{"id":"bdba6bfb-34b5-4e2f-8b95-14139b736408","arxiv_id":"hep-th/0506177","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":8.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Extremizing a single entropy function built from the Lagrangian over the near-horizon geometry yields both the horizon data and the entropy of an extremal black hole, with no appeal to supersymmetry.","lead":"Sen shows that for an extremal black hole whose near-horizon geometry is AdS2 times a sphere, the horizon values of the scalar fields, the sizes of the two factors, and the entropy all follow from extremizing one function built from the Lagrangian. The construction gives a charge-only recipe for black hole entropy that tolerates arbitrary higher-derivative corrections and does not assume supersymmetry.","discovery_kind":"unclear","skeptic_critique":{"model":"selfhosted/deepseek-v4.1-flash","headline":"The derivation is sound within its stated scope; the one soft spot is the unexamined assumption that the extremization equations (2.17) have a solution and that it is the horizon configuration, which the paper asserts generically but does not establish.","rationale":"The reader's weakest_assumption (absence of Chern-Simons terms, footnote 3) is a real restriction and the paper acknowledges it explicitly, so it is not a load-bearing concern in the sense of an unstated gap. The reader's own rationale identifies the more interesting soft spot: no worked example applying the machinery to a specific higher-derivative Lagrangian and checking against a known entropy. I agree that this is the main thing missing, and I would frame it slightly more sharply than the reader does. The paper proves a necessary condition (horizon data is a stationary point of F) and then states a determination claim (abstract:s2) that implicitly assumes existence and, for the attractor reading, uniqueness of the stationary point. Footnote 4 concedes a case where uniqueness fails. The entropy formula survives degeneracy, so the central numerical claim is not threatened; the attractor-mechanism framing is. This is a scope issue, not an error, and the derivation of (2.14)/(2.18) is correct as algebra. I therefore leave the verdict at ACCEPT; the concrete test above would convert the scope concern into either confirmation or a needed caveat, and it is the check I would most want to see run. No ad hominem or soundness challenge is warranted: the paper is a clean, self-aware derivation whose limitations are stated in the text.","tokens_in":31544,"tokens_out":99267,"duration_ms":1878315,"concrete_test":"Take the simplest explicit higher-derivative example the formalism should handle, e.g. the D=4 Gauss-Bonnet or an R^2-corrected Einstein-Maxwell-dilaton Lagrangian with a known extremal black hole solution, and carry out the program end to end: compute f(u,v,e,p), extremize F over u and v via (3.3), and check that the resulting S_BH matches the Wald entropy evaluated on the independently known solution. If the extremization reproduces the known entropy, the 'determined by extremizing' claim is confirmed in a nontrivial case; if the extremum is degenerate or misses the solution, the scope of abstract:s2 needs narrowing.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim (abstract:s2, eqs. 3.1-3.4) is a derivation, and the derivation itself is tight: Wald's entropy (2.7) plus the homogeneity identity (2.13) plus the v1 equation of motion gives (2.14)/(2.18) essentially by algebra, and the general-D extension is immediate. The step I would press on is the move from 'the equations of motion are (2.17)' to 'the horizon values of u, v are determined by extremizing F' (section 3, eqs. 3.3-3.4). This requires that (2.17) admit a solution and that the solution be the actual near-horizon data. The paper itself flags the failure mode in footnote 4: in N=2 theories the hypermultiplet scalars drop out of f at lowest order, so (2.17) does not fix them and the extremum is degenerate along those directions. That is an acknowledged limitation, not a new finding, but it shows the 'determined by extremizing' language in abstract:s2 is stronger than what is proved: what is proved is that any horizon configuration must be a stationary point of F, not that a stationary point exists or is unique. For the entropy formula (3.4) this is harmless (F is flat along the degenerate directions, so its value at the extremum is still S_BH), but for the attractor-mechanism reading it matters. A second, smaller point: the derivation of (2.13) assumes f_lambda has the scaling form v1 g(u, v2, p, lambda/v1, e/v1), which rests on the vanishing of covariant derivatives in the background (2.1) and on the absence of undifferentiated gauge fields. The paper states the latter condition in footnote 3 and the closing paragraph of section 1, so it is acknowledged; the former is argued from symmetry and is fine. Neither of these is a defect in the algebra; the concern is scope, and the paper is largely honest about it.","agreement_with_reader":"partial"},"referee_report":{"model":"selfhosted/deepseek-v4.1-flash","summary":"The manuscript proposes a variational formulation of extremal black hole entropy in D dimensions for near-horizon geometries of the form AdS_2 × S^{D-2}. The author defines a function f(u, v, e, p) by integrating the Lagrangian density over the sphere for a general symmetric background, with v_1, v_2 the AdS_2 and sphere radii, u the scalar values, e_i the radial electric fields and p_a the magnetic fluxes. The scalar and metric equations of motion are shown to be equivalent to stationarity of f with respect to u and v, while the gauge equations give q_i = ∂f/∂e_i. Using Wald's entropy formula (2.7) together with a homogeneity identity (2.13) that follows from r-t reparametrization invariance, the author derives S_BH = 2π(e_i ∂f/∂e_i − f) in eq. (2.14)/(2.18). Legendre transforming in e_i yields the entropy function F(u, v, q, p) of eq. (3.1), whose extremum over u and v is claimed to give both the horizon data and the entropy (eqs. 3.3–3.4). Section 4 checks that the construction reproduces the known relation between the Legendre-transformed entropy and the imaginary part of the N=2 generalized prepotential. No supersymmetry is assumed and no specific structure of the higher-derivative terms is used.","tokens_in":10954,"tokens_out":4123,"duration_ms":82755,"significance":"If correct, this is a general and useful result: a variational principle for extremal black hole entropy that holds for any Lagrangian built from gauge-invariant field strengths, independent of supersymmetry and of the detailed structure of the higher-derivative terms. It unifies the attractor mechanism with Wald's covariant entropy formula in a single statement, and it makes the entropy a function of charges alone once the extremization is carried out. The derivation is short and each step is checkable, and it has no free parameters and no fitted quantities. The cross-check in §4 against the Ooguri–Strominger–Vafa relation to the generalized prepotential is a genuine, falsifiable consistency test, and the corollary (2.19) for tree-level heterotic black holes gives a concrete prediction that can be compared with microscopic counting. The paper also correctly credits the entropy formula it builds on to the published work of Wald, Jacobson, Kang and Myers (refs. 34–37).","major_comments":[{"comment":"The abstract states that the horizon values of the scalars and the sizes v_1, v_2 are 'determined by extremizing' the entropy function, but what the derivation establishes is the converse implication: any near-horizon configuration satisfying the equations of motion must be a stationary point of F. Existence and uniqueness of the stationary point are assumed rather than shown, and the author's own footnote 4 notes that in N=2 theories the hypermultiplet scalars drop out of f at lowest order, so that the extremum is degenerate along those directions. This does not affect the entropy formula (3.4), since F is flat along the degenerate directions and its value at the extremum is still S_BH, but it does affect the attractor-mechanism reading of the result. I would ask the author to add one or two sentences after eq. (3.4) stating precisely what is proved (stationarity of the horizon data) and what is assumed (that a stationary point exists and is the horizon configuration), and to soften the corresponding wording in the abstract and in item 2 of §1 accordingly.","section":"§3, eqs. (3.3)–(3.4), and abstract:s2","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The derivation rests on the Lagrangian depending on the gauge fields only through their field strengths, which is what allows every factor of F^{(i)}_{rt} to enter f_λ through the combination e_i/v_1 in the scaling identity (2.13). The author acknowledges in footnote 3 that string theory violates this through Chern-Simons couplings of gauge fields to three-form strengths, and offers dualization of the two-form to an axion as a workaround, conceding that the analysis is incomplete if dualization cannot be carried out for all fields. Since this is an acknowledged limitation, my concern is with what remains unresolved rather than with the acknowledgement itself: dualization exchanges electric and magnetic charges, so it is not immediate that the entropy function computed in the dual frame reproduces the same S_BH(q, p) with the same identification of q_i as ∂f/∂e_i. A short paragraph stating explicitly whether the dual formulation preserves the Legendre structure and the charge identification, or a worked check in a case where the dual is known (for instance a four-dimensional N=2 theory with a Chern-Simons term), would materially strengthen the scope claim.","section":"§2, footnote 3, and the closing paragraph of §1","authors_acknowledge":null,"authors_acknowledge_p":null}],"minor_comments":[{"comment":"There is a typographical error in the sentence listing the v_1 dependence: 'every factor of F^{(i)}_{θφ} = e_i' should read 'F^{(i)}_{θφ} = p_i', since the magnetic flux is p_i sinθ/4π by eq. (2.1).","section":"§2, paragraph following eq. (2.12)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The step asserting that f_λ(u, v, e, p) has the form v_1 g(u, v_2, p, λ/v_1, e/v_1) is the most compressed part of the derivation. A sentence or two spelling out why no other v_1 dependence can arise (vanishing covariant derivatives in the background (2.1), absence of undifferentiated gauge fields, and the overall √−det g ∝ v_1) would make the homogeneity identity (2.13) self-evident to the reader.","section":"§2, eq. (2.13) and its derivation","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The statement that at string tree level and in the absence of Ramond-Ramond backgrounds the Lagrangian density at the horizon vanishes by the dilaton field equation is quoted without derivation. A brief indication of how the dilaton equation forces f = 0 (up to total derivatives) would help the reader assess the corollary S_BH = 2π q_i e_i.","section":"§1, item 3, and eq. (2.19)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The paper defines extremal black holes to be those whose near-horizon geometry is AdS_2 × S^{D-2}, and defines their entropy as the extremal limit of the non-extremal entropy. It would be useful to state explicitly which classes of extremal solutions are excluded by this definitional choice, so that the scope of the entropy function is unambiguous.","section":"§1, definition of extremal black holes","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The two observations listed in §4 are attributed to ref. [12], and the statement that 'all the normalization factors also work out correctly' is asserted rather than shown. Since this is the paper's main independent check against known results, a few lines indicating how the normalizations of the scalar fields against the magnetic and electric fields are fixed would make the comparison verifiable.","section":"§4","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The manuscript does not discuss whether the relevant stationary point of F is a maximum, minimum or saddle, nor whether this distinction has physical content (for example a stability or attractor-basin interpretation). A remark on this point would be a useful addition, even if only to say that the present analysis does not address it.","section":"§3, eq. (3.3)","authors_acknowledge":null,"authors_acknowledge_p":null}],"recommendation":"minor_revision","confidential_remarks":"The physics is sound and the result is likely to be widely used; my recommendation of minor revision reflects claim wording and scope statements rather than any problem with the derivation. The intermediate identity (2.12) overlaps with the author's companion paper cited as ref. [27], but the present manuscript derives it in full from eqs. (2.9)–(2.11), so the overlap does not raise a novelty concern for this submission."},"author_rebuttal":null,"desk_editor":{"model":"selfhosted/deepseek-v4.1-flash","letter":"This is the paper that gave everyone the entropy function, and it holds up. The move is simple and it is the right one: integrate the Lagrangian over the sphere on a general AdS2 x S^{D-2} background, Legendre transform in the electric fields, multiply by 2 pi, and extremize. Wald's formula (2.7) plus the scaling identity (2.13) plus the v1 equation of motion gives S = 2 pi (e df/de - f) in a couple of lines, and the general-D version (2.18) follows immediately. No supersymmetry, no assumptions about which higher-derivative terms appear. That generality is the contribution, and it is why the thing became a workhorse. The derivation is checkable line by line; I did not find a step that does not close. Section 4's recovery of the Ooguri-Strominger-Vafa relation is a genuine consistency check, not decoration, and the normalization works out, which is not nothing. The paper is also honest about its main restriction: no Chern-Simons terms, stated in footnote 3 and again at the end of section 1, with dualization to an axion offered as the workaround and the incompleteness conceded. That is the right way to handle it. On the stress-test concern: it lands, but softly. What is proved is that any horizon configuration is a stationary point of F, not that a stationary point exists or is unique. Footnote 4 already flags the degenerate case (hypermultiplet scalars dropping out of f at lowest order), and for the entropy formula (3.4) the degeneracy is harmless since F is flat along those directions. So the abstract's 'determined by extremizing' [abstract:s2] is a bit more than the algebra delivers for the attractor reading, less so for the entropy. Minor. The one thing I would actually want and do not see is a worked example: a specific higher-derivative Lagrangian, the resulting entropy checked against something independently known. Section 4 is a consistency check on the N=2 side, not a numerical test of the machinery. That is a real gap, though it is a gap in a short note rather than a flaw in the argument. Who should read it: anyone computing horizon data or entropy in a theory with curvature-squared or higher terms, and anyone thinking about the attractor mechanism beyond supersymmetry. It deserves referees, and it deserved to be published quickly. I would cite it without hesitation.","headline":"A short, correct paper that turned Wald's entropy into a practical variational principle; the only real soft spot is that 'determined by extremizing' is a shade stronger than what the algebra proves.","tokens_in":11614,"tokens_out":838,"would_cite":true,"duration_ms":18818,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"pith_extraction":{"msc":[],"pacs":["04.70.Dy","04.50.+h","11.25.-w"],"model":"selfhosted/deepseek-v4.1-flash","headline":"Sen proposes that a single function, built by integrating the Lagrangian over the horizon sphere and Legendre-transforming with respect to the electric fields, determines both the near-horizon data of an extremal black hole and its entropy:","keywords":["black hole entropy","attractor mechanism","entropy function","higher derivative gravity","Wald entropy","extremal black holes","AdS2 near-horizon geometry","Legendre transform"],"falsifier":"Take a theory whose Lagrangian contains a term with an undifferentiated gauge field, or a Chern-Simons coupling that cannot be removed by dualization, and an extremal black hole solution in it; compute the Wald entropy directly from the horizon geometry and compare with the value of $F$ at its extremum. A disagreement would show the Legendre-transform identity is an artifact of the field-strength-only assumption rather than a general principle. A second, sharper check is already available inside the paper's own comparison: recompute the normalization relating $F$ to the imaginary part of the $N=2$ generalized prepotential and verify that the factors of $2π$ and the electric-field normalization match.","tokens_in":11881,"feed_emoji":"🕳️","tokens_out":3795,"duration_ms":82997,"temperature":0.7,"pith_summary":"Sen asks whether the entropy of an extremal black hole — one whose near-horizon geometry is $AdS_2$ times a sphere — can be extracted from one variational principle rather than from supersymmetry. He defines an entropy function: integrate the Lagrangian density over the sphere for a general near-horizon background, Legendre-transform the result with respect to the electric fields, and multiply by $2π$. The claim is that extremizing this function over the scalar values and the two length scales of the geometry reproduces the horizon configuration and returns the black hole entropy as the value at the extremum. If right, the attractor mechanism — the fact that horizon scalars depend only on the charges — stops being a supersymmetry fact and becomes a consequence of the equations of motion of any higher-derivative theory. That would make entropy computable for Lagrangians where no supersymmetric shortcut is available.","feed_headline":"Extremize one function, get a black hole's entropy","feed_subtitle":"A single variational principle fixes horizon scalars, geometry sizes and entropy, with no supersymmetry assumed.","key_machinery":"The entropy function $F(u, v, q, p) = 2π(e_i ∂f/∂e_i - f)$, where $f$ is the Lagrangian density integrated over $S^{D-2}$ on a general $AdS_2 × S^{D-2}$ background and $e_i$ is eliminated through $∂f/∂e_i = q_i$. Two ingredients make it work. First, the Wald formula for entropy in a theory with higher-derivative terms, which simplifies drastically on this background because covariant derivatives of all tensors vanish. Second, a scaling identity: since every factor of $R_{rtrt}$ and $F^{(i)}_{rt}$ enters only in the reparametrization-invariant combinations $λ v_1^{-1}$ and $e_i v_1^{-1}$, the derivative with respect to the scaling parameter can be rewritten in terms of $e_i ∂/∂e_i$ and $v_1 ∂/∂v_1$; imposing the $v_1$ equation of motion then converts the Wald entropy into the Legendre-transformed form.","core_discovery":"For a $D$-dimensional theory of gravity coupled to scalars, one-form gauge fields and $(D-3)$-form gauge fields, and for a background with the symmetry of $AdS_2 × S^{D-2}$ parametrized by the two length scales $v_1, v_2$, the scalar values $u_s$, the electric fields $e_i$ and the magnetic charges $p_a$, the paper shows that the scalar and metric equations of motion are exactly the stationarity conditions $∂f/∂u_s = 0$ and $∂f/∂v_i = 0$ of the sphere-integrated Lagrangian $f$, while the gauge equations give $∂f/∂e_i = q_i$. Starting from the Wald entropy formula for higher-derivative gravity, and using a scaling identity that trades the derivative of $f$ with respect to the Riemann tensor for derivatives with respect to the electric fields and $v_1$, the entropy becomes $S_{BH} = 2π(e_i ∂f/∂e_i - f)$. Promoting this to a function $F(u, v, q, p)$ of the charges gives the stated result: extremizing $F$ over $u$ and $v$ fixes the horizon, and $F$ at that extremum is the entropy.","pith_inferences":["Because $F$ is defined off-shell, on a family of backgrounds rather than on a solution, it invites a thermodynamic reading: $F$ behaves like a free energy whose Legendre transform in the electric fields is the entropy, and one could ask whether a temperature-dependent version reproduces the near-extremal entropy as a series in $T$.","The derivation uses only the scaling of the Lagrangian under the $AdS_2$ factor, which suggests one could look for analogous entropy functions for other symmetric near-horizon geometries, such as squashed spheres or non-relativistic scaling backgrounds; the paper does not attempt this.","The Chern-Simons caveat is not a technicality: it covers physically central cases such as five-dimensional black holes with three-form couplings, so a natural follow-up is to check whether the entropy function still returns the correct Wald entropy when such terms are present but happen not to contribute to the solution at hand.","For a given Lagrangian the recipe is mechanical — evaluate a constant-background integral, differentiate, extremize — which makes it plausible to automate entropy computations for whole families of higher-derivative corrections without any case-by-case analysis of the field equations."],"forward_implications":["The horizon values of the scalar fields and the radii of $AdS_2$ and $S^{D-2}$ follow from extremizing an algebraic function of the charges, rather than from solving the full second-order field equations.","The entropy of an extremal black hole in higher-derivative gravity becomes a Legendre transform of a sphere integral, so the whole answer is carried by a function of the charges alone.","In tree-level string theory without Ramond-Ramond backgrounds, the dilaton equation forces the Lagrangian density to vanish at the horizon, reducing the entropy to $S_{BH} = 2π q_i e_i$ — the product of electric charge and horizon electric field.","The observation that the Legendre transform of the entropy with respect to electric charges equals the imaginary part of the $N=2$ generalized prepotential is recovered from this construction, with the normalization factors working out.","The argument never uses supersymmetry or any special structure of the higher-derivative terms, so it applies to any Lagrangian built from gauge-invariant field strengths."],"supporting_citations":[{"why":"Supplies the Wald entropy formula for gravity with higher-derivative terms, which the paper specializes to the $AdS_2 × S^{D-2}$ background.","marker":"[34]"},{"why":"Gives the covariant version of the entropy formula and the treatment of the Riemann tensor as an independent variable in the variation.","marker":"[36]"},{"why":"Provides the entropy formula in the form used here, where derivatives of the Lagrangian with respect to the Riemann tensor appear contracted with the horizon geometry.","marker":"[37]"},{"why":"Earlier work by the author deriving the expression of the entropy as a derivative of the sphere-integrated Lagrangian with respect to the scaling parameter, which equation (2.12) reproduces.","marker":"[27]"},{"why":"Source of the attractor mechanism for supersymmetric black holes, the phenomenon the paper reformulates as extremization of the entropy function.","marker":"[1]"},{"why":"The claim that the Legendre transform of the entropy with respect to electric charges equals the imaginary part of the generalized prepotential, which section 4 recovers from the entropy function.","marker":"[15]"},{"why":"Review of the near-horizon structure of $N=2$ black holes used in section 4 to identify which terms survive at the horizon and how scalar fields map to electric and magnetic fields.","marker":"[12]"}],"fun_headline_variants":["One entropy function extremizes to fix the horizon and entropy","Extremizing a single function yields the black hole entropy","Horizon scalars and AdS sizes follow from one extremum","No supersymmetry is needed to extremize the entropy function","Higher-derivative black holes reduce to extremizing one function"],"cache_read_input_tokens":0,"weakest_assumption_plain":"The derivation assumes the Lagrangian depends on the gauge fields only through their field strengths, so that no Chern-Simons-type term appears; the author states this condition explicitly in footnote 3 and in the closing paragraph of the introduction, notes that string theory violates it through couplings of gauge fields to three-form strengths, and offers dualizing the two-form to an axion as a workaround while conceding that the analysis is incomplete if that dualization cannot be carried out for every field.","fun_headline_variants_meta":{"raw":{"variants":["One entropy function extremizes to fix the horizon and entropy","Extremizing a single function yields the black hole entropy","Horizon scalars and AdS sizes follow from one extremum","No supersymmetry is needed to extremize the entropy function","Higher-derivative black holes reduce to extremizing one function"]},"model":"selfhosted/deepseek-v4.1-flash","effort":"low","attempts":1,"cost_usd":0.0,"raw_usage":{"total_tokens":2964,"prompt_tokens":978,"completion_tokens":1986,"prompt_tokens_details":null,"completion_tokens_details":{"reasoning_tokens":1901}},"tokens_in":978,"tokens_out":1986,"duration_ms":34321,"image_pages":[],"temperature":1.0,"reasoning_budget":50,"reasoning_tokens":1901,"transport_provider":"selfhosted","cache_read_input_tokens":0,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-10-29T16:14:10.947737+00:00","model_set":{"reader":"selfhosted/deepseek-v4.1-flash"},"falsifier":"Take a theory whose Lagrangian contains a term with an undifferentiated gauge field, or a Chern-Simons coupling that cannot be removed by dualization, and an extremal black hole solution in it; compute the Wald entropy directly from the horizon geometry and compare with the value of $F$ at its extremum. A disagreement would show the Legendre-transform identity is an artifact of the field-strength-only assumption rather than a general principle. A second, sharper check is already available inside the paper's own comparison: recompute the normalization relating $F$ to the imaginary part of the $N=2$ generalized prepotential and verify that the factors of $2π$ and the electric-field normalization match.","supporting_citations":[{"cited_title":"Black Hole Entropy is Noether Charge","cited_arxiv_id":"gr-qc/9307038","evidence_quote":"Supplies the Wald entropy formula for gravity with higher-derivative terms, which the paper specializes to the $AdS_2 × S^{D-2}$ background."},{"cited_title":"Some Properties of Noether Charge and a Proposal for Dynamical Black Hole Entropy","cited_arxiv_id":"gr-qc/9403028","evidence_quote":"Gives the covariant version of the entropy formula and the treatment of the Riemann tensor as an independent variable in the variation."},{"cited_title":"BLACK HOLE ENTROPY IN HIGHER CURVATURE GRAVITY","cited_arxiv_id":"gr-qc/9502009","evidence_quote":"Provides the entropy formula in the form used here, where derivatives of the Lagrangian with respect to the Riemann tensor appear contracted with the horizon geometry."},{"cited_title":"Stretching the Horizon of a Higher Dimensional Small Black Hole","cited_arxiv_id":"hep-th/0505122","evidence_quote":"Earlier work by the author deriving the expression of the entropy as a derivative of the sphere-integrated Lagrangian with respect to the scaling parameter, which equation (2.12) reproduces."},{"cited_title":"Ferrara, R","cited_arxiv_id":null,"evidence_quote":"Source of the attractor mechanism for supersymmetric black holes, the phenomenon the paper reformulates as extremization of the entropy function."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The claim that the Legendre transform of the entropy with respect to electric charges equals the imaginary part of the generalized prepotential, which section 4 recovers from the entropy function."},{"cited_title":"Black Hole Entropy, Special Geometry and Strings","cited_arxiv_id":"hep-th/0007195","evidence_quote":"Review of the near-horizon structure of $N=2$ black holes used in section 4 to identify which terms survive at the horizon and how scalar fields map to electric and magnetic fields."}],"review_version":1}