{"paper_id":"2611.20550","title":"Mollification Regularization and Asymptotic Correction for Analytical Expansion of BSDEs with Non-Smooth Terminal Conditions. I. Theory","authors":["Liangliang Zhang"],"paper_date":"2026-11-16","paper_type":"theoretical","paper_url":"https://arxiv.org/pdf/2611.20550","page_url":"https://thequant.space/flowcharts/mollification-regularization-and-asymptotic-correction-for-analytical-expansion/","scores":{"empirical_rigor":2,"math_complexity":9,"quadrant":"Lab Rats"},"proposes":{"status":"available","statement":"The paper proposes a mollification regularization framework that replaces a non‑smooth terminal condition ψ by its convolution ψ_ε = ψ * ρ_ε and provides joint error estimates separating regularization error and expansion error.","quote":"Contributions of This Paper The main contributions of this paper are fivefold: (1) Mollification framework with joint error estimates. We introduce a systematic mol-lification regularization approach that replaces the non-smooth terminal condition ψ by its convolution ψε = ψ ∗ρε with a standard mollifier of bandwidth ε > 0.","summary":"The analytical expansion method for BSDEs of Zhang (2017) breaks down when the terminal condition psi has kinks or discontinuities. We replace psi by psi_epsilon = psi * rho_epsilon and derive a joint error decomposition whose expansion constants explicitly track the derivative blow-up as epsilon…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"Uniform local‑to‑global Taylor gain (Assumption 4.8) that aggregates Taylor‑truncation defects to C ε^{-(m‑2l‑α)} δ^{(m‑2l)/2}_t with a constant uniform along the time grid, including terminal steps.","type":"other","stated":"explicit","quote":"Assumption 4.8 (Uniform local-to-global Taylor gain) For the parameters under consideration, the Taylor-truncation defects of the (l, m, n) expansion aggregate to Cε−(m−2l−α)δ(m−2l)/2 t , with a constant uniform along the time grid, including the terminal steps. Assumption 4.8 is standard in form for smooth data with ε-independent constants, but is not proved here with ε-dependent constants.","support":"full"},{"assumption":"The terminal condition ψ belongs to C^{0,α}(ℝ^d) with 0 < α ≤ 1.","type":"data","stated":"explicit","quote":"Theorem 4.9 (Conditional expansion error with ε-dependence) Let Assumptions 2.1–2.2 and 4.8 hold, let ψ ∈C0,α(Rd), 0 < α ≤1, and assume m > 2l, 0 ≤l ≤1, and ε2 ≥δt.","support":"full"},{"assumption":"Expansion orders satisfy m > 2l and 0 ≤ l ≤ 1 (or more generally l ≥ 0, m > 2l).","type":"model","stated":"explicit","quote":"Theorem 4.9 (Conditional expansion error with ε-dependence) Let Assumptions 2.1–2.2 and 4.8 hold, let ψ ∈C0,α(Rd), 0 < α ≤1, and assume m > 2l, 0 ≤l ≤1, and ε2 ≥δt. Proposition 4.11 (Unconditional additive propagation bound) Assume the hypotheses of Theo-rem 4.9, except Assumption 4.8, and write q = m −2l.","support":"full"},{"assumption":"Bandwidth ε satisfies ε^2 ≥ δ_t.","type":"other","stated":"explicit","quote":"Theorem 4.9 (Conditional expansion error with ε-dependence) Let Assumptions 2.1–2.2 and 4.8 hold, let ψ ∈C0,α(Rd), 0 < α ≤1, and assume m > 2l, 0 ≤l ≤1, and ε2 ≥δt.","support":"full"}]},"datasets":{"status":"none_found","items":[],"known":[],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"none"},"transaction_costs":{"status":"not_applicable","modeled":"n/a","headline_net_of_costs":null,"detail":null,"quote":null,"source":"front_matter","rule":"R3","candidates":{"front_matter":"n/a","evidence_card":"n/a","evidence_card_quoted":false},"meaning":"the paper reports no trading or execution result"},"results":{"status":"available","evaluation_type":"none","findings":[{"finding":"Under the uniform local‑to‑global Taylor gain, the benchmark (l,m,α) = (0,2,1) yields the balanced bandwidth ε* ≈ δ_t^{1/3} and a weak convergence rate of 2/3.","kind":"theoretical","quote":"The associated balanced rates are conditional on an explicitly stated uniform local-to-global Taylor-propagation assumption: under that assumption the benchmark (l, m, alpha) = (0, 2, 1) gives epsilon* ~ delta_t^{1/3} and weak rate 2/3. For (l, m, α) = (0, 2, 1) these formal values are δ1/3 t and 2/3.","support":"full"},{"finding":"Without Assumption 4.8, the additive bound gives no decay when m‑2l = 2, and only a positive polynomial rate for Hölder data when m‑2l > 2.","kind":"theoretical","quote":"Without it, the rigorous additive bound gives no decay when m - 2l = 2; an unconditional positive rate is obtained only for m - 2l > 2. Proposition 4.11 instead supplies an unconditional bound and a positive rate for q > 2.","support":"partial"},{"finding":"Asymptotic correction terms cancel the weak bias through order ε^{2J}, provided auxiliary expansion errors and Picard defect satisfy additional rate conditions.","kind":"theoretical","quote":"Asymptotic corrections cancel the weak bias through order epsilon^{2J}; their faster exponents are formal targets unless the auxiliary expansion errors and the Picard defect satisfy additional rate conditions, which we state explicitly. (3) Asymptotic correction terms and sufficient rate conditions.","support":"full"},{"finding":"For symmetric mollifiers the leading weak bias cancels, giving a regularization weak error of order O(ε^2).","kind":"theoretical","quote":"For symmetric mollifiers the leading weak bias cancels. What improves is the regularization error: for symmetric kernels the first-order bias cancels and the weak (pointwise, t < T) bias becomes O(ε2) (Proposition B.3).","support":"full"},{"finding":"The expansion error constants grow with dimension d through derivative order χ ≥ m + d + 2, limiting practical applicability to moderate dimensions.","kind":"theoretical","quote":"The expansion error constants, however, depend on d through the derivative order χ ≥m + d + 2 and combinatorial multi-index factors (Remark 2.15); accordingly, the practical scope of the method is moderate dimensions, as reflected in the numerical experiments reported in Part II (Zhang, 2026b). 3.","support":"full"}],"reported_performance":null,"out_of_sample":{"method":"none","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":16,"citations":{"cited_by_in_corpus":0,"by_independent_authors":0,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":0,"mentions":0,"unclassified":0},"independent_papers":[],"scope":"Citations between papers in this corpus only; citations from outside it are not counted."}},"implementation":{"status":"available","public":false,"code_url":null,"note":"No repository presented by the authors as their own was found in the paper. Absence here is not proof that no code exists."},"replication":{"status":"none_found","card":null,"note":"No replication of this paper has been run by thequant.space yet."},"extraction":{"text_source":"fulltext","model":"openai/gpt-oss-120b","version":3,"extracted_at":"2026-11-17T15:29:16Z"},"provenance":{"claims":"Everything here is what the authors state in the paper. Nothing was re-run or reproduced by the platform.","quotes":"Assumptions, datasets, cost detail and findings were extracted by a language model. Every quote is verbatim paper text. Where an item has `support`, \"full\" means every figure in it appears in the quoted sentences and \"partial\" means a figure is stated elsewhere in the paper; items with a figure found nowhere in the paper were discarded.","datasets_known":"datasets.known lists well-known vendors and public datasets by a stable id, found by a dictionary over the full text where the sentence says the data is used; it is independent of the model and covers only names in the dictionary.","flags":"transaction_costs.modeled, results.evaluation_type and results.out_of_sample.method are platform classifications (structured-decision models); see https://thequant.space/score-guide/#penalty-flags","comparisons":"supports / contradicts / qualifies are a language model's reading of two papers' quoted findings, restricted to closely related papers that share no author. They are pointers to read, not verified replications, and coverage is partial.","implementation":"code_url is a repository the paper itself presents as its own (rule-scored, ambiguous cases judged by a model), checked for liveness when it was recorded."},"docs":"https://thequant.space/agents/"}