{"id":"03f03759-06ce-44dc-9e85-00b1e09948c3","arxiv_id":"1908.03212","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For two interacting scalar fields, low-energy ghosts are equivalent to tachyonic instabilities and high-energy ghosts to gradient instabilities, with rare cases where interactions stabilize two ghosts.","lead":"This paper studies instabilities in cosmological models with a dynamical dark energy field, using two interacting scalar fields as a stand-in for dark energy and matter. It shows that so-called ghost, gradient, and tachyon instabilities are different mathematical masks of the same physical problem, and that interactions between fields can sometimes stabilize seemingly unstable systems.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Two-field 'generic' equivalence is demonstrated only for frozen coefficients; the single-field section itself concedes that time-dependent canonical transformations can remove a negative kinetic term, so the abstract-level claim needs qualification or proof.","rationale":"This is the weakest point because the paper's headline result is a 'generic' statement, but the proof is a set of frozen-coefficient examples and the authors themselves flag the time-dependence loophole in the single-field case. The diagonal-form premise (eq. 24) is secondary; even granting it, the sign-shuffling argument only shows equivalence for constant coefficients. Time-dependent canonical transformations can change the Hamiltonian by more than a boundary term (eq. 17), so stability classification can be basis-dependent on a cosmological background. The proposed test either closes the gap by proving the equivalence for all time-dependent Sp(4) transformations or forces a qualification of the abstract and Section V. This supports the reader's conditional verdict; I do not see grounds to reject the paper, and the worked examples remain valid.","tokens_in":116,"tokens_out":16719,"duration_ms":466435,"concrete_test":"Re-derive Example I (Sec. IV B) allowing K_i(t), M_i^2(t), D(t) and time-dependent canonical coefficients a_i(t), b_i(t), c_i(t), d_i(t) satisfying (35). Keep the d/dt terms in the Legendre transform (the analog of eq. (17)) instead of discarding them, and take k→0. Determine whether an initial K_1<0, K_2>0, M_i^2>0 configuration can be mapped to a fully positive K'_3, K'_4, M'_3^2, M'_4^2. If yes, the low-energy ghost–tachyon equivalence is not generic; if no such time-dependent Sp(4) transformation exists, the generic claim survives.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—that low-energy ghosts are equivalent to tachyons and high-energy ghosts to gradients in two-field systems—is established in Sec. IV only under constant coefficient assumptions. After eq. (17) the authors explicitly note that a time-dependent canonical transformation can remove a single negative kinetic term, making the solution 'expected to be stable'. That caveat is not incorporated into the two-field 'generic' argument: the transformation coefficients in Sec. IV A are assumed constant in time, the equations of motion in eq. (25) are constant-coefficient, and terms of the d/dt type in eq. (17) are dropped in the Legendre transform. In cosmology, coefficients in eq. (24) evolve on a Hubble timescale; for k→0 these time-derivative contributions are not negligible. A time-dependent canonical transformation can map a growing ghost mode onto an oscillatory field, so the claimed equivalence is not invariant under the most natural class of transformations for a time-dependent background. Constant-coefficient examples do not establish the generic statement.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies cosmological models with a dynamical dark energy field and analyzes linear scalar perturbations described by two interacting scalar fields. It classifies instabilities as ghosts, gradient instabilities, and tachyons according to the signs of the kinetic, spatial-gradient, and mass coefficients in the quadratic action. Using linear canonical transformations, the paper argues that low-energy ghosts are equivalent to tachyonic instabilities and high-energy ghosts are equivalent to gradient instabilities. It also claims that interactions between the two fields generally do not determine whether an instability is present, but do affect the instability timescale, with exceptional cases in which two ghosts can interact and lead to stable solutions. The conclusion is that stability conditions in modified gravity dark energy models should be imposed on invariant combinations of coefficients rather than on individual signs.","tokens_in":21034,"tokens_out":12252,"duration_ms":122799,"significance":"If the claimed equivalence were established in full generality, this would be a useful conceptual clarification for dark energy model building, where separate positivity conditions on kinetic and gradient terms are commonly imposed. The single-field section is clear and correct, and the constant-coefficient examples in the two-field section are explicit and internally consistent. The observation that the sign of the antisymmetric coupling D does not change the inertia of the Hamiltonian, while its magnitude can affect growth timescales, is a genuinely useful point. However, the paper's advertised 'generic' result is supported only by constant-coefficient examples with particular transformation choices and with D independent of k. Since the cosmological background makes the coefficients time-dependent, the domain of validity of the central claim remains unclear. The manuscript is therefore a potentially valuable contribution, but it needs either additional analysis for the time-dependent case or a careful restriction of its claims.","major_comments":[{"comment":"The central claim is stated generically in the abstract and in §V, but the two-field derivation assumes all coefficients and transformation parameters are constant in time. In §IV.A the C coefficients are explicitly assumed to be constants in time, and the assumptions before §IV.B fix D to be constant in time and independent of k; the equations of motion (25) and the solutions (26) are constant-coefficient results. This matters physically because §III.B explicitly notes that a time-dependent canonical transformation can remove a single negative kinetic term and produce a stable solution. In a FLRW background K_i, M_i^2 and D evolve on a Hubble timescale, and time derivatives of the transformation coefficients generate additional terms, as shown in Appendix 2 where \\dot{D}_1 appears as a mass mixing term. The paper therefore does not establish the 'generic' equivalence for the time-dependent coefficients that occur in cosmological models. The authors should either supply an adiabatic or otherwise general proof, or explicitly restrict the claim to time-independent coefficients and adjust the abstract and discussion accordingly.","section":"§IV.A; §III.B, after Eq. (17)"},{"comment":"The derivation relies on the assertion that any two-field quadratic action can be brought to the form (24) with only an antisymmetric coupling D by a field redefinition. This is a nontrivial statement when coefficients are time-dependent, since the simultaneous removal of kinetic and mass mixings introduces time-derivative terms and may change the form of D. The paper cites reference [37] for this reduction, but does not state the precise conditions under which (24) is valid. Because the entire subsequent analysis uses (24), the domain of the claimed equivalence is not fully specified. A derivation, or at least a precise statement of the class of theories for which (24) holds, is needed to support the central claim.","section":"§IV, Eq. (24)"},{"comment":"The examples demonstrate the claimed relations for specific sign patterns and for specific choices of canonical-transformation parameters, but they do not add up to the 'generic' statement in the abstract. For instance, §IV.B.1 says that if the ghost is in K_2 instead of K_1, the same transformation leaves a ghost and produces no tachyon; the paper does not give the general construction that maps an arbitrary sign assignment to a tachyon or gradient. In addition, the high-energy results assume D independent of k, whereas the earlier discussion around Eq. (28) explicitly allows D proportional to k. If D is relevant in the high-energy limit, the transformation used in §IV.B.2 and the resulting identification of a gradient instability are not shown to hold. The authors should either prove the equivalences for the full parameter space or moderate the claim from 'generic' to 'shown in representative examples'.","section":"§IV.B, §IV.C"},{"comment":"The claim that a two-ghost system can be stable because, after diagonalization, the positive- and negative-energy oscillators are decoupled is correct within the constant-coefficient linear analysis. However, the physical status of this stability is left unclear: the negative-energy sector is decoupled only in the chosen frame, and any small coupling to the environment or to the other field beyond linear order would make the negative-energy oscillator unstable. Since the paper presents this as one of its main results, it should state more carefully that the stability is a feature of the decoupled linear system in a special basis, and discuss whether it survives the couplings that are inevitably present in a cosmological setting, rather than presenting it as a generic stabilization mechanism.","section":"§IV.D (Example III)"}],"minor_comments":[{"comment":"The lowercase k1 and k2 are used for the kinetic coefficients K1 and K2, which is confusing because k is already the wavenumber; please use K1 and K2 or define the shorthand explicitly.","section":"Appendix A.1, Eq. (A.2)"},{"comment":"In the D' coupling term, φ_2 and φ_1 appear without primes; they should be φ'_2 and φ'_1.","section":"§IV.C, Eq. (44)"},{"comment":"The transformed fields are named φ'_1 and φ'_2, but the low-energy coefficients in Eq. (46) are labelled K'_3, M'_3, K'_4, M'_4; the labelling should be made consistent.","section":"§IV.C, after Eq. (44)"},{"comment":"The sentence stating that a ghost in K2 instead of K1 'would have kept one ghost in φ'_3 and would exhibit no tachyons' is confusing, because by relabelling the two fields the same construction would produce a tachyon; please clarify that a different transformation is required or that the result follows by symmetry.","section":"§IV.B, Eq. (42)"},{"comment":"The statement that the action (9) can absorb gradient terms into M(t,k) should be made more explicit, since a reader may wonder how a gradient instability is represented in the K,M notation used there.","section":"§II.D, Eq. (9)"},{"comment":"The factor (2K1K2)^{-1/2} is imaginary for K1K2<0; the paper should state that the expression is to be interpreted by analytic continuation when discussing sign patterns with negative kinetic coefficients.","section":"§IV, Eq. (26)"}],"recommendation":"major_revision","confidential_remarks":"The paper makes a useful point but the advertised 'generic' result goes beyond what is demonstrated. The constant-coefficient algebra appears correct and the topic is well suited to the journal, but the time-dependence issue is load-bearing because cosmological coefficients evolve on the Hubble timescale. I would encourage a revision that either proves the time-dependent case, perhaps in an adiabatic limit, or carefully restricts the claims in the abstract and discussion. The self-citation to [37] for the general action form is appropriate and does not seem to hide the central result."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a useful extension of the known single-field ghost/tachyon/gradient equivalence to two interacting fields, and the stable two-ghost example is worth knowing. But it is not the 'generic' proof the abstract advertises, and the time-dependence problem is real.\n\nThe new material is the two-field generalization. Starting from the standard quadratic action for two scalar perturbations, eq. (24), with an antisymmetric coupling D, the paper uses canonical transformations and field redefinitions to show that a low-energy ghost in one field can be rewritten as a tachyon and a high-energy ghost as a gradient. That is a natural but non-trivial extension of [44]. The examples are worked out with explicit transformed Lagrangians, the low/high-energy limits are clear, and the point that D affects instability timescales but usually not their presence is a practical takeaway for model builders. The stable two-ghost case is the most interesting observation: with two negative kinetic terms, a large antisymmetric coupling can decouple the fields into one positive-energy and one negative-energy oscillator with oscillatory solutions.\n\nThe soft spot is the word 'generic.' The two-field derivation assumes constant coefficients in the action, the canonical transformation, and the field redefinition. The equations of motion, the transformed Lagrangians, and the sign analyses are all constant-coefficient. In a cosmological background, K_i, M_i, and D evolve on a Hubble timescale, and the d/dt terms dropped in the Legendre transform are not negligible for k→0. The single-field section itself notes that a time-dependent canonical transformation can remove a negative kinetic term entirely, which undercuts the unqualified equivalence in the abstract; that caveat is not carried into the two-field argument. The stress-test note lands. Also, the 'stable' two-ghost example is low-energy only: with D assumed constant in k, the high-energy limit still has two ghosts. The abstract should either restrict the claims to constant coefficients or add the time-dependent analysis.\n\nThe citation pattern is fine: the action form is justified by [37], which is one author's own work but contains the general proof, and the single-field equivalence is credited to [44]. No circularity.\n\nWho should read it: people building or testing dark energy and modified gravity models and imposing stability conditions. It deserves a serious referee, but the referee should push for a qualified abstract and a treatment of the time-dependent case. My recommendation: send to peer review, not desk reject, with major revision expected.","headline":"Useful two-field extension of the ghost/tachyon/gradient equivalence, but the 'generic' claim holds only for constant coefficients and the abstract overreaches.","tokens_in":21501,"tokens_out":9593,"would_cite":true,"duration_ms":90743,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["04.50.Kd","95.36.+x","98.80.-k"],"model":"deepseek-v4-flash","headline":"Low-energy ghosts are tachyons; high-energy ghosts are gradient instabilities in interacting two-field dark energy models.","keywords":["dark energy","cosmological perturbations","ghost instabilities","gradient instabilities","tachyonic instabilities","canonical transformations","two-field actions","stability"],"falsifier":"Numerically evolve the linear scalar perturbations of a concrete two-field dark energy model (for instance a scalar-tensor theory) at low and high energies in two different field bases; the paper predicts identical growth timescales and the same instability structure in every basis, so finding bases that disagree on whether the system is stable or on the growth rate would refute the equivalence claim.","tokens_in":20675,"feed_emoji":"👻","tokens_out":10409,"duration_ms":96935,"temperature":0.7,"pith_summary":"This paper studies the stability of cosmological models in which dark energy is a dynamical field rather than a cosmological constant, focusing on the linear perturbations of a dark-energy field interacting with a matter field. Its central claim is that the three standard instabilities—ghosts (negative kinetic energy), gradient instabilities (negative momentum squared), and tachyons (negative mass squared)—are not independent in a two-interacting-field system. Working with a generic quadratic action, the authors show by canonical transformations that a low-energy ghost is the same physical instability as a tachyon, and a high-energy ghost is the same as a gradient instability; only the mathematical basis changes. The significance is that stability conditions for dark energy models should be imposed on invariant combinations of the action coefficients, not on the sign of any single term in a given field basis. Interactions between the fields generally do not determine whether an instability exists, but they set its growth timescale, and in exceptional two-ghost cases they can even make the system stable.","feed_headline":"Low-energy ghosts are tachyons; high-energy ghosts are gradients","feed_subtitle":"Stability checks on dark energy models should not depend on the basis you write the fields in.","key_machinery":"The machinery is the linear canonical transformation of the two-field Hamiltonian, $Q_i' = a_i Q_i + a_{i+1} P_i$, $P_i' = b_i Q_i + b_{i+1} P_i$ with unit Poisson brackets, combined with a linear field redefinition chosen to diagonalise the transformed kinetic and mass matrices. The key object being manipulated is the quadratic action of eq. (24), whose off-diagonal term is the antisymmetric coupling $D(\\dot\\phi_1 \\phi_2 - \\dot\\phi_2 \\phi_1)$. Because canonical transformations preserve the physics, a negative coefficient that appears as a ghost in one basis must reappear as a tachyon at low $k$ or a gradient at high $k$ in another basis; the transformation moves the sign rather than creating or destroying it.","core_discovery":"The paper derives, for a system of two interacting scalar perturbations of the form $S = \\int d^3k\\, dt \\left[ \\frac{1}{2} K_1 \\dot\\phi_1^2 - \\frac{1}{2} M_1^2 \\phi_1^2 + D(\\dot\\phi_1\\phi_2 - \\dot\\phi_2\\phi_1) + \\frac{1}{2} K_2 \\dot\\phi_2^2 - \\frac{1}{2} M_2^2 \\phi_2^2 \\right]$, that a linear canonical transformation followed by a field redefinition can move a negative sign in $K_1$ into the mass coefficient at low energies and into the gradient coefficient at high energies, without changing the equations of motion. Thus a low-energy ghost is equivalent to a tachyon and a high-energy ghost to a gradient instability. Since the transformation preserves the Hamiltonian equations and Poisson brackets, any instability that appears in one basis must appear in all bases; the negative coefficient simply changes location. The paper also shows that when one field is unstable, the interaction term $D$ couples the fields so that both grow, and that $|D|$ controls the instability timescale while the sign of $D$ is irrelevant. In the exceptional case of two ghosts, for restricted parameter values the interaction decouples the positive- and negative-energy sectors, giving stable oscillatory solutions.","pith_inferences":["If the equivalence is generic, then the standard practice of quoting 'no ghosts' as an independent health condition in modified-gravity cosmology is redundant: a no-ghost requirement in one basis already encodes no-tachyon and no-gradient conditions, and model builders should report basis-invariant stability criteria instead.","The same canonical-transformation argument should extend to more than two fields and to vector or tensor perturbations, because it relies on the structure of the quadratic action rather than on the spin of the fields; this would make the ghost–tachyon–gradient equivalence a general property of linearised cosmological perturbations.","A concrete check of the paper's practical message would be to compute, for a given dark energy model, the growth timescale from the eigenfrequencies of eq. (26) and compare it with numerical evolution of the full perturbation equations; agreement would confirm that the interaction term $D$ indeed sets the timescale, while disagreement would point to missing time-dependence in the constant-coeffici"],"forward_implications":["Stability conditions for dark energy models should be imposed on basis-invariant combinations of coefficients or on eigenfrequencies, not on the individual signs of $K_i$ and $M_i^2$.","If any field in a two-field model has a high-energy ghost or gradient instability, the entire interacting system is unstable, with modes growing at least as $e^{kt}$, so such models are not viable unless a high-energy completion intervenes.","At low energies, a negative kinetic term and a negative mass term are equally dangerous: both produce exponential growth on a timescale set by the masses and modified by $|D|$, so the phenomenological condition is on the growth rate, e.g. growth slower than the Hubble rate $H_0$.","The interaction term $D$ must be included when constraining dark energy models, because even when it does not decide stability it changes the instability timescale.","In the exceptional two-ghost case, for $4D^2 > m_1^2 + m_2^2$ and $(D^2 - m_1^2 - m_2^2)^2 > 4 m_1^2 m_2^2$, the low-energy system is stable despite two negative kinetic terms, because the positive- and negative-energy sectors decouple."],"supporting_citations":[{"why":"Supplies the generic proof that linear scalar perturbations of scalar-tensor, vector-tensor, and bimetric models can be written in the two-field quadratic form of eq. (24).","marker":"[37]"},{"why":"Established the low-energy ghost–tachyon equivalence for a single massless canonical scalar in general relativity, the result this paper generalises to two interacting fields.","marker":"[44]"},{"why":"Provides the explicit reduction of scalar-tensor theories to the form of eq. (24) and the high/low-energy scalings of D, K_i, and M_i^2 used in the discussion.","marker":"[33]"},{"why":"Supplies the definition of ghost instabilities and the argument that a quantum vacuum with high-energy instabilities decays exponentially, underwriting the high-energy equivalence claim.","marker":"[19]"}],"fun_headline_variants":["Ghosts are low-energy tachyons, high-energy gradients","Dark energy instabilities: basis changes turn ghosts into tachyons or gradients","Ghost, tachyon, gradient: same instability, different disguise","Two dark energy ghosts can be saved by interactions","Interactions set the timescale for dark energy instabilities"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The central assumption is that any two-field perturbation theory can be rewritten in a standard diagonal form with one antisymmetric interaction term, and that the coefficients are effectively constant at very low and very high energies; if either fails, the claimed equivalences may not be generic.","fun_headline_variants_meta":{"raw":{"variants":["Ghosts are low-energy tachyons, high-energy gradients","Dark energy instabilities: basis changes turn ghosts into tachyons or gradients","Ghost, tachyon, gradient: same instability, different disguise","Two dark energy ghosts can be saved by interactions","Interactions set the timescale for dark energy instabilities"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001063,"raw_usage":{"total_tokens":4515,"prompt_tokens":1064,"completion_tokens":3451,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":680,"completion_tokens_details":{"reasoning_tokens":3367}},"tokens_in":680,"tokens_out":3451,"duration_ms":25986,"temperature":1.0,"reasoning_tokens":3367,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:20:39.590754+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Numerically evolve the linear scalar perturbations of a concrete two-field dark energy model (for instance a scalar-tensor theory) at low and high energies in two different field bases; the paper predicts identical growth timescales and the same instability structure in every basis, so finding bases that disagree on whether the system is stable or on the growth rate would refute the equivalence claim.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the explicit reduction of scalar-tensor theories to the form of eq. (24) and the high/low-energy scalings of D, K_i, and M_i^2 used in the discussion."},{"cited_title":"Also, we have defined y = (4D2 + m2 1 + m2","cited_arxiv_id":null,"evidence_quote":"Supplies the definition of ghost instabilities and the argument that a quantum vacuum with high-energy instabilities decays exponentially, underwriting the high-energy equivalence claim."}],"review_version":1}