Reproduce “Multiresolution Graph Transformers and Wavelet Positional Encoding for Learning Hierarchical Structures” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Nhat Khang Ngo, Truong Son Hy, Risi Kondor
MGT predicts DFT-derived polymer properties using atom-level attention, learned atom groups, and a Transformer over those groups. Wavelet positional encoding supplies multiscale graph context. On the tested polymer dataset, MGT with WavePE reports GAP test mean absolute error 0.0387 electron volts, below the 0.043 electron-volt chemical-accuracy target; this result applies to the full system, not a demonstrated speedup.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.