattering at q² ~ m_ψ², no J/ψ–ρ⁰ mixing) are stated in the text.

The reader's weakest_assumption correctly identifies the load-bearing input: the 1/N_c ranges |v8| < 1/3, |v0| = 1 ± 0.15, |a8| ≤ 1/3. I agree and would sharpen it quantitatively. Because P/T is dominated by the v8 term in Eq. (12), the bound is essentially linear in |v8|; the branching ratio (Eq. (8)) fixes only v8 − a8, so the data alone permit |v8| up to ~0.5 once |a8| ≤ 1/3 is allowed, which would move the bound toward 1°. This is a real, checkable soft spot in the abstract's headline number, not a matter of taste. The reader's second reason (the 'PROOF OF FACTORIZATION' label overstating rigor) is weaker: the argument in the text is a standard leading-power diagrammatic power-counting proof of the kind accepted in this subfield, and the paper is explicit about its assumptions (perturbative rescattering at q² ~ m_ψ², no J/ψ–ρ⁰ mixing, neglect of electroweak penguins and a0). Those are acknowledged limitations, not new findings. The OPE claim itself is not internally inconsistent and is supported by the diagram-by-diagram analysis in the 'PROOF OF FACTORIZATION' section. The one additional gap worth noting is that the 'higher-order terms in our OPE' uncertainty included in the quoted numbers is never quantified, so its size cannot be independently checked. Neither issue invalidates the central OPE result; both bear on how the numerical bound should be presented. CONDITIONAL is the right verdict, and my read does not move it: the concern reinforces the reader's reason (ii) rather than adding a new failure mode. The concrete test above would settle whether the 0.68° survives relaxation of the |v8| cap, and if it does not, the abstract and conclusions should carry the qualification explicitly.

The result addresses a real and important systematic for the LHCb and Belle II programs: the penguin pollution of the golden modes B_d → J/ψ K_S and B_s → J/ψ φ. The OPE-based approach is a genuine alternative to the standard SU(3)_F method, which cannot be applied to B_s → J/ψ φ and whose accuracy is poorly known. The predictions for b → ccd modes are falsifiable and provide a nontrivial test of the framework. The derivation is a standard diagrammatic power-counting argument rather than a rigorous theorem, which is the norm in this area, and the paper is transparent about the 1/N_c ranges used in the body. The main strengths are the explicit IR-safety analysis and the falsifiable b → ccd predictions; the main weakness is that the abstract presents the numerical bound without the 1/N_c qualification that the body states.

The measurements that pin down the CP-violating phases of $B_d$ and $B_s$ mixing use decays such as $B_d$ to $J/ψ K_S$ and $B_s$ to $J/ψ φ$, whose interpretation is spoiled by a loop diagram in which an up quark replaces the charm quark. That loop is a long-distance QCD object, and until now its size was estimated only by assuming an approximate flavour symmetry among the light quarks. This paper argues that the loop can instead be written, through an operator product expansion, as a sum of local four-quark operators already present in the tree amplitude, with the leftover corrections suppressed by $Λ_{QCD}/m_ψ$. Because the same hadronic parameters then appear in both the tree and the penguin, the measured branching ratios and the standard large-$N_c$ counting of QCD suffice to bound the penguin-induced shift of the phases. The payoff is a numerical ceiling on how far new physics could be hiding behind this pollution: at most $0.68^°$ on $2β$, and about $1^°$ on the three polarization amplitudes of $2β_s$.

What carries the argument. The load-bearing device is the operator product expansion of the up-quark penguin matrix element onto the four local operators $Q_{0V}$, $Q_{0A}$, $Q_{8V}$, $Q_{8A}$ defined in the text, with Wilson coefficients computed from the hard loop momentum $q^2 ∼ m_ψ^2$. Its practical value comes from a second ingredient: the normalization of those matrix elements to the factorized value $V_0 = 2 f_{J/ψ} m_{B_d} p_{cm} F_1^{B→K}(m_ψ^2) = (4.26 ± 0.16)$ GeV$^3$, and the large-$N_c$ counting $v_0 = 1 + O(1/N_c^2)$, $v_8, a_8 = O(1/N_c)$, $a_0 = O(1/N_c^2)$. The branching ratio then reads $B/B_{exp} = [1 ± 0.08] |0.47 v_0 + 7.8 (v_8 - a_8)|^2$, which is satisfied for $0.07 ≤ |v_8 - a_8| ≤ 0.19$ at $v_0 = 1$ — the numerical bridge between the formal expansion and the quoted degree-level bounds.

Core claim. The central claim is an operator identity: $⟨J/ψ K_S|Q^u_j|B_d⟩ = Σ_k C_{j,k} ⟨J/ψ K_S|Q_k|B_d⟩$ plus terms suppressed by $Λ_{QCD}/m_ψ$, where the operators $Q_k$ are the four colour-singlet and colour-octet vector and axial four-quark operators built from $s γ^μ(1-γ_5) b$ times $c γ_μ c$ or $c γ_μ γ_5 c$. The authors argue that this expansion is infrared safe: soft divergences cancel between diagrams or are reproduced on the effective-theory side, collinear divergences cancel by the QCD Ward identity when the gluon attachments are summed, spectator scattering is suppressed by $Λ_{QCD}/m_ψ$, and diagrams in which a hard gluon bypasses the penguin loop are suppressed by three powers of $Λ_{QCD}/√{q^2}$. The leading coefficient is the standard hard-penguin function $P(q^2) = (2/3)(α_s/4π)[ln(q^2/μ^2) - iπ - 2/3]$ evaluated at $q^2 ∼ m_ψ^2$, plus the gluon-penguin contribution proportional to $m_b^2/q^2$. Because the penguin and the tree then share the hadronic parameters $v_0$, $v_8$, $a_8$, the measured $B(B_d → J/ψ K_S)$ fixes one combination and the $1/N_c$ hierarchy fixes the rest, giving $|Δφ_d| ≤ 0.68^°$ for $B_d → J/ψ K_S$, $0.74^°$ for $ψ(2S) K_S$, and $0.85^°$, $1.13^°$, $0.93^°$ for the three $B_d → J/ψ K^*$ polarizations, together with $|Δφ_s| ≤ 0.97^°$, $1.22^°$, $0.99^°$ for the longitudinal, parallel and perpendicular $B_s → J/ψ φ$ amplitudes. The same machinery also covers the Cabibbo-unsuppressed cases, where the penguin-to-tree ratio is not small.

The premise it rests on. The numerical ceilings rest on assumed ranges for the hadronic parameters taken from large-$N_c$ counting — $|v_8| < 1/3$, $|v_0| = 1 ± 0.15$ and $|a_8| ≤ 1/3$ — rather than on a first-principles determination, and the quoted angles grow roughly in proportion to how large $v_8$ is actually allowed to be.

The penguin shift in the $B_d → J/ψ K_S$ determination of $\sin 2β$ is bounded below the current experimental error of about $1.6^°$, so the discrepancy between the charmonium measurement and other determinations of the unitarity-triangle angle cannot be blamed on penguin pollution at a level larger than $0.68^°$.

For $B_s → J/ψ φ$ the three polarization amplitudes get separate ceilings near $1^°$, which means the $B_s$ CP phase can be interpreted as a clean new-physics probe at the precision LHCb is now reaching.

Direct CP asymmetries in these modes are predicted to be small: $|C_f| ≤ 1.33 × 10^{-2}$ for $B_d → J/ψ K_S$ and $≤ 1.92 × 10^{-2}$ for the perpendicular $B_s → J/ψ φ$ amplitude, so a measured asymmetry above a percent would signal physics outside this framework.

The Cabibbo-unsuppressed channels $B_d → J/ψ π^0$, $B_d → J/ψ ρ^0$, $B_s → J/ψ K_S$ and $B_s → J/ψ K^*$ acquire definite predicted ranges, giving a test of the whole approach that the usual flavour-symmetry estimates cannot supply.

The authors note that their $B_d → J/ψ π^0$ ranges sit closer to the Belle values of $C$ and $S$ than to the BaBar values, so improved data on that mode discriminate between the two measurements.

The quoted angles scale roughly linearly with the assumed $|v_8| < 1/3$; a lattice or data-driven determination of $v_8$ that halves the allowed range would tighten every bound in the tables by about a factor of two, which makes $v_8$ the single most valuable target for follow-up work.

The authors flag that the expansion captures all hadronic effects only if the $J/ψ$ has no intrinsic light-quark component, for instance no $J/ψ$–$ρ^0$ mixing; a measured violation of the $b → ccd$ predictions in $B_d → J/ψ π^0$ or $B_s → J/ψ K_S$ would be the natural place for such a component to show up.

Because the same expansion pattern was used for charm-loop effects in $B → K^{(*)} ℓ^+ ℓ^-$, where no four-quark operators appear, the two cases could be cross-checked against each other: a consistent description of both would strengthen confidence in the perturbative treatment of rescattering above the pair-production threshold.

The bounds are stated as conservative because the allowed parameter ranges are wide and even permit cancellations in the branching-ratio formula; a global fit that uses several charmonium modes at once to pin a common $v_8$ would convert these ceilings into sharper, correlated predictions.

This review was written by a machine. No person has reviewed this paper on Pith yet. A scientist listed on Pith can review it for a fee, and their review is published here whatever it concludes.

Request a human review