✗ Failed on benchmark 2026

ISS-Certified Sampled Optimizer Wrapper

Implementation & benchmark of arXiv:2607.18500 — Large-Signal Stability Analysis of Optimization-Based Secondary Control for Distributed Energy Resources

Usefulness7/10

Difficulty6/10

Novelty7/10

Dynamics Stability Accuracy Optimizer Rnn Ssm Training dynamics Dynamical systems Control theory Optimization Numerical analysis

Source paper: Large-Signal Stability Analysis of Optimization-Based Secondary Control for Distributed Energy Resources arXiv:2607.18500 ⓘ · analyzed Aug 30, 2026

AI-generated research hypothesis, automatically tested. Not peer-reviewed.

Idea description

Wrap a recurrent or state-space neural network in a sampled-data feedback loop: latent states evolve continuously or at every fine solver step, while a constrained optimizer updates the control, adapter, or residual-gating vector only every M steps. Between optimizer updates, use zero-order hold or linear interpolation and reject updates that violate a learned Lyapunov decrease condition. This should prevent large transient latent explosions caused by aggressive optimizer updates while retaining larger update steps than a globally contractive architecture.

Formulas

$$x_{k+1}=F_T(x_k,u_k,d_k),\qquad u_{k+1}=\Pi_{\mathcal U}\left[u_k-\eta g_k\right]$$

$$u(t)=(1-\tau)u_k+\tau u_{k+1},\qquad \tau=\frac{t-kT}{T},\quad kT\leq t<(k+1)T$$

$$V(x_{k+1})-V(x_k)\leq-\alpha(\|x_k\|)+\gamma(\|d_k\|)$$

$$\Delta V_k+\lambda V(x_k)-c\|d_k\|^2\leq0,\qquad V(x)=x^\top P x,\quad P\succ0$$

Mathematical statement

Let x_k be the neural latent state at optimizer sample k, u_k the constrained control or residual-gating vector, d_k an external perturbation, and T the sampling interval. The sampled closed loop is x_{k+1}=F_T(x_k,u_k,d_k), with optimizer update u_{k+1}=Pi_U[u_k-eta g_k], where Pi_U projects onto feasible set U and g_k is a gradient or policy direction. Between samples, the applied control is u(t)=(1-tau)u_k+tau u_{k+1}, with tau=(t-kT)/T, or u(t)=u_k under zero-order hold. The transferable stability condition is an input-to-state inequality: V(x_{k+1})-V(x_k)<=-alpha(||x_k||)+gamma(||d_k||), where V is positive definite, alpha is a positive dissipation function, and gamma measures disturbance amplification. For V(x)=x^T P x with P positive definite, use the empirical certificate Delta V_k+lambda V(x_k)-c||d_k||^2<=0, where lambda>0 and c>=0. Accept an update only when this inequality holds; otherwise reduce eta or shrink the trust region. The predicted ultimate bound is limsup V(x_k)<=(c/lambda)sup_k||d_k||^2.

Implementation notes

1. Integration point: use a GRU, residual RNN, or neural state-space model whose latent state x is updated at every fine step. Add a low-dimensional vector u controlling residual strength, damping, normalization gain, or an input-feedback correction. Every M fine steps, compute a gradient direction g_k from the task loss, a prediction-error objective, or a one-step MPC surrogate; constrain u to a box or norm ball. 2. Pseudocode: initialize P as diagonal positive, u in U, and sampling period T=M*dt. At sample k, save x_k and V_k=x_k^T P x_k; propose u_prop=project_U(u-eta*g). Roll the network forward for M steps using linear interpolation or hold between u and u_prop. Compute x_next, DeltaV=x_next^T P x_next-V_k, and a disturbance estimate d_k from input residuals or teacher-forcing error. Accept u_prop if DeltaV+lambda*V_k-c*||d_k||^2<=0; otherwise halve eta and retry, up to a fixed number of attempts. 3. Computed from the paper's mechanism: sampled update, constrained projection, interpolation, Lyapunov decrease, and ISS residual. Estimated empirically: P, lambda, c, and the worst-case local dissipation margin. Fit P by minimizing certificate violations over rollout data, then validate it on held-out initial states. 4. First cheap experiment: compare a 2-layer tanh RNN on sequential MNIST or a synthetic nonlinear oscillator-prediction task against the same RNN with Adam and gradient clipping. Sweep eta and M with bounded impulse disturbances. The quantitative prediction is a sharp degradation boundary when the accepted dissipation margin crosses zero; the maximum stable eta should decrease as M increases, and the empirical ultimate bound should scale approximately linearly with disturbance energy, with V plateau near (c/lambda)*sup||d_k||^2.

Verification

Failed on benchmark

Stage 1 · Toy mechanism gate: Passed ✓

Stage 2 · Mechanism transferred to benchmark: Not reproduced ✗

Stage 2 · Practical benchmark result: No measurable effect

Methodology: 8 paired seeds, permutation test (p<0.05) against a learning-rate-tuned baseline. How verification works

Stage 1 — Mechanism check agent confidence 8/10

Built an ISS-certified sampled optimizer wrapper with box projection, linear control interpolation, Lyapunov-certificate rejection/backtracking, and safe fallback control. All three toy predictions were confirmed: the fixed-control stability boundary matched |a-u|<1 within grid resolution, the certificate sampling transition occurred at M=2 as predicted, and disturbance plateau energy scaled exactly as d^2. The wrapper reduced the baseline's state explosion from 2.22e22 maximum magnitude to a maximum of 1.0 and final magnitude 2.09e-3, demonstrating a real stabilization effect in this toy setting.

Agent confidence: 8/10

Baseline: Scalar sampled system, M=1, eta=0.35, d=0.005: final_abs=2.22e22, max_abs=2.22e22, u=3.0.
Idea: Same setup with certification: final_abs=2.09e-3, max_abs=1.0, 92 accepted and 8 rejected updates, mean eta=0.32025, u=3.0.
Prediction checks: Stability boundary: predicted [0.20,2.20], observed [0.25,2.20] on a 0.05 grid. Sampling transition: predicted and observed minimum M=2. Disturbance scaling: V/d^2=4.0 for every tested amplitude.

Limitations: Only a scalar linear state-space toy was tested, not a GRU/SSM trained on a task. The optimizer direction is synthetic, P is fixed to the scalar identity rather than learned, disturbances are constant, and no FLOP/accuracy or larger-scale neural-network comparison was performed.

How to run: python3 iss_sampled_optimizer.py

Stage 2 — Benchmark vs. tuned baseline (arXiv:2607.18500) Sep 2, 2026

No significant difference vs. the tuned baseline (−-0.0%, p=1, wins 0 / 8 paired seeds).

Benchmark: Dynamics forecasting
Model: rnn_small
Paired seeds: 8
Baseline mean: 0.0006
Idea mean: 0.0006
Effect (Δ): 0 (−-0.0%; negative = idea better)
Wins: 0 / 8 paired seeds
p-value: 1 (permutation test, 20 000 shuffles)
Smallest detectable effect: ±0.0%
Mechanism: Not confirmed ✗
Practical effect: No effect
Baseline tuning: swept over 3 configs

Limitations:

Only the built-in dynamics track, rnn_small model, M=1, and 18 epochs were tested. The certificate did not encounter aggressive or unstable proposals, so larger sampling intervals, stronger updates, learned P, disturbances, and transient-state stability under held-out rollouts were not evaluated.

How to run: python3 stage2_iss_bench.py

Verdict computed by deterministic test code from paired-seed statistics — not by the language model.

Stage 2 — Benchmark Aug 31, 2026 latest Failed ✗

Agent confidence: 9/10

Implemented the ISS-certified sampled optimizer as a local modified training loop on the structurally matched dynamics track using rnn_small, paired seeds, equal-budget learning-rate sweeps, and trained-model Lyapunov measurements. The certificate was satisfied on every checked update, so no updates were rejected and the idea matched baseline exactly: delta_mean=0 and permutation p=1.0. This is no neural-training win.

Baseline: dynamics/rnn_small; best Adam lr=0.01; test MSE mean=0.0005546553857, std=0.0002139676899
Idea: ISS-certified loop; best shared-grid lr=0.01, M=1; test MSE mean=0.0005546553857, std=0.0002139676899; paired delta=0.0, p=1.0
Mechanism signature: Across 8 seeds, every seed had 72 accepted, 0 rejected, and 0 certificate violations. Mean certificate values ranged from -0.5226766 to -0.4818276; prediction that rejected updates would be observed was not confirmed.

Benchmark result