Upload matrixpfn-base v0.1.0

README.md

@@ -0,0 +1,65 @@
+---
+library_name: matrixpfn
+tags:
+  - preconditioner
+  - sparse-linear-systems
+  - graph-neural-network
+  - pytorch
+---
+
+# matrixpfn-base
+
+GNN-based learned preconditioner for sparse linear systems.
+
+**Version**: 0.1.0
+
+## Usage
+
+```python
+import numpy as np
+from scipy.io import mmread
+from matrixpfn import MatrixPFN
+
+pfn = MatrixPFN.from_pretrained("Csed-dev/matrixpfn-base")
+
+A = mmread("matrix.mtx")  # any scipy sparse matrix
+b = A @ np.random.randn(A.shape[0])
+
+result = pfn.solve(A, b)  # accepts scipy sparse directly
+print(f"Converged: {result.converged} in {result.iterations} iterations")
+```
+
+## Architecture
+
+| Parameter | Value |
+|-----------|-------|
+| Network | ContextResGCN |
+| Layers | 12 |
+| Embed | 64 |
+| Hidden | 256 |
+| Context pairs | 10 |
+| Parameters | 419,074 |
+| dtype | float32 |
+
+## Training
+
+- **epochs**: 2000
+- **best_loss**: 0.071401
+- **loss_function**: l1_direct
+- **batch_size**: 512
+- **domains**: diffusion, diffusion_advection
+- **grid_sizes**: [16, 24, 32, 48]
+
+## Benchmark
+
+| domain | grid | converged | avg_iters | avg_residual |
+|--------|------|-----------|-----------|--------------|
+| diffusion | 16x16 | 0/20 | 300.0 | 5.06e-02 |
+| diffusion | 24x24 | 0/20 | 300.0 | 3.02e-02 |
+| diffusion | 32x32 | 0/20 | 300.0 | 2.51e-02 |
+| diffusion | 48x48 | 0/20 | 300.0 | 1.71e-02 |
+| diffusion | 64x64 | 0/20 | 300.0 | 9.03e-03 |
+| diffusion_advection | 16x16 | 20/20 | 87.7 | 6.12e-09 |
+| diffusion_advection | 24x24 | 7/20 | 263.1 | 5.34e-04 |
+| diffusion_advection | 32x32 | 0/20 | 300.0 | 5.15e-03 |
+| diffusion_advection | 48x48 | 0/20 | 300.0 | 2.14e-02 |