# Fisher-Geodesic Finite-Step Annealing

- ID: 2733
- Canonical URL: https://synthcore.org/idea/2733/fisher-geodesic-finite-step-annealing
- API JSON: https://synthcore.org/api/idea/2733.json
- API Markdown: https://synthcore.org/api/idea/2733.md
- Verification status: mech_ok_no_baseline
- Source: [arXiv:2608.25207](https://arxiv.org/abs/2608.25207)
- Category: sampling
- Solves: speedup, stability, sample-efficiency
- ML areas: diffusion-sampling, scheduler, training-dynamics
- Math tags: statistical-mechanics, information-theory, geometry, stochastic-processes, dynamical-systems
- Ratings: usefulness 7/10; difficulty 5/10; novelty 7/10

## Idea description

Replace uniformly spaced diffusion or energy annealing schedules by a finite sequence of quench-relax stages whose intermediary distributions are approximately equally spaced in Fisher-Rao distance. Each stage abruptly changes the model energy or noise level and then runs a short relaxation phase; the schedule concentrates stages where the distribution changes most sharply. This should reduce nonequilibrium mismatch at a fixed number of sampler evaluations and avoid large distributional jumps that destabilize Langevin or reverse-diffusion integration.

## Mathematical statement

For Hamiltonian H(x), inverse temperature beta, and state x, the equilibrium distribution is pi^H(x)=exp(-beta H(x))/Z_H, where Z_H=sum_x exp(-beta H(x)). For an arbitrary distribution rho, the excess free energy satisfies F_H(rho)-F_H(pi^H)=beta^{-1}D_KL(rho||pi^H), so nonequilibrium error is measured by KL divergence. For a differentiable path p_theta(x), the Fisher-Rao line element is ds^2=dtheta^T I(theta)dtheta, where I(theta)=E_{x~p_theta}[grad_theta log p_theta(x) grad_theta log p_theta(x)^T] is the Fisher information matrix. The paper's finite-protocol result states that optimal intermediary equilibria converge to a Fisher-Rao geodesic as the number N of intermediary states increases. Thus choose theta_k so that the arc lengths Delta s_k are approximately equal, with Delta s_k approximately L/N and L=int_0^1 sqrt(theta'(u)^T I(theta(u))theta'(u))du. Locally, KL(p_{theta_k}||p_{theta_{k+1}})=1/2 Delta s_k^2+O(Delta s_k^3), so equal arc-length spacing gives cumulative leading-order mismatch approximately L^2/(2N), whereas a nonuniform parameter schedule wastes steps in low-curvature regions.

## Key formulas

- $$\pi^{H}(x)=\frac{e^{-\beta H(x)}}{Z_H},\qquad Z_H=\sum_x e^{-\beta H(x)},\qquad F_H(\rho)-F_H(\pi^H)=\beta^{-1}D_{\mathrm{KL}}(\rho\|\pi^H).$$
- $$ds^2=d\theta^{\mathsf T}I(\theta)d\theta,\qquad I(\theta)=\mathbb E_{x\sim p_\theta}\left[\nabla_\theta\log p_\theta(x)\nabla_\theta\log p_\theta(x)^{\mathsf T}\right].$$
- $$\theta_k:\quad s(\theta_k)\approx \frac{k}{N}L,\qquad L=\int_0^1\sqrt{\dot\theta(u)^{\mathsf T}I(\theta(u))\dot\theta(u)}\,du.$$
- $$\sum_{k=0}^{N-1}D_{\mathrm{KL}}(p_{\theta_k}\|p_{\theta_{k+1}})=\frac{1}{2}\sum_{k=0}^{N-1}(\Delta s_k)^2+O(N^{-2})\approx\frac{L^2}{2N}.$$

## Implementation notes

1. Integration point: use this as a scheduler for a diffusion sampler, annealed Langevin sampler, or energy-based model. Let p_theta denote the model distribution along a one-dimensional annealing path theta in [0,1], where theta can be log-SNR, inverse temperature, noise scale, or an interpolation coefficient between a base Gaussian and the data distribution. Keep the neural score or energy network unchanged and replace only the usual linear or cosine schedule with an adaptive list of N schedule values.

2. Pseudocode: construct a dense pilot grid theta[0:M]. At every grid point estimate the scalar Fisher speed v_j=sqrt(E[(d/dtheta log p_theta(x))^2]) using minibatch samples and automatic differentiation through the energy, score, or denoising parameterization. Form cumulative lengths s_j=s_{j-1}+v_j(theta_j-theta_{j-1}), then select N+1 schedule points by inverse interpolation at s=k*s_M/N. During sampling, quench to theta_k and apply r_k relaxation updates, such as x<-x+epsilon*s_theta_k(x)+sqrt(2epsilon)xi for Langevin. Initially use constant r_k; later test r_k proportional to the local estimated KL or Fisher speed.

3. Computed from the paper are the Fisher metric, equal-arc-length rule, and inverse-N leading-order dissipation scaling. Estimated empirically are Fisher speeds, one-step KL proxies, and relaxation errors. A scalar or diagonal Fisher estimate is sufficient for the first implementation.

4. First cheap experiment: compare linear, cosine, and Fisher-geodesic schedules using a small pretrained CIFAR-10 diffusion score model at N=10, 20, 40 function evaluations. Measure FID, reverse-process KL proxies, and cumulative one-step KL. The quantitative prediction is cumulative KL approximately C/N for the geodesic schedule, while each local KL should be nearly equal across stages. The measured local-KL coefficient of variation should be lower than for parameter-uniform schedules, and the fitted exponent in cumulative KL versus N should approach -1 over the discretization-dominated regime.

## Verification

- Status: mech_ok_no_baseline
- Mechanism evidence: yes
- Mechanism confirmed: yes
- Practical verdict: harms
- Verdict: The Fisher identity matched Monte Carlo within 0.066%. Equal-Fisher spacing reduced local-KL coefficient of variation to approximately zero, versus 0.406 to 4.531 for linear spacing, and the widest-range cumulative KL approached the predicted L^2/(2N) scaling.

### Mechanism check

- Verdict: The Fisher identity matched Monte Carlo within 0.066%. Equal-Fisher spacing reduced local-KL coefficient of variation to approximately zero, versus 0.406 to 4.531 for linear spacing, and the widest-range cumulative KL approached the predicted L^2/(2N) scaling.
- Confidence: 9/10
- Limitations: Only a one-dimensional exact Gaussian path was tested. No neural diffusion model, Langevin relaxation, FID, runtime, noisy Fisher estimation, or multidimensional learned-energy experiment was tested.

### Practical benchmark

- Paired seeds: 8
- Baseline mean: 0
- Idea mean: 0
- p-value: 0.03075
- Paired wins: 1/8
- Benchmark verdict: idea worse (significant)

## Artifacts

- [bench_report.json](https://synthcore.org/code/1082/bench_report.json)
- [extra_check.py](https://synthcore.org/code/1082/extra_check.py)
- [fisher_annealing_toy.py](https://synthcore.org/code/1082/fisher_annealing_toy.py)
- [fisher_bench_track.py](https://synthcore.org/code/1082/fisher_bench_track.py)
- [report.md](https://synthcore.org/code/1082/report.md)
- [report_bench_2026-09-02T000156.md](https://synthcore.org/code/1082/report_bench_2026-09-02T000156.md)
- [run_bench.py](https://synthcore.org/code/1082/run_bench.py)
- [toy_results.json](https://synthcore.org/code/1082/toy_results.json)
- [Download all files as ZIP](https://synthcore.org/download/1082)

## Disclaimer

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