# Minkowski-Additive Convex Latents

- ID: 206
- Canonical URL: https://synthcore.org/idea/206/minkowski-additive-convex-latents
- API JSON: https://synthcore.org/api/idea/206.json
- API Markdown: https://synthcore.org/api/idea/206.md
- Verification status: beats_baseline
- Source: [arXiv:2608.26615](https://arxiv.org/abs/2608.26615)
- Category: architecture
- Solves: stability, generalization, scalability
- ML areas: world-model, graph-nn, regularization, training-dynamics
- Math tags: convex-analysis, geometry, metric-geometry
- Ratings: usefulness 6/10; difficulty 4/10; novelty 8/10

## Idea description

Store a convex object as a direction-indexed vertex tuple and implement composition of objects through componentwise Minkowski addition and nonnegative scaling. This creates a structured residual or compositional layer where convexification is nonexpansive, making perturbation amplification controllable and avoiding repeated generic geometric optimization.

## Mathematical statement

For a compact convex set K, define the multivalued directional projector $\pi^{\flat}_{\boldsymbol{a}}(K)=\{\boldsymbol{x}:x_i\in\arg\max_{x\in K}a_i^T x\ \text{for all }i\}$, where a_i are fixed directions. Lemma 29 gives exact compatibility with Minkowski addition and nonnegative scaling: $\pi^{\flat}_{\boldsymbol{a}}(K+L)=\pi^{\flat}_{\boldsymbol{a}}(K)+\pi^{\flat}_{\boldsymbol{a}}(L)$ and $\pi^{\flat}_{\boldsymbol{a}}(\lambda K)=\lambda\pi^{\flat}_{\boldsymbol{a}}(K)$, with tuple operations componentwise. Thus if x_i and y_i are selected directional support maximizers, x_i plus y_i represents the corresponding support maximizer of K plus L. The decoded set is $P(\boldsymbol{x})=\operatorname{co}\{x_i\}$. Lemma 6 states that convexification is 1-Lipschitz: perturbing the point tuple cannot increase the induced directed set distance beyond the tuple perturbation. Arbitrary compositions can reach the boundary of the cone, so the implementation should use a strict margin or a projection step.

## Key formulas

- $$\pi^{\flat}_{\boldsymbol{a}}(K):=\{\boldsymbol{x}\in\mathbb{R}^{d\times N}:x_i\in\arg\max_{x\in K}a_i^T x\ \text{for all }i\},$$
- $$\pi^{\flat}_{\boldsymbol{a}}(K+L)=\pi^{\flat}_{\boldsymbol{a}}(K)+\pi^{\flat}_{\boldsymbol{a}}(L),\qquad \pi^{\flat}_{\boldsymbol{a}}(\lambda K)=\lambda\pi^{\flat}_{\boldsymbol{a}}(K),\quad \lambda\geq0,$$
- $$\operatorname{dist}(\operatorname{co}\{\boldsymbol{x}\},\operatorname{co}\{\boldsymbol{y}\})\leq\operatorname{dist}(\{\boldsymbol{x}\},\{\boldsymbol{y}\}),$$
- $$\boldsymbol{z}=\alpha\boldsymbol{x}+\beta\boldsymbol{y},\quad z_i=\alpha x_i+\beta y_i,\quad P_{\mathrm{out}}=\operatorname{co}\{z_1,\ldots,z_N\},\qquad \alpha,\beta\geq0.$$

## Implementation notes

1. Integration point: add a structured convex-object composition block to a world model, relational graph network, or diffusion denoiser that combines multiple predicted convex regions. Represent each object by a tensor X with shape batch by N by d, where N indexes fixed directions and d indexes coordinates. 2. Pseudocode: encode objects as X and Y; predict nonnegative mixture weights alpha and beta using softplus of two scalar network outputs; form z_i equal to alpha times x_i plus beta times y_i for every direction i; apply one or more projected-gradient steps toward the constraints a_i transpose z_i minus a_i transpose z_j at least epsilon; decode with a convex-hull layer or support queries. Add a cone-margin penalty consisting of the sum over i not equal to j of max(0, epsilon minus a_i transpose z_i plus a_i transpose z_j) squared. 3. The paper supplies the exact Minkowski and scaling rules and the nonexpansive convexification guarantee. Estimate empirical error amplification by perturbing X and Y, measuring the maximum tuple perturbation, and comparing it with the Hausdorff distance between decoded outputs. Do not assume uniqueness on the cone boundary. 4. First experiment: compose pairs of random two-dimensional and three-dimensional convex polygons, train a small predictor to match their Minkowski sum, and compare this block with a generic MLP followed by convex-hull decoding. Evaluate Hausdorff error, runtime, gradient norm through repeated compositions, and error after a chain of 5 to 10 compositions. Success is near-linear error growth with chain length, lower gradient variance, and lower latency than repeatedly solving support-point or polygon-sum optimization.

## Verification

- Status: beats_baseline
- Mechanism evidence: yes
- Mechanism confirmed: yes
- Practical verdict: helps
- Verdict: Built a NumPy MVP for direction-indexed convex latents with componentwise nonnegative Minkowski addition, scaling, convex-hull decoding, and Hausdorff evaluation. Across 300 trials, additive and scaling support identities had zero measured error, while convexification had maximum perturbation ratio 1.0, confirming the claimed nonexpansiveness. In chain tests of length 2–10, structured addition was 7.5–19.5x faster than repeatedly forming pairwise vertex sums and hulls, although finite-direction decoding introduced nonzero approximation error that reached about 0.13 in the longest test.

### Mechanism check

- Verdict: Built a NumPy MVP for direction-indexed convex latents with componentwise nonnegative Minkowski addition, scaling, convex-hull decoding, and Hausdorff evaluation. Across 300 trials, additive and scaling support identities had zero measured error, while convexification had maximum perturbation ratio 1.0, confirming the claimed nonexpansiveness. In chain tests of length 2–10, structured addition was 7.5–19.5x faster than repeatedly forming pairwise vertex sums and hulls, although finite-direction decoding introduced nonzero approximation error that reached about 0.13 in the longest test.
- Confidence: 8/10
- Limitations: Only small 2D random polygons were tested. No learned predictor, projected-gradient margin enforcement, 3D experiment, GPU benchmark, or MLP baseline was included. The generic baseline computes exact hull-based sums, while the structured method uses a finite direction set and trades exactness for speed.

### Practical benchmark

- Paired seeds: 8
- Baseline mean: 0
- Idea mean: 0
- p-value: 0.0081
- Paired wins: 8/8
- Benchmark verdict: idea better (significant)

## Artifacts

- [bench_report.json](https://synthcore.org/code/1203/bench_report.json)
- [minkowski_latents.py](https://synthcore.org/code/1203/minkowski_latents.py)
- [minkowski_track.py](https://synthcore.org/code/1203/minkowski_track.py)
- [report.md](https://synthcore.org/code/1203/report.md)
- [report_bench_2026-09-03T035727.md](https://synthcore.org/code/1203/report_bench_2026-09-03T035727.md)
- [results.json](https://synthcore.org/code/1203/results.json)
- [stage2_bench.py](https://synthcore.org/code/1203/stage2_bench.py)
- [Download all files as ZIP](https://synthcore.org/download/1203)

## Disclaimer

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