{"id":"87f52829-e078-4445-8b60-9cd3f544bbc0","arxiv_id":"1908.05330","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A new level-truncation method reaches level 24 in the Schnabl gauge and shows the tachyon vacuum energy has a local minimum at level 12 before extrapolating toward the analytic value -1.","lead":"This paper computes, to much higher precision than before, the numerical tachyon vacuum solution in the Schnabl gauge of open bosonic string field theory, and shows its energy approaches the predicted value minus one. The result strengthens numerical evidence that this gauge-truncated solution matches the exact analytic solution at infinite level, and introduces a numerical method usable in other gauges.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Extrapolation to infinite level is the load-bearing step: for the Ellwood invariant the generic maximal-order 1/L fit fails (giving 3.36), so the paper switches post hoc to separate 4k and 4k+2 fits; higher-level data are needed to confirm the infinite-level identification.","rationale":"The reader's weakest-assumption identification is on target: Appendix A's polynomial ansatz and the additional 1/N fit are the only bridge from L = 24 data to infinite level. The Ellwood-invariant analysis shows the problem in its starkest form: the standard maximal-order fit gives +3.36, so the paper changes the extrapolation function after the fact. That is not circular in a strict logical sense, but it means the claimed 0.15% agreement with E0 = 0 is not a prediction of the method. The energy conclusion is somewhat better supported, since the last few maximal-order extrapolants (−0.9997, −0.9996, −0.9995 for N = 8, 9, 10, 11) are stable at the 1e-4 level, but larger-N fits have not converged and Section 7 explicitly admits that non-analytic asymptotic behavior may be the real situation. The finite-level results, especially the confirmed local energy minimum at L = 12, are solid and are not in question. The reported consistency checks (symmetries in Table 5.2 and out-of-gauge equations in Table 6.1) point in the right direction, although the out-of-gauge checks inspect only five specific dependent-variable equations, so they do not certify that the whole projected-out subspace vanishes. The single most useful check is therefore to extend the numerical data to L = 26–30; this directly tests whether the chosen asymptotic form is correct and would settle whether the infinite-level identification is more than a plausible extrapolation. No change to the reader's CONDITIONAL verdict is needed: the concern is the same one, and the paper's own Section 7 already flags it.","tokens_in":21930,"tokens_out":6588,"duration_ms":66326,"concrete_test":"Compute the L = 26, 28, and 30 numerical solutions with the same factorized-vertex Newton solver and canonical projector. For the energy, add the new points to Table 4.1 and recompute the maximal-order extrapolants E^(2,26)_12, E^(2,28)_13, and E^(2,30)_14: if the infinite-level limit moves by more than the claimed ~1e-4 from −0.99949, the polynomial-in-1/L ansatz is not supported. For the Ellwood invariant, add L = 26 and L = 30 to the 4k+2 subsequence and L = 28 to the 4k subsequence, refit f5 and g5, and check whether the two asymptotic values remain within ~5e-4 of each other and of 0; if the limits shift significantly or the two subsequences diverge, the post-hoc split-and-average procedure was not predictive.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that the L≤24 numerical Schnabl-gauge vacuum is the L→∞ analytic vacuum depends entirely on extrapolating the finite-level data. For the energy, Table 4.2 shows maximal-order extrapolants E_N(∞) = −1.0299, −1.0190, …, −0.999486 for N = 1, …, 11; the spread is ~3×10^-2 and the sequence has not plateaued. Section 7 explicitly concedes that the numerical solution may be non-analytic in 1/L, in which case the ansatz q(Lmin,Lmax)_M(L) = a0 + Σ a_n/L^n in Eq. (A.1) is not the correct asymptotic expansion. The Ellwood invariant is more exposed: a single maximal-order polynomial fit of all twelve data points gives an infinite-level value of 3.3629, far from the expected 0. To obtain −0.0015, the authors split the data into L = 4k and L = 4k + 2 subsequences and fit each with order N = 5 (Eqs. 4.2–4.3), obtaining −0.00172 and −0.00129, then average them. This fit order and data split are chosen after the standard fit fails, and the two subsequence limits differ by about 25% of the final answer. With only six points per subsequence, the claimed 0.15% precision of E0 is not established. The out-of-gauge checks in Table 6.1 cover only the five dependent variables fw, fw1, fw2, fw3, fw4, not the full projected-out subspace, so they do not independently certify that all unsolved equations vanish. Thus the infinite-level identification rests on a partially unstable and, for E0, post hoc extrapolation procedure.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a level-truncation-based numerical method for Witten's open bosonic string field theory in the Schnabl gauge, allowing computations up to level L=24, and uses it to study the tachyon vacuum. It reports that the vacuum energy has a local minimum at L=12 (confirming the earlier prediction of [36]) and that extrapolations in 1/L give an infinite-level energy near -0.9995, an Ellwood invariant near -0.0015, and coefficients close to Schnabl's analytic values. The paper also checks K1 matter symmetries and a subset of out-of-gauge equations, and discusses other numerical solutions. The authors conclude that the numerical solution is most likely the infinite-level analytic Schnabl solution, but they explicitly acknowledge partial instability and the possibility that the solution is non-analytic around infinite level.","tokens_in":22406,"tokens_out":5163,"duration_ms":48150,"significance":"If the identification with Schnabl's analytic solution is correct, the paper provides the first high-level numerical confirmation in the Schnabl gauge and a general method for nontrivial linear b-gauges. The finite-level data are valuable and internally consistent: the local minimum at L=12 is visible directly in Table 4.1 without any extrapolation, and the method reproduces the earlier L<=10 results. The paper is also commendably explicit about the limitations of its extrapolations. However, the infinite-level claims are the load-bearing part of the paper and currently rest on extrapolation procedures whose stability is not established, especially for the Ellwood invariant and for the energy.","major_comments":[{"comment":"The claimed infinite-level value of the Ellwood invariant E0 is not supported by the data alone. A single maximal-order 1/L fit to all twelve points of Table 4.3 gives 3.3629, which the authors discard; they then split the data into L=4k and L=4k+2 subsequences and fit each with order N=5, obtaining -0.00172 and -0.00129, and average these to -0.0015. This split and fit order are selected after the standard fit fails, and the two subsequence limits differ by about 25% of the final answer. With only six points per subsequence, the statement that E0 is reproduced with 0.15% precision is not established. The authors should present the full set of extrapolants for all choices of Lmin, Lmax, and order M, justify the 4k/4k+2 split independently of the target value, and report a systematic uncertainty that includes the spread of the two subsequence fits.","section":"§4, Eqs. (4.2)–(4.3)"},{"comment":"The energy extrapolation is not stable enough to justify the claimed 10^-4 error. The maximal-order extrapolants E_N(infinity) in Table 4.2 run from -1.0299 (N=1) to -0.999486 (N=11), a spread of about 3×10^-2, and the sequence has not plateaued. Section 7 explicitly concedes that the numerical solution may be a non-analytic function of 1/L, in which case the ansatz (A.1) is not the correct asymptotic expansion. The conclusion that Sen's first conjecture holds at the 0.05% level should be reformulated as a tentative statement, with an uncertainty estimate that includes the spread over extrapolation orders and the possibility of non-polynomial asymptotics.","section":"§4, Table 4.2 and §7"},{"comment":"The second, N-to-infinity extrapolation for the string field coefficients is calibrated against the known analytic answer: the degree M=5 is chosen because it best reproduces the analytic tachyon coefficient, and the authors themselves call the resulting 0.01% agreement 'probably coincidental' (Section 3). Since the same M=5 fit yields values for u, v, w, and w1 that are less accurate, the evidence that the numerical coefficients converge to the analytic ones is weaker than the tachyon coefficient alone suggests. Please quantify the dependence on M and report the fit quality for all coefficients without using the known target values as a selection criterion.","section":"§3, Table 3.3"},{"comment":"The out-of-gauge checks cover only fw, fw1, fw2, fw3, and fw4, the five equations associated with the dependent variables at level 4. At higher levels the projector removes many more equations, so these five components do not certify that the full projected-out subspace vanishes asymptotically. The text should state this limitation explicitly, or extend the check to a larger set of out-of-gauge equations.","section":"§6, Table 6.1"}],"minor_comments":[{"comment":"The index in the sum should be t(I)_j, not t(I)_i; as written the equation is circular.","section":"§2, Eq. (2.17)"},{"comment":"The notation <V|Psi> and <I|V(i)|Psi> is inconsistent; please define V(i) and the normalization used for the identity string field.","section":"§4, Eq. (4.1)"},{"comment":"The column for w1 is described as 'from level 4', but the first numerical entry appears at L=6; please clarify why the level-2 and level-4 entries are omitted.","section":"Table 3.1"},{"comment":"Add axis labels and a legend identifying which curve is f5(L) and which is g5(L).","section":"Figure 4.2"},{"comment":"The use of Mathematica's NonlinearModelFit is implementation-specific; state the least-squares formulation and the handling of numerical precision in a way that makes the procedure reproducible.","section":"Appendix A"},{"comment":"The tachyon coefficient is listed under the heading c1|0> in Table 1.1 but as t in Section 3; align the notation across the two tables.","section":"Table 1.1"}],"recommendation":"major_revision","confidential_remarks":"The paper is a solid numerical study and the finite-level results are worth publishing after the extrapolation claims are tightened. I recommend major revision rather than rejection because the core computational contribution is reproducible and the issues are concentrated in the interpretation of the infinite-level limits. The citation to the unpublished thesis [38] for the extrapolation procedure should be supplemented with enough detail for the present claims to be checked."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a genuinely useful numerical methods paper with honest caveats, not a physics breakthrough. The new elements are the canonical-projector Newton method (section 2) that pushes Schnabl-gauge level truncation from L=10 to L=24, and the data that come with it. The infinite-level extrapolations are the weak joint, and the authors mostly know it.\n\nWhat is actually good. The projector construction is clean and applies to any linear b-gauge; the factorized-vertex handling is the key trick that makes L=24 feasible. They verify the method reproduces the old L≤10 results, so the pipeline is credible. The L=24 energy data in table 4.1 directly show the predicted L=12 local minimum, which settles a specific conjecture from [36]. The finite-level tables look internally consistent, and appendix A's predictive test (fitting lower levels to forecast L=24) is a good check that the data are smooth. The out-of-gauge residual checks are limited to the five dependent variables, but as a spot check they behave sensibly.\n\nWhere it gets soft. Everything about L=∞ sits on 1/L polynomial extrapolation, and that load-bearing step is only partially controlled. The energy extrapolants in table 4.2 bounce around in the tail by ~10^-4 to 10^-3; the paper quotes an error of order 10^-4 while the central value sits about 5×10^-4 from -1, so the error estimate is not really earned. The Ellwood invariant is more exposed: the single maximal-order fit of all twelve points gives 3.36, nowhere near zero. The paper recovers a sensible value by splitting into 4k and 4k+2 subsequences and averaging two order-5 fits. That split is not purely invented — the 4-level separation is standard for Ellwood invariants (footnote 5, ref [38]) — but two six-point fits whose limits differ by about a quarter of the answer cannot support the claimed 0.15% precision. Similarly, the N→∞ extrapolations for the tachyon vev use M=5 partly because it lands near the analytic value; the authors admit this is likely coincidental. So the central identification — that the numerical solution converges to Schnabl's analytic vacuum — is supported by converging evidence but not established by controlled error estimates. The paper says as much in section 7, which is to its credit.\n\nWho this is for: anyone doing numerical open string field theory, especially in non-Siegel gauges; the projector method is the reusable takeaway. The physics conclusions are already known analytically, so the value is in the technique and in the L=24 reference data. This deserves a proper referee — the method should be checked and the extrapolation claims tightened. I would also ask the authors to make code and data available, since none are included.\n\nRecommendation: send to peer review with a request to soften the infinite-level precision claims and to justify or remove the post-hoc fit choices.","headline":"Genuinely reusable numerical method and L=24 Schnabl-gauge data; the infinite-level claims are suggestive, but the extrapolation error bars are not controlled.","tokens_in":22894,"tokens_out":5750,"would_cite":true,"duration_ms":52840,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper claims that a level-24 numerical solution in the Schnabl gauge converges to the analytic tachyon vacuum, with extrapolated energy within 0.05 percent of -1.","keywords":["tachyon vacuum","Schnabl gauge","level truncation","open string field theory","Sen's first conjecture","Ellwood invariant","Newton's method","gauge fixing"],"falsifier":"Compute the same Schnabl-gauge solution at levels 26, 28, and 30 with the projector-based Newton method and rerun the maximal-order $1/L$ polynomial extrapolations; if the predicted infinite-level energy moves away from $-1$ by more than the stated $10^{-4}$ error, or if the coefficients stop tracking the analytic values, the claimed identification is falsified.","tokens_in":21737,"feed_emoji":"⚛️","tokens_out":10605,"duration_ms":93224,"temperature":0.7,"pith_summary":"This paper tries to establish that the numerical tachyon-vacuum solution obtained by level truncation in the Schnabl gauge becomes the analytic tachyon vacuum as the level goes to infinity. To get there, the authors develop a projector-based Newton method that lets them solve the level-truncated equations of motion up to level 24, far beyond the previous level-10 limit. Their evidence is that the vacuum energy has its predicted local minimum at level 12 and extrapolates to about -0.9995, matching -1 at roughly the 0.05 percent level; the gauge-invariant overlap extrapolates to about -0.0015, within 0.15 percent of zero; and the equations of motion that were projected out during gauge fixing vanish asymptotically. If this identification is right, Sen's first conjecture holds numerically in the Schnabl gauge, with somewhat lower precision than in the Siegel gauge.","feed_headline":"Schnabl-gauge numerics put tachyon vacuum energy within 0.05% of -1","feed_subtitle":"The extrapolated level-24 solution matches the analytic vacuum: energy -0.9995, gauge-invariant overlap -0.0015.","key_machinery":"The load-bearing machinery is a canonical projector $P_C$ constructed from the matrix $G$ of the gauge condition $B_0\\Psi = 0$. The projector divides string-field coefficients into dependent and independent variables, so Newton's method can solve only the projected equations of motion without ever storing the full cubic vertex; this reduction is what makes level 24 reachable when storing the full vertex would need more than a petabyte. The second piece of machinery is the extrapolation ansatz $q_M^{(L_{\\min},L_{\\max})}(L) = a_0 + \\sum_{n=1}^{M} a_n/L^n$, supplemented by a second fit in $1/N$ of order $M=5$, used to estimate infinite-level values of the energy, the coefficients, and the Ellwood invariant.","core_discovery":"On its own terms, the central claim is that the level-truncated numerical solution in the Schnabl gauge, computed up to level $L=24$ with a new projector-based Newton's method, can be identified with the analytic tachyon-vacuum solution at infinite level. The two solutions differ noticeably at finite levels, but maximal-order polynomial extrapolations in $1/L$ send the vacuum energy to $-0.9995$, the Ellwood invariant to $-0.0015$, the tachyon vacuum expectation value and other coefficients toward their analytic values, and the out-of-gauge equations of motion to zero. The paper therefore concludes that, while the numerical precision is lower than in the Siegel gauge, the Schnabl-gauge solution satisfies Sen's first conjecture at the level of numerical evidence.","pith_inferences":["If the instability of the extrapolations is caused by the gauge condition coupling fields at different levels, then other non-diagonal linear b-gauges should show similar or worse convergence as they move away from the Siegel gauge; the same projector method could test this quantitatively.","The two-stage $1/L$-then-$1/N$ extrapolation that works for the tachyon coefficient suggests the finite-level corrections follow a systematic asymptotic expansion; if so, the same procedure could sharpen predictions for all coefficients and for other observables.","The candidate second solution whose energy extrapolates near zero may be a level-truncation artifact or a gauge copy of the vacuum; continuing it to higher levels with the new method would decide which.","The failure of the KBc restricted-space truncation suggests that numerical level truncation cannot be trusted for highly constrained ansatze; a fully analytic treatment is needed there, so any future numerical check of KBc-type solutions would need a different projection."],"forward_implications":["Sen's first conjecture holds numerically in the Schnabl gauge: the extrapolated vacuum energy is $-0.9995$ (about $0.05\\%$ from $-1$) and the Ellwood invariant is $-0.0015$ (about $0.15\\%$ from zero).","The predicted local minimum of the energy at level 12 is confirmed directly from the data, and the tachyon vev minimum predicted at level 26 is ruled out; the data suggest the minimum is closer to levels 42-46.","The projector-based Newton method is general for linear b-gauges, so high-level numerical solutions can now be searched for in gauges other than Siegel and Schnabl.","The projected-out equations of motion and the symmetry identities of the analytic solution are satisfied asymptotically, which supports calling the numerical solution physical."],"supporting_citations":[{"why":"Defines the Schnabl gauge, constructs the analytic tachyon vacuum solution, and supplies the expected coefficients, identities, and energy the numerical solution is compared with.","marker":"[2]"},{"why":"Introduces the level-truncation scheme for open bosonic string field theory and the Siegel-gauge approach that the new gauge handling extends.","marker":"[34]"},{"why":"Provides the experimental level-truncation methodology, including Newton's method and the idea of out-of-gauge equations.","marker":"[35]"},{"why":"Supplies the previous level-10 numerical Schnabl-gauge solution and the predictions this paper confirms (energy minimum at level 12) and revises (tachyon vev minimum).","marker":"[36]"},{"why":"Provides the Siegel-gauge level-30 energy data and the out-of-gauge equation baseline used for comparison.","marker":"[37]"},{"why":"Describes the numerical algorithms and extrapolation rules the paper follows, including the choice of maximal-order fits.","marker":"[38]"},{"why":"Introduced the $1/L$ extrapolation ansatz used to estimate infinite-level values.","marker":"[42]"},{"why":"Supplies the conservation laws used to evaluate the Ellwood invariant from the numerical solution.","marker":"[45]"}],"fun_headline_variants":["Schnabl gauge tachyon vacuum: numerics hit -0.9995 energy","Numerics in Schnabl gauge approach tachyon vacuum energy -1","Level-24 Schnabl solution: energy -0.9995, moving to -1","Tachyon vacuum in Schnabl gauge: numerical energy approaches -1"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The infinite-level conclusions rest on assuming the sequence of numerical results from levels up to 24 can be smoothly extrapolated by polynomial curves in the reciprocal of the level, an assumption the paper itself finds only partially stable.","fun_headline_variants_meta":{"raw":{"variants":["Schnabl gauge tachyon vacuum: numerics hit -0.9995 energy","Numerics in Schnabl gauge approach tachyon vacuum energy -1","Level-24 Schnabl solution: energy -0.9995, moving to -1","Tachyon vacuum in Schnabl gauge: numerical energy approaches -1"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001057,"raw_usage":{"total_tokens":4397,"prompt_tokens":866,"completion_tokens":3531,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":482,"completion_tokens_details":{"reasoning_tokens":3444}},"tokens_in":482,"tokens_out":3531,"duration_ms":20862,"temperature":1.0,"reasoning_tokens":3444,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:16:42.517004+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the same Schnabl-gauge solution at levels 26, 28, and 30 with the projector-based Newton method and rerun the maximal-order $1/L$ polynomial extrapolations; if the predicted infinite-level energy moves away from $-1$ by more than the stated $10^{-4}$ error, or if the coefficients stop tracking the analytic values, the claimed identification is falsified.","supporting_citations":[{"cited_title":"Numerical solution of open string field theory in Schnabl gauge","cited_arxiv_id":"1707.09452","evidence_quote":"Supplies the previous level-10 numerical Schnabl-gauge solution and the predictions this paper confirms (energy minimum at level 12) and revises (tachyon vev minimum)."},{"cited_title":"Level Truncation Approach to Open String Field Theory,","cited_arxiv_id":null,"evidence_quote":"Describes the numerical algorithms and extrapolation rules the paper follows, including the choice of maximal-order fits."}],"review_version":1}