# Rank-Two Motif Spectral Architecture Library

- ID: 1230
- Canonical URL: https://synthcore.org/idea/1230/rank-two-motif-spectral-architecture-library
- API JSON: https://synthcore.org/api/idea/1230.json
- API Markdown: https://synthcore.org/api/idea/1230.md
- Verification status: unverified
- Source: [arXiv:2607.21294](https://arxiv.org/abs/2607.21294)
- Category: architecture
- Solves: accuracy, stability, scalability
- ML areas: graph-nn, architecture, initialization
- Math tags: graph-theory, spectral-theory, combinatorics, dynamical-systems
- Ratings: usefulness 5/10; difficulty 6/10; novelty 7/10

## Idea description

Use the paper's three rank-two graph families as a small, analytically understood library of propagation topologies. Select or mix figure-eight, theta, and dumbbell edge-routing motifs to obtain different effective receptive-field growth rates while retaining an exact spectral-radius target for normalization and architecture search.

## Mathematical statement

A rank-two core has one of three kernel types. A figure-eight with cycle lengths a,b has exponential rate alpha determined by 2/(alpha^a+1)+2/(alpha^b+1)=1, equivalently (alpha^a-1)(alpha^b-1)=4 and R_(a,b)(alpha)=0 with R_(a,b)(x)=x^(a+b)-x^a-x^b-3. A theta graph with path lengths a,b,c satisfies 1/(alpha^a+1)+1/(alpha^b+1)+1/(alpha^c+1)=1 and has polynomial Theta_(a,b,c)(x)=x^(a+b+c)-x^a-x^b-x^c-2. A dumbbell with cycle lengths a,b and corridor c satisfies alpha^(2c)(alpha^a-1)(alpha^b-1)=4 and has polynomial D_(a,b,c)(x)=x^(a+b+2c)-x^(a+2c)-x^(b+2c)+x^(2c)-4. In every case alpha>1 is the unique root and log(alpha) is the entropy. These formulas allow discrete motif hyperparameters to be mapped to exact propagation growth before training.

## Key formulas

- $$\frac{2}{\alpha^a+1}+\frac{2}{\alpha^b+1}=1,\qquad (\alpha^a-1)(\alpha^b-1)=4,$$
- $$\frac{1}{\alpha^a+1}+\frac{1}{\alpha^b+1}+\frac{1}{\alpha^c+1}=1,$$
- $$\alpha^{2c}(\alpha^a-1)(\alpha^b-1)=4,$$
- $$\delta=\log(\alpha)$$

## Implementation notes

Implement a motif mixer as a sparse graph layer over a set of directed edge channels. Choose a small candidate set, for example figure-eight `(a,b) in {(1,1),(1,2),(2,2)}`, theta `(a,b,c) in {(1,1,1),(1,1,2),(1,2,2)}`, and dumbbell `(a,b,c) in {(1,1,1),(1,2,1),(2,2,1)}`. For each candidate, instantiate its directed non-backtracking transition matrix B_m and a separate linear map W_m. Compute alpha_m offline by bisection over x in `[1+1e-6, 10]` using the corresponding displayed equation; increase the upper bound until the polynomial changes sign. Normalize each branch as `u_m = B_m z W_m / alpha_m`, then combine `u = sum_m softmax(beta)_m u_m`, where beta are learned architecture logits. Pseudocode: `for m in motifs: alpha[m]=root_equation(m); z_m=NBProp(B_m,z); z_m=z_m/alpha[m]; u += softmax(beta)[m]*Linear_m(z_m); h=Update(h,Aggregate(u))`. The paper supplies exact alpha values for the ideal motif; on arbitrary input graphs, estimate the actual rho(B_m) by 15-step power iteration and optionally use `alpha_eff=(1-lambda)*alpha_m+lambda*rho_hat`. Measure computed math versus empirical behavior separately: root solving gives the designed entropy, while validation loss and feature norms test whether that entropy is useful. First cheap experiment: train 2-4 layer models on synthetic stochastic block graphs and molecular TU datasets, comparing a standard GIN/GCN with a parameter-matched motif mixer. Report accuracy versus depth, activation amplification, and sparse FLOPs. A useful signal is that the learned beta selects different motifs by graph family and achieves equal accuracy with fewer layers or fewer edges processed.

## Disclaimer

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