{"id":"8a0532e1-4c9c-4556-b95e-bb712a312f86","arxiv_id":"hep-th/9612115","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A matrix model with manifest ten-dimensional N=2 supersymmetry is proposed as a candidate nonperturbative definition of type IIB superstring theory, supported by one-loop matching of BPS-state interactions to supergravity.","lead":"The authors propose a matrix model with ten-dimensional N=2 supersymmetry and argue it reproduces type IIB string theory, including its massless states and their long-range interactions. They conjecture that the large-N reduced version of ten-dimensional super Yang-Mills theory gives this model a nonperturbative definition.","discovery_kind":"paradigm_shift","skeptic_critique":{"model":"selfhosted/deepseek-v4.1-flash","headline":"The reader correctly locates the load-bearing gap: the double-scaling limit (4.8) presupposes a UV fixed point that Sec. 4.2 asserts but does not demonstrate, and the paper's own Eq. (4.17) exhibits a 1/a^2 divergence requiring a counterterm absent from the bare action (4.1).","rationale":"The reader's weakest_assumption identifies the right load-bearing premise: the existence of a nontrivial UV fixed point underlying the double-scaling limit (4.8). I agree, and I would sharpen it with two internal observations the reader did not make. First, the paper's own Eq. (4.17) exhibits a 1/a^2 Tr(F F F F) divergence, a dimension-8 operator absent from the bare action (4.1); the paper acknowledges counterterms are needed but does not show that finitely many suffice. Second, the paper's appeal to the divergence structure of 10D SYM is not independent support, because the paper itself states that 10D SYM is not perturbatively renormalizable, so the analogy points at an equally open question rather than at a resolution. The logical gap between 'divergences are mild' and 'a fixed point exists' is real and is the crux. Credit where due: the N=2 supersymmetry algebra closes (Eqs. 2.29-2.33), the BPS backgrounds receive no one-loop correction (3.17), and the D-string interaction (3.31) matches the supergravity exchange (3.58) in functional form and tensor structure. Those are genuine, checkable results. The reader is also right that the matching in Sec. 3.5 fixes alpha and beta against known string quantities, so the interaction coefficient is matched rather than predicted, while the tensor structure and the BPS non-renormalization are independent. The paper is honest about its limitations, so this is an acknowledged conjecture rather than a hidden gap. The verdict CONDITIONAL is appropriate and should not change; the concern is a reason to keep the conditional qualifier, not a reason to reject a paper that is explicit about what it has and has not shown. The concrete test above would settle whether the fixed-point assumption is merely unproven or is actively obstructed by an uncontrolled counterterm tower.","tokens_in":19145,"tokens_out":7077,"duration_ms":112742,"concrete_test":"Compute the two-loop effective action of the reduced model (4.1) around the flat diagonal background (4.9) and extract the coefficient of the leading 1/a^n divergence. Check whether that coefficient is proportional to the operator Tr[A_mu,A_nu]^2 already present in (4.1), or vanishes by the equations of motion, or requires an independent dimension-8 counterterm. If an independent counterterm with a free coefficient appears already at two loops, then the double-scaling limit requires tuning couplings not fixed by (4.1), and the 'constructive definition' claim does not follow from (4.1) alone. Equivalently, verify whether the 1/a^2 Tr(F F F F) term in (4.17) can be removed by a field redefinition or is a genuine new counterterm.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim [abstract:s2] is that the large-N reduced model (4.1) gives a constructive definition of the matrix model and hence of type IIB string theory. For that claim to hold, the double-scaling limit a -> 0, g0 ~ a^-3, N ~ a^-10 (Eq. 4.8) must exist, which requires a nontrivial UV fixed point of (4.1). Sec. 4.2 asserts this ('we expect that there is a nontrivial fixed point of renormalization group of the theory (4.1)') and offers two supports: (i) one-loop divergences are mild, of degree at most two; (ii) the divergence structure matches that of ten-dimensional super Yang-Mills (4.3). Neither is sufficient. Mildness of divergences is not a zero of the beta function; a fixed point requires the beta function to vanish, not merely to be small. And the paper itself notes that 10D SYM 'is not renormalizable perturbatively', so the analogy is not independent support: it appeals to a theory whose own UV completion is unestablished. More concretely, Eq. (4.17) produces a const.(1/a^2) Tr(F F F F) term, a dimension-8 operator not present in the bare action (4.1), which contains only the dimension-4 Tr[A,A]^2 and the fermion bilinear. The paper acknowledges 'we have to add counter terms to S0' but does not show that finitely many counterterms consistent with the N=2 supersymmetry and gauge symmetry suffice. If an infinite tower of tuned couplings is required, then (4.1) alone is not the definition; the definition is the unknown fixed point. This is not a fatal flaw, and the paper is explicit that the claim is a conjecture, but it is the precise place where the central claim is unsupported. A secondary point: the abstract's claim [abstract:s1] that the massless spectrum 'is the same as that of type IIB' is supported only by the long-range 1/r^8 potential between BPS objects (3.52), not by an explicit spectrum computation, which is weaker than the abstract's wording.","agreement_with_reader":"agree"},"referee_report":{"model":"selfhosted/deepseek-v4.1-flash","summary":"The paper proposes a zero-dimensional matrix model, Eq. (2.22), with manifest ten-dimensional N=2 super-Poincaré invariance, obtained by replacing Poisson brackets with commutators and worldsheet integrals with traces in the Green–Schwarz action in the Schild gauge. The authors verify that the N=2 algebra closes up to gauge transformations and the ψ equation of motion (Eqs. 2.29–2.33), compute the one-loop effective action around general backgrounds (Eq. 3.14, derived in the Appendix), show that it vanishes for BPS-saturated backgrounds (Eq. 3.17), and evaluate the interaction between two anti-parallel and two generally located static D-string solutions (Eqs. 3.31, 3.35), matching the latter to a supergravity exchange computation (Eq. 3.41). The parameters α and β are then fixed by matching to the D-string tension and gravitational coupling (Eqs. 3.56–3.60). The central claim, stated as a conjecture in abstract:s2, is that the large-N reduced ten-dimensional super Yang–Mills model (4.1), taken in a double-scaling limit a→0, g0∼a^{-3}, N∼a^{-10} (Eq. 4.8), provides a constructive nonperturbative definition of this matrix model and hence of type IIB superstring theory.","tokens_in":19465,"tokens_out":19932,"duration_ms":321879,"significance":"If the conjecture holds, this is a nonperturbative definition of type IIB string theory, and the paper is one of the founding proposals of the matrix-model approach to string theory. The computations that are carried out are sound and, in places, genuinely predictive: the relative tensor structure of the four-field-strength terms in Eq. (3.52), which encodes the relative couplings of graviton and scalar exchange, is independent of α and β and is a parameter-free statement about the massless spectrum. The paper is also candid about the conjectural status of the constructive definition, which is a virtue. The significance is high, but it rests on the conjecture rather than on a demonstration.","major_comments":[{"comment":"The double-scaling limit (4.8) presupposes a nontrivial UV fixed point of (4.1), which the text asserts ('we expect that there is a nontrivial fixed point of renormalization group of the theory (4.1)') but does not demonstrate. The one-loop result (4.17) produces a const.(1/a^2) Tr(F F F F) term, a dimension-8 operator absent from the bare action (4.1); the paper acknowledges that counterterms must be added but does not show that finitely many counterterms consistent with the N=2 supersymmetry and gauge symmetry suffice. A concrete test would be to exhibit the one-loop counterterm structure and check its closure under the symmetries, or to compute the two-loop beta function. Without such a check, the constructive-definition claim of abstract:s2 rests on an unverified assumption, and the 'mildness' of the divergences is not by itself a zero of the beta function.","section":"Sec. 4.2, Eq. (4.17) and Eq. (4.8)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The sum over n in (2.21) is weighted by e^{-S} with S containing +β Tr 1 = βn, and β is fixed positive in Eq. (3.60). This suppresses rather than enhances large n, in tension with the stated expectation that 'large values of n dominate' and with the correspondence (2.23), which requires large n. From Eq. (4.13) the entropy factor is (l^10 α^{5/2})^N/N!, so large N requires l^10 α^{5/2} e^{-β} ≫ 1; this condition is neither stated nor checked against the double-scaling limit (4.8). Please state the condition under which the large-n saddle dominates, and verify that it is compatible with (4.8).","section":"Sec. 2, Eq. (2.21)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The general block-block interaction is reported in (3.52) as O(1/d^8) with four field strengths, whereas the exact computation for the anti-parallel D-string background gives O(f^3/b^6) with f_01 = -TL/2πn (Eq. 3.31). In the D-string background f_01 is O(1) and the F-expansion underlying (3.51) is not obviously controlled; please state the regime of validity of (3.51)–(3.52) and reconcile the two scalings. This matters because the massless-spectrum claim of abstract:s1 rests on the tensor structure extracted from (3.52).","section":"Sec. 3.4, Eq. (3.52) versus Sec. 3.3, Eq. (3.31)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"α and β are determined by matching the D-string tension ρ and the gravitational coupling κ, so the overall coefficient of the 1/b^6 interaction in (3.31) is matched to (3.58) rather than predicted. The genuine parameter-free content is the relative tensor structure in (3.52), i.e., the relative coefficients of the graviton and scalar exchanges, which is independent of α and β. The paper should state this distinction explicitly, because the abstract's claim that the massless spectrum 'is the same as that of type IIB string theory' is supported by the tensor structure, not by the normalization.","section":"Sec. 3.5, Eqs. (3.56)–(3.60)","authors_acknowledge":null,"authors_acknowledge_p":null}],"minor_comments":[{"comment":"The subscript λ in P^2_λ is not defined anywhere in the text; please define it or remove it.","section":"Eq. (3.14)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The reality and hermiticity conditions on ψ after the analytic continuation θ2 → iθ2 should be stated explicitly, since the trace over 16-component spinors with the (1+Γ_11)/2 projection and the Minkowski continuation require care.","section":"Sec. 2, footnote on p. 3","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The criterion 'P_i = 0 for at least one direction i' does not literally hold for the tilted two-D-string configuration (3.13), where P'_2 has no c-number piece but is not zero as an operator; please explain why the phase θ vanishes there.","section":"Appendix, case (2), Eq. (A.19)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The statement that the global N=2 supersymmetry 'is extended to the local one inevitably since the N=1 supersymmetry is maximal in ten-dimensional gauge theory' does not follow for a zero-dimensional matrix model; local supersymmetry is not implied by global supersymmetry plus Lorentz invariance in this setting. This does not affect the main claim, which rests on the one-loop computation, but the sentence should be softened.","section":"Sec. 2, paragraph after Eq. (2.33)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The identification of the BFSS parameters g and R with those of the matrix model is not given; please state the map explicitly.","section":"Sec. 4.3, Eqs. (4.19)–(4.20)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The identification of the Wilson loop with the string creation operator is heuristic, as the paper acknowledges ('we cannot demand the exact coincidence'); it should be presented as a consistency check rather than as evidence.","section":"Sec. 5, Eqs. (5.1)–(5.6)","authors_acknowledge":null,"authors_acknowledge_p":null},{"comment":"The error terms in the approximate commutation relation [q,p] = i for finite n should be estimated, since they propagate into the one-loop results of Sec. 3.3.","section":"Sec. 3.1, Eqs. (3.6)–(3.7)","authors_acknowledge":null,"authors_acknowledge_p":null}],"recommendation":"major_revision","confidential_remarks":"This is a well-known and influential proposal, and the computations that are carried out are sound. My concerns are about the strength of the claims relative to the evidence, not about the correctness of the one-loop algebra. The authors may prefer to reframe the paper as an explicit proposal with clearly labelled conjectures, in which case the load-bearing gap in Sec. 4.2 would be less of an issue; as written, the abstract's 'therefore is equivalent to superstring theory' overstates what is demonstrated."},"author_rebuttal":null,"desk_editor":{"model":"selfhosted/deepseek-v4.1-flash","letter":"This is the IKKT paper, and it earns its place. The model in Eq. (2.22) is new: a matrix integral with manifest ten-dimensional N=2 invariance, obtained by pushing the Green-Schwarz action in Schild gauge through the Eguchi-Kawai correspondence. The supersymmetry algebra closes up to gauge transformations and the psi equation of motion (2.29-2.33), the one-loop effective action (3.14) is actually computed in the appendix, BPS backgrounds get no corrections (3.17), and the interaction between two anti-parallel D-strings (3.31) reproduces the supergravity exchange in tensor structure and distance dependence (3.58). Those are checkable results, not hand-waving. The observation in Sec. 4.3 that the same model contains BFSS at another point of moduli space is a nice bonus and, in hindsight, the most consequential paragraph in the paper. The writing is honest about what is conjecture and what is derived, which I appreciate more now than I did at the time. Two soft spots, in proportion. First, the load-bearing one: the constructive definition needs the double-scaling limit (4.8), which needs a nontrivial UV fixed point of (4.1). Sec. 4.2 asserts the expectation and offers mildness of one-loop divergences plus an analogy to 10D SYM, which is not independent support since that theory's own UV completion is exactly what is at issue. Eq. (4.17) even exhibits a 1/a^2 Tr(FFFF) divergence requiring a counterterm absent from the bare action, and the paper says counterterms must be added without showing finitely many suffice. That is the precise gap. It is acknowledged, not hidden, so I would not call it a flaw in honesty, only in completeness. Second, and minor: the abstract's claim that the massless spectrum 'is the same as that of type IIB' is supported by long-range 1/r^8 potentials between BPS objects, not by a spectrum computation. The wording runs slightly ahead of the evidence. Also note the alpha/beta matching in Sec. 3.5 fixes two parameters against known string quantities, so the overall coefficient is matched rather than predicted; the tensor structure and the non-renormalization are the independent content. Worth a serious referee, and worth citing.","headline":"The IKKT paper: a genuinely new supersymmetric matrix model with real one-loop checks, whose constructive-definition claim rests on an unproven UV fixed point the authors openly flag as an expectation.","tokens_in":20307,"tokens_out":1015,"would_cite":true,"duration_ms":16106,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"pith_extraction":{"msc":[],"pacs":["11.25.Sq","11.15.Pg","11.30.Pb"],"model":"selfhosted/deepseek-v4.1-flash","headline":"The authors propose that a matrix model built from ten hermitian matrices and one fermion, with manifest ten-dimensional $N=2$ supersymmetry, is equivalent to type IIB superstring theory, and that the large-$N$ reduced model of ten-­­­­­­­­","keywords":["matrix model","large-N reduction","type IIB superstring","Green-Schwarz action","Schild gauge","BPS-saturated states","D-branes","nonperturbative formulation of string theory"],"falsifier":"A renormalization-group analysis of the reduced model showing that its couplings flow to a free theory or to strong coupling with no fixed point in the required universality class, or equivalently a demonstration that the $U(1)^d$ symmetry is spontaneously broken in the weak-coupling limit, would remove the continuum theory the conjecture needs. A sharper and more local test is a two-loop computation of the force between two separated blocks: if it departs from the $1/r^8$ graviton-plus-antisymmetric-tensor form once $alpha'$ is fixed by the relation $TL/2pi n = 2sqrt{beta/alpha}$, the identification with type IIB fails at the first quantum correction.","tokens_in":20400,"feed_emoji":"🧵","tokens_out":3152,"duration_ms":47169,"temperature":0.7,"pith_summary":"The paper asks whether a finite-dimensional matrix integral can stand in for an entire string theory. Its authors propose an action built from ten hermitian matrices $A_$ and a fermion $A_$, namely $S = -$, and choose it because it carries the same ten-dimensional $N=2$ supersymmetry algebra as the type IIB Green-Schwarz string written in the Schild gauge. Their evidence is twofold: the one-loop effective action around BPS-saturated backgrounds vanishes exactly, and the leading interaction between two widely separated blocks of eigenvalues has the $1/r^8$ form produced by exchanging a graviton and an antisymmetric tensor field, with the string tension fixed by the eigenvalue spacing. If the conjecture holds, string theory acquires a definition that is finite-dimensional at every step, manifestly supersymmetric, and automatically includes multi-string states as block-diagonal configurations.","feed_headline":"Matrices may give string theory a nonperturbative definition","feed_subtitle":"A ten-dimensional supersymmetric matrix model reproduces the massless spectrum and the forces between D-strings.","key_machinery":"The load-bearing device is the correspondence $-i[~,~]$, which sends matrix commutators to Poisson brackets and traces to integrals over a worldsheet, converting the matrix action into the Schild form of the Green-Schwarz string. On top of that sits the one-loop effective action around a general background, $W = 1/2 Tr log(P^2 - 2iF) - 1/4 Tr log((P^2 + 1/2 i F_{munu} Gamma^{munu})(1+Gamma_{11})/2) - Tr log(P^2)$, in which $P$ and $F$ are the adjoint actions of the background. This formula is what makes the BPS argument work: when the background field strength is a c-number the bosonic, fermionic and ghost contributions cancel, giving $W=0$, and when two blocks are separated the expansion of the same formula produces the $1/r^8$ graviton and antisymmetric-tensor exchange that lets the authors match $TL/2pi n$ to the string slope $alpha'$ and thereby fix the two couplings of the model.","core_discovery":"The central claim is that the large-$N$ reduced model of ten-dimensional super Yang-Mills theory, taken in a double-scaling limit, is a constructive definition of type IIB superstring theory. The chain runs as follows. The zero-volume reduction of the super Yang-Mills action, plus a term proportional to the identity, is a matrix model whose commutators and traces become Poisson brackets and worldsheet integrals in a smooth large-$N$ limit, so the matrix action turns into the Green-Schwarz action in Schild gauge. The model has a manifest $N=2$ ten-dimensional supersymmetry, which the authors verify by computing the commutators of the two supersymmetry variations up to gauge transformations and the fermion equation of motion; since $N=1$ is maximal for a ten-dimensional gauge theory, the second supersymmetry is argued to be unavoidable. Straight D-strings appear as classical solutions in which two matrices satisfy the canonical commutation relation $[q,p]=i$ and the remaining matrices are diagonal constants, and the one-loop corrections around these BPS backgrounds cancel between bosons and fermions. Space-time itself is not an input: the eigenvalues of the matrices are reinterpreted as coordinates, and the trace parts of diagonal blocks become the centres of mass of the objects they describe.","pith_inferences":["The BPS argument is a one-loop statement, so the natural next test is a two-loop computation of the force between separated blocks; a deviation from the $1/r^8$ graviton-plus-antisymmetric-tensor form at that order would be the cleanest place to see whether the correspondence survives quantum corrections.","The identification of the string slope with $TL/2pi n$ ties the model's coupling to the density of eigenvalues, which suggests that the string coupling itself is not a free parameter but is determined by how the eigenvalues distribute in the double-scaling limit.","If the emergent-space picture is right, then the notorious difficulty of formulating quantum gravity in fixed backgrounds is replaced by the difficulty of showing that a smooth eigenvalue distribution is a stable saddle, which is a question about the eigenvalue gas rather than about geometry.","The paper's own remark that the moduli space can be read either as type IIB or as type IIA invites the question of what distinguishes the two readings at finite $N$, where T-duality is not exact."],"forward_implications":["String theory would have a definition that is a finite-dimensional matrix integral at every stage, with no worldsheet path integral to regulate and no sum over genera inserted by hand.","Multi-string states would come for free: whenever the matrices are block-diagonal the action splits into a sum of traces, one per block, so the single model already contains arbitrary numbers of strings and their interactions.","Space-time coordinates would be emergent rather than assumed, arising as the eigenvalues of the matrices, with the trace parts of diagonal blocks playing the role of centres of mass.","Because the model's moduli space contains a point whose action coincides with the Banks-Fischler-Shenker-Susskind matrix quantum mechanics, the same construction would also contain type IIA string theory, with the two readings related by T-duality.","Wilson loops of the reduced model would act as creation operators for string states with definite momentum, while the classical matrix solutions would act as the D-objects on which those strings end."],"supporting_citations":[{"why":"Supplies the large-$N$ reduction of degrees of freedom (Eguchi-Kawai and follow-ups) that the whole constructive proposal rests on.","marker":"[4]"},{"why":"The Banks-Fischler-Shenker-Susskind matrix quantum mechanics, which the paper recovers as a point in its own moduli space and uses to argue for type IIA content.","marker":"[5]"},{"why":"The Green-Schwarz covariant type IIB action, the starting point that the matrix model is built to reproduce.","marker":"[6]"},{"why":"Schild's gauge, the form of the string action that the matrix commutator-and-trace replacement lands on.","marker":"[7]"},{"why":"Polchinski's D-brane identification, which underwrites reading the classical matrix solutions as D-strings.","marker":"[3]"},{"why":"The Witten-Olive notion of BPS-saturated states, which is the criterion used to select the backgrounds whose one-loop corrections vanish.","marker":"[9]"}],"fun_headline_variants":["Large-N reduced super Yang-Mills is conjectured to define type IIB strings","Ten-dimensional matrix model reproduces the type IIB massless spectrum","BPS D-string interactions match type IIB in the matrix model","Matrix model equivalence to superstrings is conjectured, not proved","Ten-dimensional N=2 matrix model has manifest super Poincare invariance"],"cache_read_input_tokens":0,"weakest_assumption_plain":"The argument needs the large-$N$ reduced model to possess a nontrivial ultraviolet fixed point, so that the double-scaling limit $a to 0$, $g_0 sim a^{-3}$, $N sim a^{-10}$ actually defines a continuum theory; the paper states this as an expectation and does not demonstrate it.","fun_headline_variants_meta":{"raw":{"variants":["Large-N reduced super Yang-Mills is conjectured to define type IIB strings","Ten-dimensional matrix model reproduces the type IIB massless spectrum","BPS D-string interactions match type IIB in the matrix model","Matrix model equivalence to superstrings is conjectured, not proved","Ten-dimensional N=2 matrix model has manifest super Poincare invariance"]},"model":"selfhosted/deepseek-v4.1-flash","effort":"low","attempts":1,"cost_usd":0.0,"raw_usage":{"total_tokens":1592,"prompt_tokens":846,"completion_tokens":746,"prompt_tokens_details":null,"completion_tokens_details":{"reasoning_tokens":666}},"tokens_in":846,"tokens_out":746,"duration_ms":11606,"image_pages":[],"temperature":1.0,"reasoning_budget":50,"reasoning_tokens":666,"transport_provider":"selfhosted","cache_read_input_tokens":0,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-10-29T19:08:14.674741+00:00","model_set":{"reader":"selfhosted/deepseek-v4.1-flash"},"falsifier":"A renormalization-group analysis of the reduced model showing that its couplings flow to a free theory or to strong coupling with no fixed point in the required universality class, or equivalently a demonstration that the $U(1)^d$ symmetry is spontaneously broken in the weak-coupling limit, would remove the continuum theory the conjecture needs. A sharper and more local test is a two-loop computation of the force between two separated blocks: if it departs from the $1/r^8$ graviton-plus-antisymmetric-tensor form once $alpha'$ is fixed by the relation $TL/2pi n = 2sqrt{beta/alpha}$, the identification with type IIB fails at the first quantum correction.","supporting_citations":[{"cited_title":"Polchinski, Phys","cited_arxiv_id":null,"evidence_quote":"Supplies the large-$N$ reduction of degrees of freedom (Eguchi-Kawai and follow-ups) that the whole constructive proposal rests on."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The Banks-Fischler-Shenker-Susskind matrix quantum mechanics, which the paper recovers as a point in its own moduli space and uses to argue for type IIA content."},{"cited_title":"M Theory As A Matrix Model: A Conjecture","cited_arxiv_id":"hep-th/9610043","evidence_quote":"The Green-Schwarz covariant type IIB action, the starting point that the matrix model is built to reproduce."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Schild's gauge, the form of the string action that the matrix commutator-and-trace replacement lands on."},{"cited_title":"Witten, Nucl","cited_arxiv_id":null,"evidence_quote":"Polchinski's D-brane identification, which underwrites reading the classical matrix solutions as D-strings."},{"cited_title":"Bars, Phys","cited_arxiv_id":null,"evidence_quote":"The Witten-Olive notion of BPS-saturated states, which is the criterion used to select the backgrounds whose one-loop corrections vanish."}],"review_version":1}