TY  - EJOU
AU  - Hu, Haipeng 
AU  - Hong, Tao 
AU  - Zhu, Junjie 
AU  - Yao, Xinjie 
AU  - Guo, Zhoupeng 
AU  - Xia, Dahai 

TI  - CALPHAD-Informed MAP Priors for Cold-Start Composition-Space Partitioning in Active Alloy Design
T2  - Computers, Materials \& Continua

PY  - 2026
VL  - 89
IS  - 2
SN  - 1546-2226

AB  - Cold-start alloy-design campaigns often have too few labeled compositions to reliably locate phase boundaries for tree-structured composition-space Gaussian process regression (TCGPR). We study a controlled way to incorporate external CALPHAD-like boundary information into this partitioning step. The proposed MP-TCGPR method adds a Gaussian MAP penalty centered on a thermodynamic boundary estimate and uses an adaptive width <math id="mml-ieqn-1"><msub><mi>σ</mi><mi>j</mi></msub><mo stretchy="false">(</mo><mi>N</mi><mo stretchy="false">)</mo><mo>=</mo><msub><mi>σ</mi><mn>0</mn></msub><msqrt><mn>1</mn><mo>+</mo><mi>N</mi><mrow><mo>/</mo></mrow><msub><mi>N</mi><mrow><mrow><mi mathvariant="normal">c</mi><mi mathvariant="normal">r</mi><mi mathvariant="normal">o</mi><mi mathvariant="normal">s</mi><mi mathvariant="normal">s</mi></mrow></mrow></msub></msqrt></math> to reduce prior influence as node-level data accumulate. The revised theory distinguishes asymptotic convergence from convergence rate: a fixed-width prior is also asymptotically negligible under local regularity, whereas the adaptive schedule accelerates finite-sample prior release. Across 900 one-dimensional simulation runs spanning six Al-alloy cross-sections, MP-TCGPR reduces normalized Hausdorff distance at <math id="mml-ieqn-2"><msub><mi>N</mi><mn>0</mn></msub><mo>=</mo><mn>10</mn></math> from <math id="mml-ieqn-3"><mn>0.2233</mn><mo>±</mo><mn>0.2939</mn></math> to <math id="mml-ieqn-4"><mn>0.0613</mn><mo>±</mo><mn>0.0487</mn></math>. Additional fixed-width MAP, CALPHAD-only, misspecification, <math id="mml-ieqn-5"><msub><mi>N</mi><mrow><mrow><mi mathvariant="normal">c</mi><mi mathvariant="normal">r</mi><mi mathvariant="normal">o</mi><mi mathvariant="normal">s</mi><mi mathvariant="normal">s</mi></mrow></mrow></msub></math>, two-dimensional, and active-learning diagnostics show that most cold-start gain comes from a sufficiently accurate thermodynamic prior, while downstream AL benefits are modest in the tested budgets. The method is therefore presented as a reproducible cold-start partitioning strategy whose reliability depends on prior calibration, not as validation of a specific production CALPHAD database.
KW  - Active learning; CALPHAD; Gaussian process regression; aluminum alloy design; composition-space partitioning; MAP prior

DO  - 10.32604/cmc.2026.086475