# Low-Rank Radiation Resonant State Space

- ID: 2042
- Canonical URL: https://synthcore.org/idea/2042/low-rank-radiation-resonant-state-space
- API JSON: https://synthcore.org/api/idea/2042.json
- API Markdown: https://synthcore.org/api/idea/2042.md
- Verification status: unverified
- Source: [arXiv:2608.09367](https://arxiv.org/abs/2608.09367)
- Category: dynamics
- Solves: stability, memory, accuracy
- ML areas: ssm, rnn, training-dynamics
- Math tags: dynamical-systems, spectral-theory, linear-algebra, control-theory
- Ratings: usefulness 6/10; difficulty 5/10; novelty 6/10

## Idea description

Replace an unconstrained recurrent transition with a second-order resonant state whose restoring matrix is full-rank but whose damping is low-rank. The low-rank damping creates a small set of rapidly controlled bright modes and a large dark subspace with long memory, while a small optional damping term prevents numerical drift in completely dark modes.

## Mathematical statement

The paper reduces the singular resolvent to the effective matrix \(\mathcal{M}_{\delta}^{\pm}(\omega)=\delta K-\omega^{2}I_{m}\mp\mathrm{i}\delta\omega\Gamma_{0}+\mathcal{O}(\delta^{2}+\delta\omega^{2})\), where \(m\) is the reduced resonant dimension, \(K\) is the static stiffness matrix, \(I_m\) is the identity, \(\delta>0\) is the contrast parameter, \(\omega\) is frequency, and \(\Gamma_0\) is a positive semidefinite radiation matrix. The paper states that \(\operatorname{rank}\Gamma_0=3\) and \(\dim\ker\Gamma_0=6N-3\), because radiation factors through a three-dimensional total-force map. Transfer this as \(\Gamma_0=BB^{\mathsf T}\), with \(B\in\mathbb{R}^{m\times r}\) and \(r\ll m\). The corresponding time-domain oscillator is \(\ddot h+\delta BB^{\mathsf T}\dot h+\delta Kh=Ux(t)\), whose Fourier operator is exactly the adapted effective matrix. Bright modes are in \(\operatorname{range}(B)\), while dark modes are in \(\ker(B^{\mathsf T})\).

## Key formulas

- $$\mathcal{M}_{\delta}^{\pm}(\omega)=\delta K-\omega^{2}I_{m}\mp\mathrm{i}\delta\omega\Gamma_{0}+\mathcal{O}(\delta^{2}+\delta\omega^{2}),\qquad \operatorname{rank}\Gamma_{0}=3,\quad \dim\ker\Gamma_{0}=6N-3.$$
- $$\Gamma_0=BB^{\mathsf T},\qquad \ddot h(t)+\delta BB^{\mathsf T}\dot h(t)+\delta Kh(t)=Ux(t).$$
- $$\widehat h(\omega)=\left(\delta K-\omega^2I_m-\mathrm{i}\delta\omega BB^{\mathsf T}\right)^{-1}U\widehat x(\omega).$$
- $$\ddot h+\delta BB^{\mathsf T}\dot h+\delta Kh+\eta P_{\mathrm{dark}}\dot h=Ux(t),\qquad P_{\mathrm{dark}}=I_m-B(B^{\mathsf T}B+\epsilon I_r)^{-1}B^{\mathsf T}.$$

## Implementation notes

(1) Integration point: implement this as a recurrent or state-space block between the input projection and output projection. Map each input token \(x_t\in\mathbb{R}^{d_{in}}\) to resonant coordinates with \(Ux_t\). Maintain displacement \(h_t\in\mathbb{R}^m\) and velocity \(v_t\in\mathbb{R}^m\). Parameterize \(K=LL^{\mathsf T}+\kappa I_m\), with \(\kappa>0\), and parameterize damping as \(BB^{\mathsf T}\), where \(B\in\mathbb{R}^{m\times r}\) and \(r\ll m\). (2) Pseudocode: compute \(a_t=Ux_t\); compute \(p_t=B^{\mathsf T}v_t\); compute the regularized dark projector \(P_{dark}=I-B(B^{\mathsf T}B+\epsilon I)^{-1}B^{\mathsf T}\); update \(v_{t+1/2}=v_t+\Delta t(a_t-\delta Kh_t-\delta Bp_t-\eta P_{dark}v_t)\); update \(h_{t+1}=h_t+\Delta t v_{t+1/2}\); then use a linear or gated readout of \(h_{t+1}\), optionally with a residual path from \(x_t\). Use semi-implicit integration if explicit Euler becomes unstable. (3) The mathematical structure fixes the low-rank damping and bright/dark decomposition. Estimate empirically the rank \(r\), stiffness spectrum, \(\delta\), \(\eta\), and step size. Monitor the eigenvalues of the discrete transition matrix and the condition number of \(\delta K-\omega^2I_m-\mathrm{i}\delta\omega BB^{\mathsf T}\) over frequencies found by an FFT of training sequences. (4) First experiment: use \(m=64\), \(r=4\), and two layers on sequential MNIST, the Copying task, and a synthetic long-memory task requiring a delayed weighted sum. Compare against a same-width GRU and a diagonal SSM with matched parameter count and optimizer. Measure validation loss at equal FLOPs, gradient norm, hidden-state retention versus sequence length, and the largest discrete-transition eigenvalue. The idea succeeds if it improves long-context loss or memory retention without exploding gradients, and if low-rank damping outperforms both full-rank damping and no damping at equal parameter count.

## Disclaimer

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