The Equilibrium Is the Initialization: Lazy Identity Collapse in Physics-Structured Deep Equilibrium Reasoning

• URL: http://arxiv.org/abs/2607.11116v1
• Date: Mon, 13 Jul 2026 05:45:35 GMT
• Title: The Equilibrium Is the Initialization: Lazy Identity Collapse in Physics-Structured Deep Equilibrium Reasoning
• Abstract summary: Deep equilibrium models promise input-adaptive implicit computation.<n>We report a cautionary study of a port-Hamiltonian DEQ with learned reasoning tasks.<n>We show that the standard zeroing ablation is confounded and gives wildly seed-dependent answers.
• Score: 0.0
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Deep equilibrium models promise input-adaptive implicit computation: harder problems should demand more solver iterations, and the solved equilibrium should encode the result of genuine iterative inference. We report a cautionary study of a port-Hamiltonian DEQ with a learned initialization on two reasoning tasks -- ProofWriter entailment over frozen DeBERTa embeddings and a BFS-verified graph-reachability benchmark -- in which the implicit computation is a silent no-op. Across tasks, seeds, and controlled ablation arms, the solved equilibrium equals the solver's start point to numerical precision, and bypassing the solver entirely changes test accuracy by +0.00 percentage points in 18 of 19 training runs. Controlled interventions falsify the tempting explanation: removing the anchoring term reproduces every result, and retraining with noise-decoupled starts yields a solver that converges to the noisy start while the decoder learns to ignore it. The single escaping run diverges instead ($\|h^{*}-z_0\|=171$), producing a co-adapted noise channel whose removal improves accuracy. Iteration counts are uncorrelated with ground-truth difficulty ($r=0.009$), and the full apparatus never outperforms a two-layer MLP on either task. We trace the mechanism to gradient starvation along two distinct routes, show that the standard zeroing ablation is confounded and gives wildly seed-dependent answers where the correct substitution test gives a stable zero, and distill a four-test diagnostic protocol for auditing claimed implicit computation. All experiments run on a single free Colab GPU; code, raw logs, and analysis scripts are released.

Related papers

• Diffusion as a Training Curriculum for Timestep-Free Iterative Reasoning [1.16829782470264]
  We introduce a hidden state diffusion denoiser that can be run to arbitrary depth.<n>We reach 99.90% exact solve on Sudoku-Extreme and 98.93% on Maze-Unique.<n>We find that diffusion's primary contribution to our anytime solver is not a sampling procedure at inference, but a denoising training curriculum.
  arXiv  Detail & Related papers  (2026-09-01T15:55:01Z)
• Caliber: Cross-Architecture Extraction-Cost Control for Score-Returning APIs [27.21644231867863]
  Caliber is a defense against model extraction that formulates noise selection as a calibration problem.<n>To defend against an attacker that uses returned scores for knowledge distillation, Caliber adds independent and identically distributed Gaussian noise to the internal logits.<n>Caliber normalizes noise variance by the squared median top-two logit margin and fits the resulting noise-utility relationship with a logistic curve.
  arXiv  Detail & Related papers  (2026-08-02T05:58:00Z)
• Search as Computation Allocation [52.080489413778]
  Many algorithms spend an internal resource before returning a decision and are evaluated only by the quality of that terminal output.<n>We formalize such procedures as terminal computation-allocation problems.<n>Bellman equations characterize optimal allocation under fixed budgets, priced computation, and exact certification.
  arXiv  Detail & Related papers  (2026-07-30T08:45:14Z)
• Cost Accounting for Reactive Computational Graphs: Exhaustive Sweeps, Sequential Mutation, and the Backward-Locality Gap [0.0]
  We show how Julia handles interventions on a neural network's computational graph.<n>We prove the exact cost of a sequence of persistent mutations, never undone between insertions.<n>We also show how batched application is order-independent and sub-additive.
  arXiv  Detail & Related papers  (2026-07-18T00:53:22Z)
• Stabilizing Extrapolation in Looped Transformers via Learned Stochastic Stopping [56.14767235650558]
  Looped Transformers, which repeatedly apply a shared transformer block, are an architecturally natural fit for variable-length computation tasks.<n>We trace this variance to the spurious correlation in simple algorithmic tasks between sequence length and number of loops.<n>Our work suggests that "when to stop" should be treated as a training design choice, not merely an inference-time-allocation rule.
  arXiv  Detail & Related papers  (2026-06-29T08:58:09Z)
• Data and Evaluation Closed-Loop for Model Capability Enhancement [8.245706308161504]
  We show that evaluation can be routine, auditable, and experimentally validated rather than intuitive.<n>We show that the evaluation-to-data inference can be routine, auditable, and experimentally validated rather than intuitive.
  arXiv  Detail & Related papers  (2026-06-26T14:45:57Z)
• Survive or Collapse: The Asymmetric Roles of Data Gating and Reward Grounding in Self-Play RL [76.45061154544568]
  Self-play reinforcement learning trains language models on their own generated tasks, co-evolving a proposer and solver without human labels.<n>Recent systems report strong reasoning gains, but collapse and instability are widely observed and poorly understood.<n>We argue instead that self-play stability is governed by two distinct levers: a data-level gate that decides which proposer-generated tasks enter the training pool, and the reward signal that updates the policy on tasks already admitted.
  arXiv  Detail & Related papers  (2026-05-21T09:19:23Z)
• PAIR-CI: Calibrated Conditional Independence Testing for Causal Discovery with Incomplete Data [0.0]
  We introduce PAIR-CI, a nonparametric conditional independence (CI) test that restores calibration by integrating multiple imputation directly into the inferential procedure.<n>A provably consistent variance estimator jointly accounts for uncertainty arising from cross-validation and multiple imputation.
  arXiv  Detail & Related papers  (2026-05-06T12:34:37Z)
• Correction and Corruption: A Two-Rate View of Error Flow in LLM Protocols [51.56484100374058]
  We propose a paired-outcome measurement interface for auditing a single protocol step on exact-match tasks.<n>For each instance, the interface records a baseline correctness bit and a post-step correctness bit.<n>These rates predict accuracy changes and define a reusable empirical interface testable across seeds, mixtures, and pipelines.
  arXiv  Detail & Related papers  (2026-04-20T13:25:40Z)
• Test-Time Scaling with Diffusion Language Models via Reward-Guided Stitching [66.39914384073145]
  We propose a self-consistency framework that turns cheap diffusion-sampled reasoning into a reusable pool of step-level candidates.<n>We find that step-level recombination is most beneficial on harder problems.<n>Our training-free framework improves average accuracy by up to 2 across six math and coding tasks.
  arXiv  Detail & Related papers  (2026-02-26T11:08:39Z)
• Learning Shrinks the Hard Tail: Training-Dependent Inference Scaling in a Solvable Linear Model [2.7074235008521246]
  We analyze neural scaling laws in a solvable model of last-layer fine-tuning where targets have intrinsic, instance-heterogeneous difficulty.<n>We show that learning shrinks the hard tail'' of the error distribution.
  arXiv  Detail & Related papers  (2026-01-07T10:00:17Z)
• INC: An Indirect Neural Corrector for Auto-Regressive Hybrid PDE Solvers [61.84396402100827]
  We propose the Indirect Neural Corrector ($mathrmINC$), which integrates learned corrections into the governing equations.<n>$mathrmINC$ reduces the error amplification on the order of $t-1 + L$, where $t$ is the timestep and $L$ the Lipschitz constant.<n>We test $mathrmINC$ in extensive benchmarks, covering numerous differentiable solvers, neural backbones, and test cases ranging from a 1D chaotic system to 3D turbulence.
  arXiv  Detail & Related papers  (2025-11-16T20:14:28Z)
• Reinforcement Learning with Verifiable yet Noisy Rewards under Imperfect Verifiers [90.50039419576807]
  Reinforcement Learning with Verifiable Rewards (RLVR) trains policies against automated verifiers to avoid costly human labeling.<n>To reduce vulnerability to verifier hacking, many RLVR systems collapse rewards to binary $0,1$ during training.<n>This choice carries a cost: it introduces textitfalse negatives (rejecting correct answers, FNs) and textitfalse positives (accepting incorrect ones, FPs)
  arXiv  Detail & Related papers  (2025-10-01T13:56:44Z)
• Abduct, Act, Predict: Scaffolding Causal Inference for Automated Failure Attribution in Multi-Agent Systems [20.846301581161978]
  Failure attribution in multi-agent systems is a critical yet unsolved challenge.<n>Current methods treat this as a pattern recognition task over long conversation logs.<n>A2P Scaffolding transforms failure attribution from pattern recognition into a structured causal inference task.
  arXiv  Detail & Related papers  (2025-09-12T16:51:15Z)
• Differentiable Bilevel Programming for Stackelberg Congestion Games [47.60156422249365]
  In a Stackelberg congestion game (SCG), a leader aims to maximize their own gain by anticipating and manipulating the equilibrium state at which the followers settle by playing a congestion game. Here, we attempt to tackle this computational challenge by marrying traditional methodologies with the latest differentiable programming techniques in machine learning. We propose two new local search algorithms for SCGs. The first is a gradient descent algorithm that obtains the derivatives by unrolling ILD via differentiable programming. The second algorithm adds a twist by cutting short the followers' evolution trajectory.
  arXiv  Detail & Related papers  (2022-09-15T21:32:23Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.