{"paper_id":"2501.12600","title":"Breaking the Dimensional Barrier for Constrained Dynamic Portfolio Choice","authors":["Jeonggyu Huh","Jaegi Jeon","Hyeng Keun Koo","Byung Hwa Lim"],"paper_date":"2025-01-22","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2501.12600v6","page_url":"https://thequant.space/flowcharts/breaking-the-dimensional-barrier-for-constrained-dynamic-por/","scores":{"empirical_rigor":7.5,"math_complexity":9,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes a scalable, policy‑centric framework that integrates neural policies with Pontryagin's Maximum Principle and a log‑barrier‑regularized Hamiltonian to enforce inequality constraints, yielding the PG‑DPO algorithm and its barrier‑projected variant P‑PGDPO.","quote":"We propose a scalable, policy-centric framework for continuous-time multi-asset portfolio-consumption optimization under inequality constraints. Our method integrates neural policies with Pontryagin's Maximum Principle (PMP) and enforces feasibility by maximizing a log-barrier-regularized Hamiltonian at each time-state pair, thereby satisfying KKT conditions without value-function grids.","summary":"We propose a scalable, policy-centric framework for continuous-time multi-asset portfolio-consumption optimization under inequality constraints. Our method integrates neural policies with Pontryagin’s Maximum Principle (PMP) and enforces feasibility by maximizing a log-barrier-regularized…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"The market follows the standard continuous‑time self‑financing condition with Brownian motions.","type":"market","stated":"explicit","quote":"according to the standard continuous-time self-financing condition: dXt = \u0010 Xt π⊤ t eµ −Ct \u0011 dt + Xt π⊤ t ev dWt, X0 = x0 > 0, where Wt = (W1,t, ..., Wn,t)⊤is an Rn-valued standard Brownian motion adapted to the filtra-tion F = {Ft} under the filtered probability space (Ω, F, P).","support":"full"},{"assumption":"The investor maximizes expected discounted utility of consumption and terminal wealth with a strictly increasing, strictly concave C² utility function.","type":"model","stated":"explicit","quote":"The investor’s aim is to maximize the expected discounted utility from consumption over time and terminal wealth at the planning horizon T: J(π, C) = E \u0014Z T 0 e−ρtU(Ct) dt + Ke−ρT U(XT ) \u0015 , where ρ > 0 is the subjective discount rate and K ≥0 determines the relative weight placed on terminal wealth. We assume that the utility function U : R+ →R is C2, strictly increasing, and strictly concave.","support":"full"},{"assumption":"Inequality constraints such as non‑negative portfolio weights and bounded consumption satisfy standard interior‑point regularity (LICQ, Slater’s condition, strict complementarity).","type":"other","stated":"explicit","quote":"Such constraints are standard in interior-point methods and Karush–Kuhn–Tucker (KKT) analyses (Boyd, 2004; Nesterov and Nemirovskii, 1994; Bertsekas, 1997). (Slater’s condition is already assumed in Assumption 1(ii).)","support":"full"},{"assumption":"Barrier‑regularized Hamiltonian maximization at each time‑state pair is well‑posed with bounded level sets and smoothness near the solution.","type":"other","stated":"explicit","quote":"In addition, for each (t, x) we assume: (i) Bounded level sets for the barrier objective: For any fixed εbar > 0, the superlevel sets of e Hbar are bounded in (π, C). (iii) Smoothness near the solution and central-path interiority: The gradients ∇(π,C)H and ∇(π,C)Γj are locally Lipschitz in a neighborhood of the solution.","support":"full"}]},"datasets":{"status":"none_found","items":[],"known":[],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"synthetic"},"transaction_costs":{"status":"not_applicable","modeled":"n/a","headline_net_of_costs":null,"detail":null,"quote":null,"source":"evidence_card","rule":"R2","candidates":{"front_matter":"unclear","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":"PG‑DPO and P‑PGDPO recover KKT‑optimal policies in short‑sale and consumption‑cap problems while maintaining strict feasibility across dimensions, with runtime scaling linearly in the number of assets.","kind":"theoretical","quote":"Empirically, PG-DPO and its projected variant (P-PGDPO) recover KKT-optimal policies in canonical short-sale and consumption-cap problems while maintaining strict feasibility across dimensions; unlike PDE/BSDE solvers, runtime scales linearly with the number of assets and remains practical at n=100.","support":"full"},{"finding":"The barrier‑projected variant reduces policy errors by up to two orders of magnitude relative to activation‑only training and remains numerically stable for 100 or more assets.","kind":"performance","quote":"The barrier-projected variant (P-PGDPO) reduces policy errors by up to two orders of magnitude relative to activation-only training and remains numerically stable in large-scale environments with 100 or more assets. Across all dimensions, the manifold-projection variant P-PGDPO enforces strict feasibility and first-order optimality, reducing policy error by up to two orders of magni-tude relative to the baseline","support":"full"},{"finding":"Value‑based solvers become computationally infeasible beyond moderate dimensions, whereas the proposed methods remain practical up to 100–1000 assets.","kind":"theoretical","quote":"This scalability represents a fundamental advance over value-based solvers, which become computationally infeasible even at moderate dimensions. Crucially, the computational complexity of both algorithms scales linearly with the number of assets, enabling accurate and feasible training in large-scale environments with 100 or even 1000 assets—a regime where classical DP, HJB, or deep-BSDE methods become","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":17,"citations":{"cited_by_in_corpus":2,"by_independent_authors":0,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":0,"mentions":0,"unclassified":2},"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":2,"extracted_at":"2026-11-14T19:52:46Z"},"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/"}