https://hunterheidenreich.com/research/gutenocr-grounded-vision-language-frontend/2026-02-21T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/guten-mascot.webpGutenOCR Mascothttps://hunterheidenreich.com/research/2026-03-07T00:00:00+00:00monthly0.5https://hunterheidenreich.com/research/deconstructing-recurrence-attention-gating/2026-02-21T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/deconstructing-sequence-prediction/multiscale-lorenz.webpForecasting comparison of different neural architectures on the Multiscale Lorenz-96 systemhttps://hunterheidenreich.com/projects/2026-02-26T00:00:00+00:00monthly0.5https://hunterheidenreich.com/posts/2026-02-26T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/2026-02-26T00:00:00+00:00monthly0.5https://hunterheidenreich.com/videos/2026-02-26T00:00:00+00:00monthly0.5https://hunterheidenreich.com/leaderboards/2026-02-26T00:00:00+00:00monthly0.5https://hunterheidenreich.com/tags/graph-theory/2026-03-24T00:00:00+00:00monthly0.5https://hunterheidenreich.com/categories/machine-learning-fundamentals/2026-03-24T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/machine-learning/model-architectures/neural-scaling-of-deep-chemical-models/2026-03-24T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/neural-scaling-deep-chemical-models-cover.webpLog-log plots showing power-law scaling of ChemGPT validation loss versus model size and GNN force field loss versus dataset sizehttps://hunterheidenreich.com/tags/neural-networks/2026-03-24T00:00:00+00:00monthly0.5https://hunterheidenreich.com/tags/optimization/2026-03-24T00:00:00+00:00monthly0.5https://hunterheidenreich.com/tags/simulation/2026-03-24T00:00:00+00:00monthly0.5https://hunterheidenreich.com/tags/algorithms/2026-03-23T00:00:00+00:00monthly0.5https://hunterheidenreich.com/tags/benchmark/2026-03-23T00:00:00+00:00monthly0.5https://hunterheidenreich.com/tags/cheminformatics/2026-03-23T00:00:00+00:00monthly0.5https://hunterheidenreich.com/categories/computational-chemistry/2026-03-23T00:00:00+00:00monthly0.5https://hunterheidenreich.com/tags/generative-models/2026-03-23T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/computational-chemistry/molecular-modeling/genetic-algorithms-molecule-generation-baselines/2026-03-23T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/ga-molecules-cover.webpDiagram showing a genetic algorithm for molecules where a parent albuterol molecule undergoes mutation to produce two child molecules, with a selection and repeat loophttps://hunterheidenreich.com/tags/machine-learning/2026-03-23T00:00:00+00:00monthly0.5https://hunterheidenreich.com/tags/molecular-representation/2026-03-23T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/computational-chemistry/chemical-language-models/molgensurvey-molecule-design/2026-03-23T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/molgensurvey-cover.webpTaxonomy diagram showing the three axes of MolGenSurvey: molecular representations (1D string, 2D graph, 3D geometry), generative methods (deep generative models and combinatorial optimization), and eight generation tasks (1D/2D and 3D)https://hunterheidenreich.com/notes/computational-chemistry/benchmark-problems/smina-docking-benchmark/2026-03-23T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/smina-docking-benchmark-cover.webpBar chart comparing SMINA docking scores of CVAE, GVAE, and REINVENT against a random ZINC 10% baseline across eight protein targetshttps://hunterheidenreich.com/tags/survey/2026-03-23T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/computational-chemistry/benchmark-problems/tartarus-inverse-molecular-design/2026-03-23T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/tartarus-inverse-molecular-design-cover.webp2D structure of a phenyl-quaterthiophene, a conjugated organic molecule representative of the photovoltaic donor materials benchmarked in the Tartarus platformhttps://hunterheidenreich.com/notes/computational-chemistry/chemical-language-models/tied-two-way-transformers-retrosynthesis/2026-03-23T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/tied-two-way-tf-cover.webpDiagram of the tied two-way transformer architecture with shared encoder, retro and forward decoders, latent variables, and cycle consistency, alongside USPTO-50K accuracy and validity resultshttps://hunterheidenreich.com/tags/transformers/2026-03-23T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/computational-chemistry/chemical-language-models/bartsmiles-molecular-representations/2026-03-22T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/bartsmiles-ablation-summary.webpBARTSmiles ablation study summary showing impact of pre-training strategies on downstream task performancehttps://hunterheidenreich.com/notes/computational-chemistry/chemical-language-models/lm-complex-molecular-distributions/2026-03-22T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/lm-complex-distributions-cover.webpThree distribution plots showing RNN language models closely matching training distributions across peaked, multi-modal, and large-scale molecular generation tasks while graph models failhttps://hunterheidenreich.com/notes/computational-chemistry/chemical-language-models/limo-latent-inceptionism/2026-03-22T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/limo-cover.webpDiagram of the LIMO pipeline showing gradient-based reverse optimization flowing backward through a frozen property predictor and VAE decoder to optimize the latent space zhttps://hunterheidenreich.com/notes/computational-chemistry/chemical-language-models/regression-transformer/2026-03-22T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/regression-transformer-cover.webpRegression Transformer dual-masking concept showing property prediction (mask numbers) and conditional generation (mask molecules) in a single modelhttps://hunterheidenreich.com/notes/computational-chemistry/chemical-language-models/retmol-retrieval-molecule-generation/2026-03-22T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/retmol-cover.webpDiagram of the RetMol pipeline showing input molecule and retrieval database feeding into a frozen encoder, cross-attention fusion module, and frozen decoder to produce optimized molecules with iterative refinementhttps://hunterheidenreich.com/notes/computational-chemistry/chemical-language-models/uncorrupt-smiles/2026-03-22T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/uncorrupt-smiles-cover.webpDiagram showing the UnCorrupt SMILES pipeline: invalid SMILES are corrected by a transformer seq2seq model into valid SMILES, with correction rates of 62-95% across generator typeshttps://hunterheidenreich.com/projects/kabsch-horn-cookbook/2026-03-20T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/scientific-computing/kabsch-alignment-before-and-after.webpBefore and after visualization of point-set alignment using the Kabsch algorithmhttps://hunterheidenreich.com/tags/molecular-dynamics/2026-03-20T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/computational-chemistry/chemical-language-models/molgen-molecular-generation-chemical-feedback/2026-03-20T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/molgen-overview.webpMolGen overview showing two-stage pre-training (molecular language syntax learning and domain-agnostic prefix tuning) and chemical feedback paradigmhttps://hunterheidenreich.com/tags/python/2026-03-20T00:00:00+00:00monthly0.5https://hunterheidenreich.com/tags/pytorch/2026-03-20T00:00:00+00:00monthly0.5https://hunterheidenreich.com/categories/scientific-computing/2026-03-20T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/computational-chemistry/llms-for-chemistry/2026-03-18T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/computational-chemistry/chemical-language-models/molecular-transformer/2026-03-18T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/molecular-transformer-overview.webpMolecular Transformer architecture showing atom-wise tokenized SMILES input through encoder-decoder with multi-head attention to predict reaction productshttps://hunterheidenreich.com/notes/interdisciplinary/computational-biology/arun-svd-point-fitting/2026-03-16T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/arun-cover.webpThree-panel diagram showing input point sets, SVD factorization of the cross-covariance matrix, and the aligned resulthttps://hunterheidenreich.com/categories/computational-biology/2026-03-16T00:00:00+00:00monthly0.5https://hunterheidenreich.com/tags/computer-vision/2026-03-16T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/computational-chemistry/benchmark-problems/activity-cliffs-benchmark/2026-03-17T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/activity-cliffs-cover.webpActivity cliffs benchmark showing method rankings by RMSE on cliff compounds, with SVM plus ECFP outperforming deep learning approacheshttps://hunterheidenreich.com/notes/interdisciplinary/computational-biology/horn-orthonormal-matrices/2026-03-16T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/horn88-cover.webpDiagram showing the polar decomposition of the cross-covariance matrix M into orthonormal factor U and positive semidefinite square roothttps://hunterheidenreich.com/notes/computational-chemistry/chemical-language-models/molformer/2026-03-17T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/molformer-cover.webpMoLFormer-XL architecture diagram showing SMILES tokens flowing through a linear attention transformer to MoleculeNet benchmark results and attention-structure correlationhttps://hunterheidenreich.com/notes/computational-chemistry/chemical-language-models/selformer/2026-03-17T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/selformer-cover.webpSELFormer architecture diagram showing SELFIES token input flowing through a RoBERTa transformer encoder to molecular property predictionshttps://hunterheidenreich.com/notes/interdisciplinary/computational-biology/umeyama-similarity-transformation/2026-03-16T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/umeyama-cover.webpSide-by-side comparison showing naive SVD producing a reflected alignment versus Umeyama's corrected proper rotationhttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-graph/adaptmol-2026/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/adaptmol-2026-cover.webpAdaptMol domain adaptation pipeline showing encoder-decoder with MMD alignment between labeled source and unlabeled target domain imagesThe AdaptMol domain adaptation pipeline aligns bond features via MMD between source and target domains, with validated predictions used for self-training.https://hunterheidenreich.com/tags/bioinformatics/2026-03-15T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/machine-learning/generative-models/consistency-models/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/consistency-models-cover.webpDiagram showing consistency models mapping points on a PF ODE trajectory to the same originConsistency models learn to map any point on the Probability Flow ODE trajectory back to its origin, enabling single-step generation.https://hunterheidenreich.com/notes/machine-learning/generative-models/discrete-diffusion-models/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/discrete-diffusion-models-cover.webpD3PM forward and reverse processes on a quantized swiss roll with uniform, Gaussian, and absorbing transition matricesD3PM forward and learned reverse processes applied to a quantized swiss roll, showing uniform, discretized Gaussian, and absorbing-state corruption.https://hunterheidenreich.com/tags/diffusion/2026-03-15T00:00:00+00:00monthly0.5https://hunterheidenreich.com/categories/generative-modeling/2026-03-15T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/rule-based/graphreco-2026/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/graphreco-2026-cover.webpGraphReco system architecture showing component extraction, atom and bond ambiguity resolution, and graph reconstruction stagesThe GraphReco pipeline: component extraction, probabilistic ambiguity resolution via Markov network, and molecule graph assembly.https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-graph/grasp-2026/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/grasp-2026-cover.webpGraSP feed-forward architecture showing GNN, FiLM-conditioned CNN, and MLP classification headThe GraSP architecture: a GNN produces a graph embedding that conditions a CNN via FiLM layers, with an MLP head predicting subgraph membership.https://hunterheidenreich.com/notes/interdisciplinary/computational-biology/horn-absolute-orientation/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/horn-cover.webp3D scatter plot showing left and right point sets with rotation axis and quaternion rotation archttps://hunterheidenreich.com/notes/interdisciplinary/computational-biology/kabsch-algorithm/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/kabsch-cover.webp3D scatter plot showing source points, target points, and Kabsch-aligned points overlapping the targetshttps://hunterheidenreich.com/notes/machine-learning/generative-models/latent-diffusion-models/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/latent-diffusion-models-cover.webpLDM architecture diagram showing conditioning via concatenation and cross-attentionLDMs condition on various inputs via concatenation or cross-attention layers injected into the UNet backbone.https://hunterheidenreich.com/tags/optical-chemical-structure-recognition/2026-03-15T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/benchmarks-and-reviews/uni-parser-2025/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/uni-parser-2025-cover.webpUni-Parser pipeline diagram showing document pre-processing, layout detection, semantic parsing, content gathering, and format conversion stagesThe Uni-Parser pipeline converts unstructured PDFs into clean, hierarchical, multimodal outputs.https://hunterheidenreich.com/notes/machine-learning/model-architectures/can-recurrent-neural-networks-warp-time/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/rnn-warp-time-cover.webpThree-panel diagram showing an original sequence, its time-warped version, and the gate values derived from requiring time warping invariancehttps://hunterheidenreich.com/tags/comparative-analysis/2026-03-14T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/machine-learning/model-architectures/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/machine-learning/model-architectures/relational-inductive-biases-deep-learning-graph-networks/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/relational-inductive-biases-graph-networks-cover.webpGraph network block diagram showing input graph transformed through edge, node, and global update steps to produce an updated graphhttps://hunterheidenreich.com/notes/machine-learning/model-architectures/scaling-laws-vs-model-architectures/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/scaling-laws-vs-model-architectures-cover.webpLog-log plot comparing scaling laws across six architectures showing the vanilla Transformer has the steepest slopehttps://hunterheidenreich.com/notes/machine-learning/geometric-deep-learning/se3-transformers/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/se3-transformers-cover.webpSE(3)-Transformer architecture showing invariant attention weights modulating equivariant value messages on a 3D point cloudhttps://hunterheidenreich.com/notes/machine-learning/geometric-deep-learning/spherical-cnns/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/spherical-cnns-cover.webpComparison of planar CNN (translation only) versus spherical CNN (SO(3)-equivariant) showing how filters rotate on the spherehttps://hunterheidenreich.com/notes/machine-learning/model-architectures/quarks-of-attention/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/quarks-of-attention-cover.webpThe three quarks of attention: multiplexing (additive), output gating (multiplicative output), and synaptic gating (multiplicative weight)https://hunterheidenreich.com/notes/computational-chemistry/molecular-modeling/2026-03-02T00:00:00+00:00weekly0.8https://hunterheidenreich.com/archives/2026-03-01T00:00:00+00:00monthly0.5https://hunterheidenreich.com/tags/datasets/2026-03-15T00:00:00+00:00monthly0.5https://hunterheidenreich.com/tags/evaluation/2026-03-15T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/computational-chemistry/benchmark-problems/molecular-sets-moses/2026-03-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/moses-distribution.webpDensity plot showing training vs generated physicochemical property distributionhttps://hunterheidenreich.com/categories/document-processing/2026-03-07T00:00:00+00:00monthly0.5https://hunterheidenreich.com/posts/reliability-trap-document-automation/2026-02-21T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/page-stream-segmentation-throughput.webpChart showing the trade-off between accuracy and throughput in document automationThe Reliability Trap: High accuracy is only valuable when paired with confident volume automation. This chart shows the trade-off between 'safe' automation (throughput) and error rates.https://hunterheidenreich.com/tags/history-of-science/2026-03-14T00:00:00+00:00monthly0.5https://hunterheidenreich.com/posts/history-of-page-stream-segmentation/2026-02-21T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/page-stream-segmentation/page-stream-segmentation-sorter.webpConceptual diagram of page stream segmentation sorting pages into documentsVisualizing the PSS task of transforming a continuous stream of mixed pages into discrete document packets.https://hunterheidenreich.com/research/pubmed-ocr-pmc-open-access-ocr-annotations/2026-03-07T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/pubmed-ocr-stats.webpStatistics of the PubMed-OCR dataset including number of articles, pages, words, and bounding boxes.https://hunterheidenreich.com/notes/computational-chemistry/chemical-language-models/chemberta-3/2026-03-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/chemberta-3-cover.webpChemBERTa-3 visualization showing muscular arms lifting a stack of building blocks representing molecular data with SMILES notation, symbolizing the power and scalability of the open-source training frameworkhttps://hunterheidenreich.com/notes/computational-chemistry/llms-for-chemistry/chemdfm-r/2026-03-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/chemdfm-r-cover.webpChemical structures and molecular representations feeding into a neural network model that processes atomized chemical knowledgehttps://hunterheidenreich.com/tags/reinforcement-learning/2026-03-15T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/computational-chemistry/chemical-language-models/chemberta-2/2026-03-12T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/chemberta-2-cover.webpChemBERTa-2 visualization showing flowing SMILES strings in blue tones representing molecular data streamsChemBERTa-2 processes massive streams of SMILES molecular representationshttps://hunterheidenreich.com/notes/computational-chemistry/chemical-language-models/gp-molformer/2026-03-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/gp-molformer-architecture.webpGP-MoLFormer architecture showing large-scale SMILES input, linear-attention transformer decoder, and property optimization via pair-tuning soft promptshttps://hunterheidenreich.com/notes/computational-chemistry/chemical-language-models/chemberta/2026-03-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/chemberta-masked-language-modeling.webpChemBERTa masked language modeling visualization showing SMILES string CC(=O)O with masked tokensChemBERTa learns molecular representations by predicting masked tokens in SMILES strings (Masked Language Modeling)https://hunterheidenreich.com/notes/computational-chemistry/chemical-language-models/chemformer/2026-03-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/chemformer-pretraining.webpChemformer pre-training on 100M SMILES strings flowing into BART model, which then enables reaction prediction and property prediction taskshttps://hunterheidenreich.com/notes/machine-learning/generative-models/convexity-principle-interacting-gases/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/displacement-interpolation-comparison.webpComparison of linear interpolation (teleportation) showing double peaks versus displacement interpolation (transportation) showing smooth single peakLinear interpolation creates non-convex double peaks, while displacement interpolation preserves convexityhttps://hunterheidenreich.com/notes/machine-learning/generative-models/stochastic-interpolants/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/optimal-transport-flow.webpVisualization of probability density flow from initial distribution ρ₀ to target distribution ρ₁ over time through spaceStochastic interpolants define smooth probability flows between arbitrary distributionshttps://hunterheidenreich.com/notes/computational-chemistry/chemical-language-models/2026-03-18T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/machine-learning/generative-models/flow-matching-for-generative-modeling/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/flow-matching-vs-diffusion.webpVisualization comparing Optimal Transport (straight paths) vs Diffusion (curved paths) for Flow MatchingFlow Matching with Optimal Transport paths (green lines) provides straighter trajectories compared to Diffusion paths (gray dashed), enabling faster sampling with fewer function evaluations.https://hunterheidenreich.com/notes/machine-learning/generative-models/neural-odes/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/resnet-vs-ode.webpComparison of Residual Network vs ODE Network architectures showing discrete layers versus continuous transformationsResidual Networks use discrete layers (left) while ODE Networks parameterize continuous transformations (right), enabling adaptive depth and smooth trajectories.https://hunterheidenreich.com/notes/machine-learning/generative-models/rectified-flow/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/rectified-flow-visualization.webpVisualization showing linear interpolation, learned ODE trajectories, and the reflow straightening process for rectified flowhttps://hunterheidenreich.com/notes/machine-learning/generative-models/score-matching-denoising-autoencoders/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/denoising-score-matching-intuition.webpDenoising Score Matching Intuition - Vectors point from corrupted samples back to clean data, approximating the scoreDenoising Score Matching learns the score by training vectors to point from corrupted samples ($\tilde{x}$) back to their clean originals ($x$)https://hunterheidenreich.com/notes/machine-learning/generative-models/score-based-generative-modeling-sde/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/sde-trajectory-visualization.webpForward and Reverse SDE trajectories showing the diffusion process from data to noise and backForward SDE gradually adds noise to data (left), while the reverse SDE generates data from noise (right). White lines show the deterministic Probability Flow ODE, red lines show stochastic SDE paths.https://hunterheidenreich.com/notes/computational-chemistry/llms-for-chemistry/chemdfm-x/2026-03-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/chemdfm-x-architecture.webpChemDFM-X architecture showing five modalities (2D graphs, 3D conformations, images, MS2 spectra, IR spectra) feeding through separate encoders into unified LLM decoderhttps://hunterheidenreich.com/notes/interdisciplinary/computational-biology/dynamicflow/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/dynamicflow.webpDynamicFlow illustration showing the transformation from apo pocket to holo pocket with ligand molecule generationhttps://hunterheidenreich.com/tags/embeddings/2026-03-14T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/benchmarks-and-reviews/image-to-sequence-comparison/2026-03-03T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/image-to-sequence-comparison-cover.webpComparative analysis of image-to-sequence OCSR methodsEvolution and comparison of image-to-sequence architectures for chemical structure recognition.https://hunterheidenreich.com/notes/computational-chemistry/llms-for-chemistry/instructmol/2026-03-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/instructmol-architecture.webpInstructMol architecture showing molecular graph and text inputs feeding through two-stage training to produce property predictions, descriptions, and reactionshttps://hunterheidenreich.com/notes/interdisciplinary/computational-biology/invmsafold/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/invmsafold-cover.webpInvMSAFold generates diverse protein sequences from structure using a Potts modelhttps://hunterheidenreich.com/tags/materials/2026-03-14T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/computational-chemistry/llms-for-chemistry/mermaid/2026-03-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/mermaid-pipeline.webpMERMaid pipeline diagram showing PDF processing through VisualHeist segmentation, DataRaider VLM mining, and KGWizard graph construction to produce chemical knowledge graphshttps://hunterheidenreich.com/notes/computational-chemistry/molecular-modeling/mofflow/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/mofflow-cover.webpMOFFlow assembles metal nodes and organic linkers into Metal-Organic Framework structureshttps://hunterheidenreich.com/notes/computational-chemistry/llms-for-chemistry/shah-multimodal-search-2025/2026-03-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/multimodal-search-overview.webpDiagram showing text, molecular structures, and reactions feeding into a multimodal index and search system that outputs passages with contexthttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/hand-drawn/ocsaug/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/ocsaug-cover.webpOverview of the OCSAug pipeline showing DDPM training, masked RePaint augmentation, and OCSR fine-tuning phases.The OCSAug pipeline using diffusion models for hand-drawn chemical structure data augmentation.https://hunterheidenreich.com/notes/computational-chemistry/chemical-language-models/stout-v2/2026-03-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/stout-v2-overview.webpDiagram showing molecular structure passing through a neural network to produce IUPAC chemical nomenclature documenthttps://hunterheidenreich.com/notes/computational-chemistry/chemical-language-models/stout/2026-03-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/stout-molecular-interpreter.webpVintage wooden device labeled 'The Molecular Interpreter - Model 1974' with vacuum tubes, showing SMILES to IUPAC name translationhttps://hunterheidenreich.com/notes/computational-chemistry/chemical-language-models/struct2iupac-2021/2026-03-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/struct2iupac-workflow.webpDiagram showing Struct2IUPAC workflow: molecular structure (SMILES) passing through Transformer to generate IUPAC name, with round-trip verification loophttps://hunterheidenreich.com/notes/computational-chemistry/chemical-language-models/handsel-inchi-iupac-2021/2026-03-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/inchi-iupac-transformer.webpTransformer encoder-decoder architecture processing InChI string character-by-character to produce IUPAC chemical namehttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/hand-drawn/atomlenz/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/atomlenz-cover.webpAtomLenz learns atom-level detection from hand-drawn molecular images with weak supervisionhttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/benchmarks-and-reviews/krasnov-ocsr-benchmark-2024/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/krasnov-ocsr-benchmark-2024-cover.webpPrecision and recall comparison of 8 OCSR tools on patent imagesA comprehensive benchmark of 8 OCSR tools on manually curated patent images.https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/hand-drawn/chemreco/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/chemreco-cover.webpOverview of the ChemReco pipeline showing synthetic data generation and EfficientNet+Transformer architecture for hand-drawn chemical structure recognitionhttps://hunterheidenreich.com/notes/computational-chemistry/llms-for-chemistry/chemvlm/2026-03-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/chemvlm-architecture.webpChemVLM architecture showing molecular structure and text inputs flowing through vision encoder and language model into multimodal LLM for chemical reasoninghttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-sequence/decimer-ai/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/decimer-ai-cover.webpOverview of the DECIMER.ai platform combining segmentation, classification, and image-to-SMILES recognitionhttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-sequence/dgat/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/dgat-cover.webpArchitecture diagram of the DGAT model showing dual-path decoder with CGFE and SDGLA moduleshttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/hand-drawn/decimer-hand-drawn/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/decimer-hand-drawn-cover.webpDiagram showing the DECIMER hand-drawn OCSR pipeline from hand-drawn chemical structure image through EfficientNetV2 encoder and Transformer decoder to predicted SMILES outputhttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-sequence/image2inchi/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/image2inchi-cover.webpDiagram of the Image2InChI architecture showing a SwinTransformer encoder connected to an attention-based feature fusion decoder for converting molecular images to InChI strings.The Image2InChI architecture using SwinTransformer and attention-based feature fusion.https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/markush/markushgrapher/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/markushgrapher-cover.webpArchitecture diagram of the MarkushGrapher dual-encoder system combining VTL and OCSR encoders for Markush structure recognition.The MarkushGrapher architecture for joint visual and textual recognition of Markush structures.https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-sequence/mmssc-net/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/mmssc-net-cover.webpDiagram of the MMSSC-Net architecture showing the SwinV2 encoder and GPT-2 decoder pipeline for molecular image recognitionThe MMSSC-Net architecture for drug molecule recognition using SwinV2 and GPT-2.https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-graph/molgrapher/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/molgrapher-cover.webpMolGrapher: Graph-based Visual Recognition of Chemical StructuresThe MolGrapher pipeline for graph-based visual recognition of chemical structures.https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-graph/molmole/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/molmole-cover.webpOverview of the MolMole pipeline showing ViDetect, ViReact, and ViMore processing document pages to extract molecules and reactions.The MolMole unified vision pipeline for molecule detection, reaction parsing, and OCSR.https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-graph/molscribe/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/molscribe-cover.webpOverview of the MolScribe encoder-decoder architecture predicting atoms with coordinates and bonds from a molecular image.The MolScribe architecture for robust molecular structure recognition with image-to-graph generation.https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-sequence/molsight/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/molsight-cover.webpThree-stage training pipeline for MolSight showing pretraining, multi-granularity fine-tuning, and RL post-training stagesThe MolSight framework for OCSR using reinforcement learning and multi-granularity learning.https://hunterheidenreich.com/tags/research-methodology/2026-03-15T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-graph/abc-net/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/abc-net-cover.webpABC-Net detects atom and bond keypoints to reconstruct molecular graphs from imageshttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/hand-drawn/chempix/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/chempix-cover.webpOverview of the ChemPix CNN-LSTM pipeline converting a hand-drawn hydrocarbon sketch to a SMILES stringhttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-sequence/decimer-1.0/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/decimer-1.0-cover.webpArchitecture diagram showing the DECIMER 1.0 transformer pipeline from chemical image input to SELFIES outputhttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-sequence/vit-inchi-transformer/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/vit-inchi-transformer-cover.webpArchitecture diagram showing Vision Transformer encoder processing image patches and Transformer decoder generating InChI stringsThe Vision Transformer (ViT) encoder and Transformer decoder architecture for molecular image-to-InChI translation.https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/hand-drawn/hu-handwritten-rcgd-2023/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/hu-handwritten-rcgd-2023-cover.webpHandwritten chemical structure recognition with RCGD and SSMLThe RCGD framework for recognizing complex handwritten chemical structures.https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-sequence/icmdt/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/icmdt-cover.webpChemical structure diagram representing the ICMDT molecular translation systemThe ICMDT architecture for image-to-InChI translation.https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-graph/image-to-graph-transformers/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/image-to-graph-transformers-cover.webpDiagram showing a pixelated chemical image passing through a multi-layer encoder to produce a molecular graph with nodes and edges.The Image-to-Graph (I2G) architecture translating chemical images directly into molecular graphs.https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-sequence/image2smiles/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/image2smiles-cover.webp4-tert-butylphenol molecular structure diagram for Image2SMILES OCSRhttps://hunterheidenreich.com/tags/meta-science/2026-03-15T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-sequence/micer/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/micer-cover.webpBromobenzene molecular structure diagram for MICER OCSRhttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-graph/molminer/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/markush/jurriaans-markush-detection-2023/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/jurriaans-markush-detection-cover.webpPatch-based classification pipeline showing overlapping green and blue grids over a chemical image with Markush indicators highlighted in red.The patch-based pipeline splits chemical images into overlapping grids for Markush structure detection.https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/benchmarks-and-reviews/musazade-ocsr-review-2022/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/avatar.webphttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/benchmarks-and-reviews/rajan-string-representations-2022/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/avatar.webphttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-sequence/swinocsr/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/swinocsr-cover.webp4-chlorofluorobenzene molecular structure diagram for SwinOCSRhttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/benchmarks-and-reviews/rajan-ocsr-review-2020/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/avatar.webphttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-graph/chemgrapher-2020/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/chemgrapher-cover.webpChemGrapher pipeline overview showing segmentation and classification stagesThe ChemGrapher pipeline: from image input to digital graph output via segmentation and classification.https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-sequence/decimer/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/decimer-cover.webpEncoder-decoder architecture translating a chemical structure bitmap into a SMILES stringhttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-sequence/staker-deep-learning-2019/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/staker-deep-learning-2019-cover.webpThymol molecular structure diagram for Staker deep learning OCSRhttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/hand-drawn/hewahi-ring-recognition-2008/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/hewahi-ring-recognition-2008-cover.webpHandwritten chemical ring recognition neural network architectureThe proposed Classifier-Recognizer architecture for heterocyclic ring recognition.https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/online-recognition/tang-online-symbol-2013/2026-03-03T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/avatar.webphttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/online-recognition/zhang-hmm-handwriting-2009/2026-03-03T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/avatar.webphttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-sequence/img2mol/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/img2mol-cover.webpIbuprofen molecular structure diagram for Img2Mol OCSRhttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/online-recognition/yang-icpr-2008/2026-03-03T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/avatar.webphttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/online-recognition/wang-online-handwritten-2009/2026-03-03T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/avatar.webphttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/online-recognition/yang-online-handwritten-2008/2026-03-03T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/online-recognition/zhang-svm-hmm-2010/2026-03-03T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/avatar.webphttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/online-recognition/chang-unified-framework-2009/2026-03-04T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/chang-unified-framework-2009-cover.webpUnified framework converts handwritten chemical expressions to structured graph representationshttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/benchmarks-and-reviews/chemocr-trec-2011/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/chemocr-cover.webpDiagram of the chemoCR pipeline converting a bitmap chemical structure into a connection tablehttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/benchmarks-and-reviews/chemreader-trec-2011/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/chemreader-trec-cover.webpPipeline diagram of ChemReader chemical structure recognition from image to connection tablehttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/benchmarks-and-reviews/clef-ip-2012/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/clef-ip-2012-cover.webpOverview of CLEF-IP 2012 tasks including patent passage retrieval, flowchart recognition, and chemical structure extractionhttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/benchmarks-and-reviews/molrec-clef-2012/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/avatar.webphttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/benchmarks-and-reviews/osra-clef-2012/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/avatar.webphttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/benchmarks-and-reviews/trec-chem-2011/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/avatar.webphttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/rule-based/mlocsr/2026-03-03T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/avatar.webphttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/rule-based/hong-chemical-expression-2015/2026-03-03T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/hong-chemical-expression-2015-cover.webpOptical Chemical Structure Recognition workflow visualizationThe proposed OCSR workflow improving handling of adhesive symbols and wedge bonds.https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/rule-based/molrec-2012/2026-03-03T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/avatar.webphttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/online-recognition/chemink-2011/2026-03-03T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/chemink-cover.webpChemInk: Real-Time Recognition for Chemical Drawingshttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/rule-based/clide-pro-2009/2026-03-03T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/clide-pro-cover.webpCLiDE Pro: Optical Chemical Structure Recognition Toolhttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/benchmarks-and-reviews/imago-trec-2011/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/avatar.webphttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/rule-based/kekule-1996/2026-03-03T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/avatar.webphttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/benchmarks-and-reviews/osra-trec-2011/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/avatar.webphttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/hand-drawn/ramel-handwritten-1999/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/interdisciplinary/computational-social-science/nominate-1985/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/nominate-senate-vote-spatial-plot-1995.webpNOMINATE spatial plot showing Senate vote on Balanced Budget Amendment (1995) with legislators positioned on liberal-conservative dimensionSenate Vote on Balanced Budget Amendment (1995) visualized using NOMINATE. Image by [Chris hare84](https://commons.wikimedia.org/wiki/File:NOMINATE_3.jpg), CC BY 3.0.https://hunterheidenreich.com/categories/astrobiology/2026-03-14T00:00:00+00:00monthly0.5https://hunterheidenreich.com/tags/astrobiology/2026-03-14T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/rule-based/algorri-chemical-image-recognition-2007/2026-03-03T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/algorri-enc-2007-cover.webpAutomatic chemical image recognition pipeline from raster image to structured filehttps://hunterheidenreich.com/notes/interdisciplinary/planetary-science/chaotic-solar-system-1992/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/planetary-science/chaotic-solar-system-orbits.webpOrbital diagram showing chaotic planetary trajectorieshttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/rule-based/clide-1993/2026-03-03T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/clide-cover.webpChemical Literature Data Extraction: The CLiDE Projecthttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/rule-based/chemical-machine-vision/2026-03-03T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/chemical-machine-vision-cover.webpVisualization of Gabor wavelets and Kohonen networks for chemical image classificationThe Chemical Machine Vision pipeline uses Gabor wavelets for feature extraction and Kohonen networks for classification.https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/rule-based/chemreader-2009/2026-03-03T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/chemreader-cover.webpChemReader: Automated Structure Extractionhttps://hunterheidenreich.com/notes/interdisciplinary/computational-biology/2026-02-26T00:00:00+00:00weekly0.8https://hunterheidenreich.com/categories/computational-social-science/2026-03-14T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/interdisciplinary/computational-social-science/2026-02-26T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/machine-learning/classic-papers/distributed-representations/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/distributed-representations-binding.svgDiagram showing distributed representations with three pools of units (AGENT, RELATIONSHIP, PATIENT) connected via role/identity bindingsThe binding problem solution: true hierarchies require creating unique subpatterns that fuse an identity with its role, where the whole is represented as the sum of these combined representations.https://hunterheidenreich.com/notes/interdisciplinary/evolutionary-biology/drive-to-life-wet-icy-worlds/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/evolutionary-biology/alkaline-vent-origin-life.webpAbstract artistic representation of alkaline hydrothermal vents with spiraling geological formationsArtistic representation of alkaline hydrothermal vents as the birthplace of life.https://hunterheidenreich.com/notes/computational-chemistry/molecular-dynamics/self-diffusion-lj-fcc111-1989/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/lennard-jones-potential.webpGraph of the Lennard-Jones 12-6 potential showing the characteristic attractive and repulsive forceshttps://hunterheidenreich.com/tags/electrochemistry/2026-03-14T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/computational-chemistry/molecular-dynamics/embedded-atom-method-voter-1994/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/eam-embedding-effective-charge.webpEmbedding energy and effective charge functions for Ni and Pd from the original EAM paperhttps://hunterheidenreich.com/notes/computational-chemistry/molecular-dynamics/embedded-atom-method-review-1993/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/eam-embedding-effective-charge.webpEmbedding energy and effective charge functions for Ni and Pd from the original EAM paperhttps://hunterheidenreich.com/notes/computational-chemistry/molecular-dynamics/evans-thermal-conductivity-1986/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/lennard-jones-potential.webpGraph of the Lennard-Jones 12-6 potential showing the characteristic attractive and repulsive forceshttps://hunterheidenreich.com/categories/evolutionary-biology/2026-03-14T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/interdisciplinary/evolutionary-biology/2026-02-26T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/interdisciplinary/computational-biology/funnels-pathways-energy-landscape/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/energy-landscape-funnels.webpFour types of protein folding energy landscapes from left to right: smooth funnel, rugged funnel with kinetic traps, moat funnel, and champagne glass funnelFour proposed energy landscape topologies for protein folding. From left: idealized smooth funnel, rugged funnel with kinetic traps, moat funnel, and champagne glass funnel. Image by [Ken A. Dill](https://dillgroup.org/#/landscapes) (CC BY 4.0)https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/rule-based/contreras-ocr-1990/2026-03-03T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/contreras-ocr-1990-cover.webpGraph Perception for Chemical Structure OCRhttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/hand-drawn/ouyang-davis-aaai-2007/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-sequence/img2smi/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/ocsr/img2smiles.webpOptical chemical structure recognition examplehttps://hunterheidenreich.com/notes/computational-chemistry/molecular-dynamics/oxidation-reduction-oscillations-pt-sio2-1994/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/co-pt100-hollow.webpCarbon monoxide molecule adsorbed on Pt(100) FCC surface in hollow site configurationCO molecule adsorbed in hollow site on Pt(100) surfacehttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/rule-based/kekule-1992/2026-03-03T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/avatar.webphttps://hunterheidenreich.com/notes/computational-chemistry/molecular-dynamics/kinetic-oscillations-pt100-1985/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/co-pt100-hollow.webpCarbon monoxide molecule adsorbed on Pt(100) FCC surface in hollow site configurationCO molecule adsorbed in hollow site on Pt(100) surfacehttps://hunterheidenreich.com/tags/legislative-data/2026-03-14T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/computational-chemistry/molecular-dynamics/self-diffusion-metal-surfaces-1994/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/ir-001-surface-diffusion.webpIridium fcc(001) surface with adatomIridium (001) surface with adatom showing the loosely-packed structure where exchange mechanism occurshttps://hunterheidenreich.com/notes/machine-learning/generative-models/mixture-density-networks/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/inverse-problem-multimodal.webpVisualization of inverse problem showing one input mapping to multiple valid outputsThe inverse problem that motivates MDNs: a single input value maps to multiple valid target valueshttps://hunterheidenreich.com/tags/molecular-evolution/2026-03-14T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/benchmarks-and-reviews/ocsr-methods/2026-03-03T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/avatar.webphttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/rule-based/casey-ocsr-1993/2026-03-04T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/casey-ocsr-1993-cover.webpEarly optical recognition system converts scanned chemical diagrams to connection tableshttps://hunterheidenreich.com/tags/origin-of-life/2026-03-14T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/computational-chemistry/molecular-dynamics/oscillatory-co-oxidation-pt110-1992/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/oscillatory-co-pt110-replication.webpReplication of Figure 7 showing stable oscillations in CO oxidation on Pt(110)Simulated oscillations showing anti-phase relationship between CO coverage and surface reconstructionhttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/rule-based/osra/2026-03-03T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/ocsr/2008239616_449_chem.webpChemical structure diagram for optical recognitionhttps://hunterheidenreich.com/notes/interdisciplinary/computational-social-science/party-matters-hiptm/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/party-matters-embeddings.webpVisualization of party-based legislative embeddingshttps://hunterheidenreich.com/tags/phylogenetics/2026-03-14T00:00:00+00:00monthly0.5https://hunterheidenreich.com/categories/planetary-science/2026-03-14T00:00:00+00:00monthly0.5https://hunterheidenreich.com/tags/prebiotic-chemistry/2026-03-14T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/rule-based/algorri-reconstruction-2007/2026-03-03T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/algorri-reconstruction-2007-cover.webpFive-stage pipeline for reconstructing chemical molecules from raster imageshttps://hunterheidenreich.com/notes/computational-chemistry/molecular-dynamics/second-order-langevin-1987/2026-03-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/wiener-process-3d.webpThree-dimensional Brownian motion trajectory showing random walk behaviorA 3D Wiener process (Brownian motion) illustrating stochastic trajectory evolution. While this shows pure diffusion, the Langevin and Hyperbolic algorithms combine such random noise with deterministic forces from the potential energy landscape. Image: [Sullivan.t.j](https://commons.wikimedia.org/wiki/File:Wiener_process_3d.png), CC BY-SA 3.0.https://hunterheidenreich.com/notes/computational-chemistry/molecular-dynamics/stillinger-weber-1985/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/stillinger-weber-potential.webpVisualization of the Stillinger-Weber potential showing the two-body radial term and three-body angular penaltyhttps://hunterheidenreich.com/tags/surface-science/2026-03-14T00:00:00+00:00monthly0.5https://hunterheidenreich.com/tags/systematics/2026-03-14T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/interdisciplinary/computational-social-science/tea-party-hiptm/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/tea-party-hiptm.webpHierarchical Ideal Point Topic Model visualization showing political polarizationHierarchical topic structure and legislator ideal pointshttps://hunterheidenreich.com/tags/text-classification/2026-03-14T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/interdisciplinary/evolutionary-biology/woese-three-domain-1990/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/planetary-science/pyrolobus-fumarii.webpElectron microscope image of Pyrolobus fumarii showing irregular coccoid cell structureElectron microscope image of _Pyrolobus fumarii_, an archaeon that defines the upper temperature limit for known life at 113°C. (Manfred Rohde, CC BY-SA 4.0)https://hunterheidenreich.com/tags/tutorial/2026-03-14T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/computational-chemistry/molecular-dynamics/diffusion-adatom-dimers-1984/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/argon-dimer-diffusion.webpVisualization of argon dimer on fcc(111) surfaceInitial configuration showing an adatom dimer on an fcc(111) surfacehttps://hunterheidenreich.com/notes/interdisciplinary/research-methods/ai-physical-sciences-paper-taxonomy/2026-03-08T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/research-methods/paper-taxonomy-vectors.webpAbstract visualization of seven basis vectors represented as curved lines with dots, centered around a psi symbolhttps://hunterheidenreich.com/notes/computational-chemistry/molecular-dynamics/correlations-motion-atoms-liquid-argon/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/rahman-1964-argon-molecular-dynamics/rahman-argon-delayed-convolution.webpDelayed convolution approximation for distinct Van Hove function showing comparison between simulated data and theoretical modelhttps://hunterheidenreich.com/tags/paper-classification/2026-03-08T00:00:00+00:00monthly0.5https://hunterheidenreich.com/categories/research-methods/2026-03-08T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/interdisciplinary/research-methods/2026-02-26T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/interdisciplinary/planetary-science/cloud-continents/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/planetary-science/cloud-continent-cross-section.webpConceptual cross-section of the Cloud Continent proposal showing three layers: the CO2 atmosphere below, the nitrogen-filled honeycomb structure at 50 km altitude, and the habitable atmosphere aboveConceptual cross-section of the Cloud Continent terraforming proposal.https://hunterheidenreich.com/notes/interdisciplinary/planetary-science/venus-evolution-through-time/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/planetary-science/venus-mariner-10.webpVenus as seen by Mariner 10, showing swirling cloud patterns in the dense atmosphereVenus captured by Mariner 10 in 1974. (NASA/JPL-Caltech)https://hunterheidenreich.com/notes/interdisciplinary/planetary-science/life-on-venus/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/planetary-science/venus-mariner-10.webpVenus as seen by Mariner 10, showing swirling cloud patterns in the dense atmosphereVenus captured by Mariner 10 in 1974. (NASA/JPL-Caltech)https://hunterheidenreich.com/notes/computational-chemistry/benchmark-problems/2026-03-01T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/benchmarks-and-reviews/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/computational-chemistry/2026-03-18T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/computational-chemistry/classic-papers/2026-03-08T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/machine-learning/classic-papers/2026-03-08T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/machine-learning/generative-models/2026-02-26T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/machine-learning/geometric-deep-learning/2026-02-26T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/hand-drawn/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-graph/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-sequence/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/interdisciplinary/2026-03-08T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/machine-learning/2026-02-26T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/markush/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/computational-chemistry/datasets/2026-03-01T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/computational-chemistry/molecular-dynamics/2026-03-08T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/computational-chemistry/molecular-representations/2026-02-26T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/online-recognition/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/interdisciplinary/planetary-science/2026-02-26T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/rule-based/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/vision-language/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/projects/molecular-string-renderer/2025-12-12T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/molecular-renderer/cover-grid.webpGrid of complex molecular structures rendered from SELFIES and SMILES stringsProduction-grade rendering pipeline handling diverse chemical formatshttps://hunterheidenreich.com/notes/machine-learning/generative-models/autoencoding-variational-bayes/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/variational-autoencoder-reparameterization-trick.webpDiagram comparing standard stochastic sampling (gradient blocked) vs the reparameterization trick (gradient flows)The reparameterization trick enables end-to-end training of VAEshttps://hunterheidenreich.com/notes/machine-learning/generative-models/importance-weighted-autoencoders/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/variational-autoencoder-vae-vs-importance-weighted-autoencoder-iwae.webpFlowchart comparing VAE and IWAE computation showing the key difference in where averaging occurs relative to the log operationhttps://hunterheidenreich.com/posts/importance-weighted-autoencoders/2026-02-21T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/vae-tut/samples.webpMNIST digit samples generated from a Variational Autoencoder latent spaceVAE-generated samples from learned latent representationshttps://hunterheidenreich.com/notes/computational-chemistry/molecular-representations/inchi-and-tautomers/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/Glucose-tautomerism.webpD-glucose open-chain aldehyde form converting to beta-D-glucopyranose ring form, illustrating ring-chain tautomerismRing-chain tautomerism in glucosehttps://hunterheidenreich.com/notes/computational-chemistry/molecular-representations/inchi-2013/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/marcel-kraken-ligand10-conf0-2d.webp2D molecular structure diagram of tricyclohexylphosphine showing a central phosphorus atom bonded to three cyclohexyl groupshttps://hunterheidenreich.com/notes/computational-chemistry/molecular-representations/inchi-2025/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/inchi-2025-cover.webpCrystal structure of Na8Si46 clathrate displaying dodecahedral and tetrakaidecahedral coordination polyhedraCrystal structure of Na8Si46 clathrate. Image by [PickledSquid](https://commons.wikimedia.org/wiki/File:Na8Si46_inorganic_clathrate_structure_with_coordination_polyhedra.png), [CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/).https://hunterheidenreich.com/notes/computational-chemistry/molecular-representations/mixfile-minchi/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/mixfile-minchi-cover.webpA cobalt sulfate and ethylenediamine mixture being preparedPhoto by Leiem, [CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/), via Wikimedia Commonshttps://hunterheidenreich.com/notes/computational-chemistry/molecular-representations/ninchi-alpha/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/nanoart-fluorescence.webpColorized electron microscope image of nanostructured indium phosphide surface showing spatially oriented cubic crystallitesNanoart: Fluorescence by Yana Sychikova & Serhii Kovachov (CC BY 4.0)https://hunterheidenreich.com/notes/computational-chemistry/molecular-representations/selfies-2023/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/selfies/benzene_selfies.webpBenzene in SELFIES notationhttps://hunterheidenreich.com/notes/computational-chemistry/molecular-representations/rinchi/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/grignard-reaction-scheme.webpChemical diagram showing a generalized Grignard reactionImage: Grignard reaction scheme (Public Domain, Wikimedia Commons)https://hunterheidenreich.com/notes/computational-chemistry/molecular-representations/selfies-original-paper/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/selfies/selfies_cover.webpSELFIES molecular representation overviewhttps://hunterheidenreich.com/notes/computational-chemistry/molecular-representations/smiles-original-paper/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/smiles2img/benzene_demo.webpBenzene molecular structure diagramhttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/benchmarks-and-reviews/molrec_at_clef/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/ocsr/img2smiles.webpOptical chemical structure recognition examplehttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/benchmarks-and-reviews/molrec_at_trec/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/ocsr/img2smiles.webpOptical chemical structure recognition examplehttps://hunterheidenreich.com/posts/what-is-ocsr/2026-02-21T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/ocsr/img2smiles.webpThe transformation from a 2D chemical structure image to a SMILES representationOCSR bridges the gap between molecular images and machine-readable chemical datahttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-sequence/alpha-extractor/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/alpha-extractor-cover.webpαExtractor extracts structured chemical information from biomedical literaturehttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/rule-based/cheminfty/2026-03-03T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/cheminfty-cover.webpChemInfty: Chemical Structure Recognition in Patent Imageshttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/vision-language/molnextr/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/molnextr-cover.webpDiagram showing MolNexTR's dual-stream architecture: a molecular image feeds into parallel ConvNext and Vision Transformer encoders, producing a SMILES string.https://hunterheidenreich.com/img/avatar.webphttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/vision-language/molparser_7m-wildmol/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/molparser-colored-example.webpA colored molecule with annotations, representing the diverse drawing styles found in scientific papers that OCSR models must handle.The MolParser system is designed to perform end-to-end recognition of molecular structures found in real-world documents like patents and scientific literature.https://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/vision-language/mol-parser/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/ocsr/img2smiles.webpOptical chemical structure recognition examplehttps://hunterheidenreich.com/notes/computational-chemistry/datasets/zinc-22/2026-03-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/zinc-22-sample.webpZINC-22 Tranche Browser showing molecular count distributionZINC-22's Tranche Browser displaying the distribution of 37.2 billion molecules organized by heavy atom count and lipophilicity (LogP)https://hunterheidenreich.com/posts/visualizing-smiles-and-selfies-strings/2026-02-21T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/smiles2img/aspirin_demo.webpAspirin molecular structure generated from SMILES stringConverting 1D molecular strings to publication-quality 2D imageshttps://hunterheidenreich.com/notes/computational-chemistry/molecular-representations/selfies/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/selfies/selfies_cover.webpSELFIES representation of 2-Fluoroethenimine moleculeA SELFIES string `[F][=C][=C][#N]` and its corresponding molecular structurehttps://hunterheidenreich.com/notes/machine-learning/classic-papers/communication-in-the-presence-of-noise/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/geometric-interpretation-of-signals-as-spheres.webpSphere packing illustration showing Shannon's geometric interpretation of channel capacityhttps://hunterheidenreich.com/notes/interdisciplinary/computational-biology/fold-graciously/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/folding-funnel.webpProtein folding energy landscape funnel showing high-energy unfolded states converging to the native stateThe energy landscape funnel resolving Levinthal's Paradoxhttps://hunterheidenreich.com/notes/computational-chemistry/datasets/marcel/2026-03-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/marcel-kraken-ligand10-conf0-3d.webpMARCEL dataset Kraken ligand example in 3D conformationExample conformer from MARCEL's Kraken subset, showcasing the dataset's focus on 3D molecular conformations for machine learning applicationshttps://hunterheidenreich.com/notes/computational-chemistry/molecular-representations/smiles/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/smiles2img/benzene_demo.webpBenzene molecule with SMILES notationhttps://hunterheidenreich.com/notes/computational-chemistry/benchmark-problems/muller-brown-1979/2026-03-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/muller-brown/muller-brown-potential-surface.webpMüller-Brown Potential Energy Surface showing the three minima and two saddle pointsThe classic Müller-Brown potential energy surface with its characteristic three minima and curved reaction pathwayshttps://hunterheidenreich.com/notes/computational-chemistry/classic-papers/number-of-isomeric-hydrocarbons/2026-03-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/number-of-isomeric-hydrocarbons-of-the-methane-series.webpLog-scale plot showing exponential growth of alkane isomer counts from C1 to C40The combinatorial explosion of alkane structural isomershttps://hunterheidenreich.com/notes/interdisciplinary/planetary-science/surface-of-venus/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/planetary-science/venus-magellan-radar.webpMagellan radar mosaic of Venus showing the northern hemisphere with volcanic plains, tesserae, and lava flows in orange-brown tonesMagellan synthetic aperture radar mosaic of Venus. (NASA/JPL-Caltech)https://hunterheidenreich.com/2026-03-24T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/computational-chemistry/datasets/geom/2026-03-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/GEOM-sample-_4-pyrimidin-2-yloxyphenyl_acetamide.webpGEOM dataset example molecule: N-(4-pyrimidin-2-yloxyphenyl)acetamideExample SARS-CoV-2 3CL protease active molecule from the GEOM datasethttps://hunterheidenreich.com/posts/random-number-tricks/2026-02-21T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/exponential_random_gens.webpComparison of exponential sampling methods showing histograms from both inverse transform and von Neumann methods overlaid with the theoretical exponential distributionBoth sampling methods perfectly reproduce the exponential distributionhttps://hunterheidenreich.com/notes/computational-chemistry/datasets/gdb-11/2026-03-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/gdb_11_sample.webpGDB-11 molecule structure showing FC1C2OC1c3c(F)coc23Example GDB-11 molecule demonstrating the systematic generation of small organic structureshttps://hunterheidenreich.com/posts/muller-brown-in-pytorch/2026-02-21T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/muller-brown/muller-brown-potential-surface.webpMüller-Brown Potential Energy Surface showing the three minima and two saddle pointsThe classic Müller-Brown potential energy surface implemented in PyTorch with performance optimizationshttps://hunterheidenreich.com/img/muller-brown/muller-brown-saddle.webphttps://hunterheidenreich.com/img/muller-brown/muller-brown-throughput-analysis.webphttps://hunterheidenreich.com/videos/muller-brown-basin-ma-simulation/2025-08-31T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/muller-brown/muller-brown-potential-surface.webpMuller-Brown potential energy surfacehttps://hunterheidenreich.com/videos/muller-brown-basin-mb-simulation/2025-08-31T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/muller-brown/muller-brown-potential-surface.webpMuller-Brown potential energy surfacehttps://hunterheidenreich.com/projects/muller-brown-pytorch/2025-12-12T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/muller-brown/muller-brown-potential-surface.webpMüller-Brown Potential Energy Surface showing the three minima and two saddle pointsThe classic Müller-Brown potential energy surface implemented in PyTorch with performance optimizationshttps://hunterheidenreich.com/videos/muller-brown-transition-simulation/2025-08-31T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/muller-brown/muller-brown-potential-surface.webpMuller-Brown potential energy surfacehttps://hunterheidenreich.com/notes/computational-chemistry/molecular-modeling/denoise-vae/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/energy-surface.webpPotential energy surface showing molecular conformation space with equilibrium and low energy conformationsMolecular potential energy surface illustrating the landscape of conformationshttps://hunterheidenreich.com/notes/computational-chemistry/molecular-modeling/beyond-atoms/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/spaceformer-adaptive-grid-benzene.webpAdaptive grid merging visualization for benzene molecule showing multi-resolution spatial discretizationhttps://hunterheidenreich.com/notes/computational-chemistry/molecular-modeling/dark-side-of-forces/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/energy-surface.webpA mathematical representation of a potential energy surface (PES)https://hunterheidenreich.com/notes/computational-chemistry/molecular-modeling/efficient-dft-hamiltonian-predicton-sphnet/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/spherical-harmonics.webpSpherical harmonics visualizationhttps://hunterheidenreich.com/tags/lammps/2026-02-21T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/computational-chemistry/molecular-modeling/learning-smooth-interatomic-potentials/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/fullerene-cover.webpAtomic structure of a spherical fullerenehttps://hunterheidenreich.com/videos/liquid-argon-lammps-simulation/2025-08-31T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/rahman-1964-argon-molecular-dynamics/rahman-argon-radial-distribution-function.webpRadial distribution function of liquid argonhttps://hunterheidenreich.com/tags/method/2026-03-14T00:00:00+00:00monthly0.5https://hunterheidenreich.com/projects/rahman-1964-replication/2025-11-30T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/rahman-1964-argon-molecular-dynamics/rahman-argon-velocity-autocorrelation.webpVelocity Autocorrelation Function showing the signature negative region characteristic of liquid dynamics and the cage effect discovered by RahmanReplication of Rahman's 1964 discovery: the velocity autocorrelation function reveals the cage effect in liquid argonhttps://hunterheidenreich.com/posts/rahman-1964-lammps-liquid-argon/2025-12-12T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/rahman-1964-argon-molecular-dynamics/rahman-argon-velocity-autocorrelation.webpVelocity Autocorrelation Function showing the signature negative region characteristic of liquid dynamicsThe 'Cage Effect': A digital restoration of Rahman's discovery using Python and LAMMPShttps://hunterheidenreich.com/img/rahman-1964-argon-molecular-dynamics/rahman-argon-radial-distribution-function.webphttps://hunterheidenreich.com/img/rahman-1964-argon-molecular-dynamics/rahman-argon-mean-square-displacement.webphttps://hunterheidenreich.com/tags/ovito/2026-02-21T00:00:00+00:00monthly0.5https://hunterheidenreich.com/tags/systematization/2026-03-14T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/computational-chemistry/molecular-dynamics/embedded-atom-method/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/eam-embedding-effective-charge.webpEmbedding energy and effective charge functions for Ni and Pd from the original EAM paperhttps://hunterheidenreich.com/notes/computational-chemistry/molecular-dynamics/umbrella-sampling/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/folding-funnel.webpProtein folding funnel diagram illustrating energy landscapehttps://hunterheidenreich.com/notes/machine-learning/generative-models/contrastive-learning-for-vae-priors/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/vae-prior-hole-problem-illustrated.webpVisualization of the VAE prior hole problem showing a ring-shaped aggregate posterior with an empty center where the Gaussian prior has highest densityThe 'prior hole' problem - high prior density where no data existshttps://hunterheidenreich.com/notes/computational-chemistry/classic-papers/processes-of-adsorption/2026-03-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/method-of-images-atom-surface.webpSchematic showing atom-surface interaction using the method of imageshttps://hunterheidenreich.com/notes/computational-chemistry/datasets/gdb-13/2026-03-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/gdb_13_sample.webpGDB-13 molecule structure showing CCCC(O)(CO)CC1CC1CNExample GDB-13 molecule demonstrating the expanded chemical space with up to 13 atomshttps://hunterheidenreich.com/notes/computational-chemistry/datasets/gdb-17/2026-03-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/gdb_17_sample.webpGDB-17 molecule structure showing complex polycyclic architectureExample GDB-17 molecule demonstrating the complex 3D diversity and polycyclic structures characteristic of the 166 billion molecule databasehttps://hunterheidenreich.com/projects/modern-word2vec/2025-12-12T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/huffman-tree.webpHuffman Tree visualization for the input 'beep boop beer!' showing internal nodes with frequency counts and leaf nodes with charactersTensorizing the tree: flattening pointer-based traversals into dense GPU operationshttps://hunterheidenreich.com/categories/natural-language-processing/2026-02-21T00:00:00+00:00monthly0.5https://hunterheidenreich.com/posts/geom-conformer-generation-dataset/2026-02-21T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/GEOM-sample-_4-pyrimidin-2-yloxyphenyl_acetamide.webp3D conformer ensemble of a drug-like molecule from the GEOM datasetA sample molecule from the GEOM dataset: 3D conformer ensembles capture the flexible, dynamic nature of drug-like compoundshttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/vision-language/gtr-mol-vlm/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/gtr-cot-cover.webpDiagram showing graph traversal chain-of-thought parsing of a molecular structure image into atom and bond predictionshttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/vision-language/subgrapher/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/ocsr/markush.webpMarkush structure diagramhttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/vision-language/ocsu/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/ocsu-cover.webpOCSU: Optical Chemical Structure UnderstandingThe OCSU framework for multi-level understanding of molecular images.https://hunterheidenreich.com/notes/machine-learning/geometric-deep-learning/3d-steerable-cnns/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/3d-cnn-versus-3d-steerable-cnn.webpComparison of standard 3D CNN versus 3D Steerable CNN for handling rotational symmetryhttps://hunterheidenreich.com/research/page-stream-segmentation-llms/2026-02-21T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/page-stream-segmentation-throughput.webpStream accuracy versus relative throughput for Mistral-7B and XGBoost modelshttps://hunterheidenreich.com/notes/computational-chemistry/optical-structure-recognition/image-to-sequence/rfl/2026-03-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/computational-chemistry/rfl-cover.webpDiagram showing how Ring-Free Language decouples a molecular graph into skeleton, ring structures, and branch informationThe Ring-Free Language (RFL) representation and Molecular Skeleton Decoder (MSD) architecture.https://hunterheidenreich.com/categories/time-series-forecasting/2026-02-21T00:00:00+00:00monthly0.5https://hunterheidenreich.com/research/llm-page-stream-segmentation/2025-11-29T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/page-stream-segmentation-diagram.webpDiagram showing page stream segmentation workflow: an input stream of pages is processed through binary classification of page pairs to predict document breaks, producing segmented output documentsPage Stream Segmentation formulated as binary classification of page pairs.https://hunterheidenreich.com/tags/genomics/2026-03-14T00:00:00+00:00monthly0.5https://hunterheidenreich.com/notes/interdisciplinary/evolutionary-biology/nature-of-luca-early-earth-system/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/notes/luca-tree-of-life.webpA reconstruction of LUCA within its evolutionary and ecological contextLUCA in the context of the tree of life. Branches that have left sampled descendants today are colored black, those that have left no sampled descendants are in grey. (Moody et al., CC BY 4.0)https://hunterheidenreich.com/notes/computational-chemistry/molecular-representations/invalid-smiles-help/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/selfies/selfies_robustness_demo.webpSELFIES robustness demonstrationhttps://hunterheidenreich.com/projects/isomer-dataset-generation/2025-11-30T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/alkane-constitutional-isomers/4-Butane-3D-balls.webp3D ball-and-stick model of butane molecule representing the structural isomer generation processFrom Formula to Physics: Enumerating structural isomers and sampling 3D conformershttps://hunterheidenreich.com/posts/modern-variational-autoencoder-in-pytorch/2026-02-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/vae-tut/vae.webpVariational Autoencoder architecture diagram showing encoder, latent space, and decoderVAE architecture: Encoder compresses data to a latent distribution, decoder reconstructs from sampleshttps://hunterheidenreich.com/img/vae-tut/phase-diagram.webphttps://hunterheidenreich.com/img/vae-tut/encoding-diagram.webphttps://hunterheidenreich.com/img/vae-tut/reconstruction-loss-graphic.webphttps://hunterheidenreich.com/img/vae-tut/kl-loss-graphic.webphttps://hunterheidenreich.com/img/vae-tut/gradient_behaviors.webphttps://hunterheidenreich.com/img/vae-tut/z2-elbo_epochs.webphttps://hunterheidenreich.com/img/vae-tut/z2-loss_scatter_epochs.webphttps://hunterheidenreich.com/posts/sarcasm-detection-with-transformers/2026-02-21T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/Word_vector_illustration-50.webpWord vector illustration showing text classification and NLP conceptsHigh accuracy requires proper evaluation: a lesson in dataset biashttps://hunterheidenreich.com/tags/social-media/2026-02-21T00:00:00+00:00monthly0.5https://hunterheidenreich.com/posts/alkane-constitutional-isomer-classification/2026-02-21T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/alkane-constitutional-isomers/4-Butane-3D-balls.webp3D ball-and-stick model of butane molecule showing linear carbon chain structureButane vs. isobutane: identical formulas, different shapes. Can eigenvalues distinguish them?https://hunterheidenreich.com/img/alkane-constitutional-isomers/2d_largest_eigenvalue_C4H10.webphttps://hunterheidenreich.com/img/alkane-constitutional-isomers/2d_log_largest_eigenvalue_C4H10.webphttps://hunterheidenreich.com/img/alkane-constitutional-isomers/4-Isobutane-3D-balls.webphttps://hunterheidenreich.com/img/alkane-constitutional-isomers/99_variance_explained.webphttps://hunterheidenreich.com/img/alkane-constitutional-isomers/99_variance_explained_log.webphttps://hunterheidenreich.com/img/alkane-constitutional-isomers/alkane_coulomb_matrix_largest_eigenvalues.webphttps://hunterheidenreich.com/img/alkane-constitutional-isomers/alkane_log_coulomb_matrix_largest_eigenvalues.webphttps://hunterheidenreich.com/img/alkane-constitutional-isomers/pdf_largest_eigenvalue_C4H10.webphttps://hunterheidenreich.com/img/alkane-constitutional-isomers/pdf_largest_eigenvalue_C5H12.webphttps://hunterheidenreich.com/img/alkane-constitutional-isomers/pdf_largest_eigenvalue_C6H14.webphttps://hunterheidenreich.com/img/alkane-constitutional-isomers/pdf_largest_eigenvalue_C7H16.webphttps://hunterheidenreich.com/img/alkane-constitutional-isomers/dunn_index_vs_num_carbon_atoms.webphttps://hunterheidenreich.com/img/alkane-constitutional-isomers/fraction_of_negative_silhouette_scores_vs_num_carbon_atoms.webphttps://hunterheidenreich.com/img/alkane-constitutional-isomers/fraction_of_negative_silhouette_scores_vs_num_carbon_atoms_individual.webphttps://hunterheidenreich.com/img/alkane-constitutional-isomers/alkane-classification-1nn.webphttps://hunterheidenreich.com/img/alkane-constitutional-isomers/alkane-classification-3nn.webphttps://hunterheidenreich.com/img/alkane-constitutional-isomers/alkane-classification-5nn.webphttps://hunterheidenreich.com/img/alkane-constitutional-isomers/alkane-classification-knn.webphttps://hunterheidenreich.com/img/alkane-constitutional-isomers/alkane-classification-lr.webphttps://hunterheidenreich.com/img/alkane-constitutional-isomers/alkane-classification-1nn-lr.webphttps://hunterheidenreich.com/posts/congressional-bill-policy-area-classification/2026-02-21T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/logreg_summary_policy_area/top-Armed_Forces_and_National_Security.webpTop features for Armed Forces and National Security policy classification showing veterans, defense, military keywordsLogistic regression achieves 86%+ F1 on congressional bill policy classificationhttps://hunterheidenreich.com/leaderboards/policy_area_classification_leaderboard/2026-02-21T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/avatar.webphttps://hunterheidenreich.com/posts/molecular-descriptor-coulomb-matrix/2025-12-12T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/cm_bicyclobutane_log.webpCoulomb matrix heatmap visualization showing molecular structure encoding on logarithmic scaleCoulomb matrices transform 3D molecular structures into ML-ready features through pairwise atomic interactionshttps://hunterheidenreich.com/img/cm_bicyclobutane.webphttps://hunterheidenreich.com/img/cm_bicyclobutane_eigenvalues.webphttps://hunterheidenreich.com/img/cm_bicyclobutane_log_eigenvalues.webphttps://hunterheidenreich.com/about/2026-03-24T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/avatar.webphttps://hunterheidenreich.com/posts/us-117th-congress-data-exploration/2026-02-21T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/nb_title_policy_area/top-Economics_and_Public_Finance.webpTop features for Economics and Public Finance policy classification across CongressesAnalyzing 15,000+ bills from the 117th Congress to understand how legislation really workshttps://hunterheidenreich.com/posts/kabsch-algorithm/2026-02-15T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/scientific-computing/rmsd-weakness-mutation.webpMolecular structure alignment showing protein conformations and RMSD calculationThe Kabsch algorithm optimally aligns molecular structures for RMSD calculations across ML frameworkshttps://hunterheidenreich.com/videos/cu-adatom-diffusion/2025-08-31T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/adatom_cu_xy.webpCopper adatom trajectory on Cu(100) surfacehttps://hunterheidenreich.com/posts/adatom-cu-diffusion/2026-02-21T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/posts/crystal-surface.webpBall model representation of a crystal surface with steps, kinks, adatoms, and vacancies showing various surface featuresCrystal surface structure showing adatoms and surface defects. This is the foundation for understanding surface diffusion.https://hunterheidenreich.com/img/adatom_cu_energy_avg.webphttps://hunterheidenreich.com/img/adatom_cu_xy.webphttps://hunterheidenreich.com/img/adatom_cu_z.webphttps://hunterheidenreich.com/videos/pt-adatom-diffusion/2025-08-31T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/adatom_cu_xy.webpAdatom surface diffusion trajectory on FCC metal surfacehttps://hunterheidenreich.com/projects/lammps-adatom-diffusion/2025-11-30T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/adatom_cu_energy_avg.webpEnergy conservation plot showing kinetic, potential, and total energy oscillations for a copper adatom diffusion simulationAutomated health-check dashboard generated by the Python analysis pipelinehttps://hunterheidenreich.com/posts/mini-proteins/2026-02-21T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/mini-protein-trajectories-cover.webpMolecular visualization of a methionine dipeptide structure from MD simulationAmino acid dipeptide trajectory snapshot from GROMACS simulationhttps://hunterheidenreich.com/img/alanine-dipeptide-molecular-dynamics.webphttps://hunterheidenreich.com/img/glycine-dipeptide-molecular-dynamics.webphttps://hunterheidenreich.com/img/isoleucine-dipeptide-molecular-dynamics.webphttps://hunterheidenreich.com/img/leucine-dipeptide-molecular-dynamics.webphttps://hunterheidenreich.com/img/methionine-dipeptide-molecular-dynamics.webphttps://hunterheidenreich.com/img/phenylalanine-dipeptide-molecular-dynamics.webphttps://hunterheidenreich.com/img/proline-dipeptide-molecular-dynamics.webphttps://hunterheidenreich.com/img/tryptophan-dipeptide-molecular-dynamics.webphttps://hunterheidenreich.com/img/valine-dipeptide-molecular-dynamics.webphttps://hunterheidenreich.com/projects/mini-protein-trajectories/2025-12-14T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/mini-protein-trajectories-cover.webpMolecular visualization of a methionine dipeptide structure from MD simulationAmino acid dipeptide trajectory snapshot from GROMACS simulationhttps://hunterheidenreich.com/projects/congressional-data-analysis/2025-11-30T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/logreg_summary_policy_area/top-Social_Welfare.webpTop features for Social Welfare policy classification showing social, poverty, benefits keywords47,000+ bills analyzed: legislative graph construction and 87% policy classification accuracyhttps://hunterheidenreich.com/notes/computational-chemistry/molecular-representations/selfies-2022/2026-03-14T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/selfies/selfies_robustness_demo.webpSELFIES robustness demonstrationhttps://hunterheidenreich.com/projects/iqcrnn-pytorch/2025-11-30T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/IQCRNN.webpComparison of IQCRNN (Our Method) vs standard Policy Gradient showing training curves, phase portraits, and state trajectories for control tasksIQCRNN achieves faster, more stable convergence with bounded state trajectories compared to unconstrained policy gradient methodshttps://hunterheidenreich.com/research/word-company-vicinity/2026-02-21T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/word-bias-iqr.webpInformation Quality Ratio plot showing statistical dependencies decay as window size increaseshttps://hunterheidenreich.com/research/eigennoise-contrastive-prior/2026-02-21T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/eigennoise_cover.webpHeatmap visualization of the EigenNoise analytical co-occurrence prior matrix showing word rank relationshipshttps://hunterheidenreich.com/research/look-dont-tweet/2026-02-21T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/look-dont-tweet-universal-message.webpDiagram of the Universal Message schema showing fields like ID, Text, Author, and Reply Sets that normalize data across platformshttps://hunterheidenreich.com/projects/pyconversations-social-media-analysis/2025-08-31T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/look-dont-tweet-universal-message.webpDiagram of the Universal Message schema showing fields like ID, Text, Author, and Reply Sets that normalize data across platformsThe 'Universal Message' schema designed to normalize disparate social media structures into a single DAG-compatible format.https://hunterheidenreich.com/tags/adversarial-machine-learning/2026-02-21T00:00:00+00:00monthly0.5https://hunterheidenreich.com/categories/ai-safety/2026-02-21T00:00:00+00:00monthly0.5https://hunterheidenreich.com/research/gpt2-adversarial-triggers/2026-02-21T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/gpt-2-universal-adversarial-triggers-flat-earth.webpA nonsensical trigger sequence 'WTC theoriesclimate Flat Hubbard Principle' is fed into GPT-2, which then generates Flat Earth conspiracy texthttps://hunterheidenreich.com/posts/a-roadmap-to-multi-arm-bandit-algorithms/2026-02-21T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/multi-arm-bandits/multi-arm-bandit-conceptual-graphic.webpVintage slot machine with multiple arms representing the multi-arm bandit problem in machine learningMulti-arm bandit algorithms balance exploration and exploitation, like choosing between slot machines at a casinohttps://hunterheidenreich.com/img/multi-arm-bandits/multi-arm-bandit-for-ecommerce.webphttps://hunterheidenreich.com/img/multi-arm-bandits/multi-arm-bandit-for-ad-bidding.webphttps://hunterheidenreich.com/research/newstweet-social-media-journalism/2026-02-21T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/newstweet.webpNewsTweet data collection pipeline: news outlets are crawled via Google News RSS feeds, articles are accessed to extract embedded tweets, and user timelines are downloaded from Twitterhttps://hunterheidenreich.com/research/coordinated-social-targeting/2026-02-21T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/social-targeting/elonmusk-realDonaldTrump-twitter-spikes-and-saws.webpSawtooth follower growth patterns for @elonmusk and @realDonaldTrump showing coordinated bot activityhttps://hunterheidenreich.com/research/semantic-network-induction/2026-02-21T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/wiktionary-wordnet-induction.webpVenn diagram showing semantic overlap between word senses for go, move, and proceed, illustrating our hierarchy induction algorithmhttps://hunterheidenreich.com/posts/neuroevolution-neat-and-hyperneat/2026-02-21T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/neat_genomes.webpNEAT genome encoding diagram showing node genes and connection genes with innovation numbersNEAT and HyperNEAT: evolving neural network topologies through direct and indirect gene representationhttps://hunterheidenreich.com/img/competing_conventions.webphttps://hunterheidenreich.com/img/neat_crossover.webphttps://hunterheidenreich.com/img/hyperneat_cppns.webphttps://hunterheidenreich.com/img/hyperneat_cppn_basics.webphttps://hunterheidenreich.com/img/hyperneat_substrate_configurations.webphttps://hunterheidenreich.com/img/hyperneat_inputs_outputs.webphttps://hunterheidenreich.com/posts/breaking-down-ml-for-the-average-person/2025-12-13T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/types-of-machine-learning.webpDiagram showing the three main types of machine learning: supervised, unsupervised, and reinforcement learningThe three fundamental approaches to machine learning that power modern AI applicationshttps://hunterheidenreich.com/posts/knowledge-based-agents-and-logic/2026-02-21T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/knowledge-base-diagram.webpDiagram illustrating knowledge-based agent architecture with knowledge base, reasoning, and action componentsKnowledge-based agents use structured knowledge bases for reasoning and decision-makinghttps://hunterheidenreich.com/projects/cgp-julia/2025-11-30T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/cartesian-genetic-programming.webpCartesian Genetic Programming graph showing input nodes, function nodes, and output nodes with active and inactive connectionsA CGP graph with active paths (bold) and inactive 'junk DNA' connections (faded)https://hunterheidenreich.com/posts/quac-question-answering-in-context/2026-02-21T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/quac_coref.webpTypes and distribution of coreferences in QuAC dataset showing dialogue complexityQuAC's challenge: 86% of questions require resolving references across conversation contexthttps://hunterheidenreich.com/img/quac_stats.webphttps://hunterheidenreich.com/img/quac_convo.webphttps://hunterheidenreich.com/img/quac_comparison.webphttps://hunterheidenreich.com/img/quac_dist.webphttps://hunterheidenreich.com/img/quac_performance.webphttps://hunterheidenreich.com/posts/coqa-conversation-question-answering/2026-02-21T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/coqa_coreferences.webpVisualization of coreference resolution in the CoQA conversational question answering datasetCoreference resolution is a key challenge in conversational QAhttps://hunterheidenreich.com/img/coqa_domains.webphttps://hunterheidenreich.com/img/squad_coqa_size.webphttps://hunterheidenreich.com/img/squad_v_coqa.webphttps://hunterheidenreich.com/img/squad_coqa_answers.webphttps://hunterheidenreich.com/img/coqa_scores.webphttps://hunterheidenreich.com/posts/what-is-a-gan/2026-02-21T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/GAN-SUMMARY-50.webpIllustration of GAN training process showing adversarial competition between generator and discriminatorGANs learn through competition: the generator creates fake data while the discriminator tries to detect ithttps://hunterheidenreich.com/img/GAN-70.webphttps://hunterheidenreich.com/img/gen_ai_types.webphttps://hunterheidenreich.com/img/wasserstein.webphttps://hunterheidenreich.com/img/lsgan-result.webphttps://hunterheidenreich.com/posts/intro-to-word-embeddings/2026-02-21T00:00:00+00:00weekly0.8https://hunterheidenreich.com/img/distributional_semantics-50.webp3D visualization of word embeddings showing semantic relationships in vector spaceWord embeddings position semantically similar words close together in high-dimensional spacehttps://hunterheidenreich.com/img/word_vector_onehot-50.webphttps://hunterheidenreich.com/img/Word_co-occurrence_network_%28range_3_words%29_-_ENG-50.webphttps://hunterheidenreich.com/img/bengio-npm-50.webphttps://hunterheidenreich.com/img/Word_vector_illustration.webphttps://hunterheidenreich.com/projects/fftw-compiler-haskell/2025-11-30T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/fftw-radix2-butterfly.webpRadix-2 DIT butterfly diagram showing the fundamental FFT operation with twiddle factor multiplicationThe butterfly operation: the atomic unit of FFT computationhttps://hunterheidenreich.com/projects/term-schedule-optimizer/2025-11-30T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/term-scheduler.webpFlowchart diagram showing the recursive backtracking algorithm for constraint satisfaction in schedule generationThe recursive_generator function implements a CSP solver using depth-first search with backtrackinghttps://hunterheidenreich.com/categories/creative-work/2025-12-13T00:00:00+00:00monthly0.5https://hunterheidenreich.com/tags/creative-coding/2025-12-13T00:00:00+00:00monthly0.5https://hunterheidenreich.com/videos/rubiks-cube-player-hackathon/2025-12-12T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/rubiks-cube-player.webpRubik's cube solver interfacehttps://hunterheidenreich.com/projects/rubiks-cube-player/2025-11-30T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/rubiks-cube-player.webpRubik's Cube sonification hackathon project showing a Raspberry Pi setup with webcam reading cube colors and Python code on screenMapping visual entropy to harmonic audio synthesishttps://hunterheidenreich.com/videos/internal-reflection-music-video/2025-12-13T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/posts/internal-reflection-cover.webpColorful overlapping circles from audio-reactive visualizationhttps://hunterheidenreich.com/videos/carnal-desires-music-video/2025-12-13T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/youthful-notion-the-edge-of-waking-ep.webpThe Edge of Waking EP album cover by Youthful Notionhttps://hunterheidenreich.com/videos/concavity-music-video/2025-12-13T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/youthful-notion-the-edge-of-waking-ep.webpThe Edge of Waking EP album cover by Youthful Notionhttps://hunterheidenreich.com/videos/false-gods-official-video/2025-12-13T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/youthful-notion-the-edge-of-waking-ep.webpThe Edge of Waking EP album cover by Youthful Notionhttps://hunterheidenreich.com/videos/false-idols-music-video/2025-12-13T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/youthful-notion-the-edge-of-waking-ep.webpThe Edge of Waking EP album cover by Youthful Notionhttps://hunterheidenreich.com/videos/in-the-grass-by-a-lake-music-video/2025-12-13T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/youthful-notion-the-edge-of-waking-ep.webpThe Edge of Waking EP album cover by Youthful Notionhttps://hunterheidenreich.com/videos/the-secrecy-in-dreaming-music-video/2025-12-13T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/youthful-notion-the-edge-of-waking-ep.webpThe Edge of Waking EP album cover by Youthful Notionhttps://hunterheidenreich.com/projects/elemental-brawl/2025-12-13T00:00:00+00:00monthly0.5https://hunterheidenreich.com/img/elemental-brawl/elemental-brawl-logo.webpElemental Brawl logo featuring stylized text with elemental imageryA high school dream project from 2014https://hunterheidenreich.com/muller-brown-optimized/2026-03-24T13:16:21+00:00