Answers are grounded in this paper and cite their sources.
Computational validation demonstrates accurate polymer blend identification from elemental composition alone, indicating reliable material deconvolution via browser-native software.
The Elemental Blend Solver (EBS v9.1.1) is a browser-native application that identifies, from a target compound’s elemental composition alone, the mass-weighted database blend that best reproduces it, with no candidate materials nominated and the number of blend components selected automatically. When candidates are known, the search can instead be restricted to up to five user-specified sources. EBS formulates the problem as constrained Non-Negative Least Squares (NNLS) and applies the Lawson–Hanson active-set algorithm to produce non-negative mixing fractions summing to unity. The software embeds a 73-material database spanning 20 elements in a single HTML file ( ∼ 276 KB), requiring no installation, server, or network access. Sixteen tools span chart views and panel diagnostics: Monte Carlo sensitivity analysis, principal component analysis (PCA) coverage mapping, ternary SSR (sum of squared residuals) contouring, exhaustive database search, and solution-space exploration. A dimensionless fit index in ( 0,1 ] , the Global Decomposability Score (GDS), gives an instrument-independent quality signal. A 20-molecule test suite confirms exact recovery ( SSR < 1 0 − 8 ) for hull-interior targets and minimum-error approximation otherwise. Validation against published CHNS combustion data confirms noise-floor recovery: root-mean-square error ≤ 0.050 percentage points (pp), within single- σ CHN instrument noise, across seven fully-represented blends ( p ≥ 0.860 ); absent-source cases are correctly flagged by the χ 2 diagnostic ( p < 0.001 ), confirming graceful degradation. Halogen mass balance constrains PVC identification to sub-0.1 pp precision. Together these capabilities form a complete, validated analytical workbench. EBS is freely available under the MIT license (DOI: 10.5281/zenodo.20818128 ).
No takes yet. Share an insight, caveat, or question.
Paul Mason (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: