AI Paper Atlas

Research map / Probabilistic and scientific modeling / Variational and generative sampling

Riemannian Variational Flow Matching for Material and Protein Design

Olga Zaghen, Floor Eijkelboom, Alison Pouplin, Cong Liu, Max Welling, Jan-Willem van de Meent, Erik J Bekkers

ICLR 2026 · Probabilistic and scientific modeling › Variational and generative sampling · approach: Flow matching and flow sampling · 13 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use endpoint (variational) geodesic-distance loss instead of velocity lossRFM (intrinsic) · Spherical checkerboard (synthetic)Coverage (%): 66.83 -> 84.21Improved
use endpoint (variational) geodesic-distance loss instead of velocity lossRFM (intrinsic) · Hyperboloid checkerboard (synthetic)Coverage (%): 60.75 -> 68.38Improved
use Riemannian Gaussian (geodesic) loss instead of Euclidean MSE with extrinsic priorVFM · Spherical checkerboard (synthetic)Coverage (%): 79.08 -> 83.10Improved
use Riemannian Gaussian loss instead of Euclidean MSE with extrinsic priorVFM · Hyperboloid checkerboard (synthetic)C2ST: 57.03 -> 63.55Hurt
Riemannian geometry instead of Euclidean for extrinsic modelsCFM/VFM · Spherical checkerboard (synthetic)distance to manifold: 0.044 -> 0.010Improved
variationalize the rotational loss with geodesic distanceMOFFlow · MOF structure prediction (stol=0.5, 1 sample)match rate (%): 30.40 -> 33.52Improved
variationalize the rotational lossMOFFlow · MOF stol=1.0, 5 samplesmatch rate (%): 100.0 -> 97.18Hurt
variationalize the rotational lossQFlow · Protein backbonedesignability fraction: 0.924 -> 0.968Improved
variationalize the rotational lossReQFlow · Protein backbonescRMSD: 0.939 -> 0.961 (reproduced); paper 1.071 -> 0.961Mixed
set sigma_t(x)=1-t instead of 1 (ablation)RG-VFM · synthetic checkerboardssample qualityNo clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy