Automated multi-objective reaction optimisation: which algorithm should I use

Abandoned without a dataset or submission. The wishlist item was Available when checked, and the pinned card, curation Skill, contract and review criteria were read. The White Rose accepted manuscript was legitimately downloaded. The publisher SI's indexed text was readable, and 104 numeric entries were preserved locally (Table S4:51; Table S5:53), but the original SI PDF URL and screenshots returned HTTP404; the publisher supplementary host returned403. No access controls were bypassed.

Full coverage could not be demonstrated: the first16 table entries are repeated between the algorithms; the manuscript states15 LHC initial experiments,35 iterations per algorithm, and six repeated-condition experiments, two added to the initial set. Figure S10 holds the reproducibility results, and no exact readable figure was available. Do not represent104 table rows as104 unique physical experiments or manufacture data to match the card estimate. The source also conflicts on steady-state timing (1.5 vs1.7 reactor volumes). The Conc1/M column is not safely an initial feed concentration; many values resemble residual sulfide and some are inconsistent. Products are sulfoxide/sulfone; outcomes are GC-MS conversion/selectivity and author-derived STY, not isolated yields. The standard is actually premixed in the feed; exact water amount is not recoverable from reported aqueous-H2O2 stock descriptions.

Future work needs a legitimate complete SI copy and reconciliation of shared seeds, repeated experiments and numerical source defects. Local source extraction and a detailed checkpoint were preserved for continuation. A separate accessible paper is being selected under the user's request to complete five curation problems. AI assistance used web/source retrieval, Python extraction and independent review planning; no ORD validator result is claimed because no dataset was built.

The White Rose accepted manuscript downloaded successfully. The publisher ESI is available as indexed full text through web retrieval at https://www.rsc.org/suppdata/d1/re/d1re00549a/d1re00549a1.pdf, but direct PDF retrieval and screenshots returned 404; an ordinary request to the publisher pubs.rsc.org supplementary path returned 403. No access controls were bypassed. Parsed indexed Tables S4 and S5 contain 51 and 53 rows respectively, with identical first 16 rows, rather than the card's 50+50 counts. Main manuscript states a shared 15-experiment initial design, six reproducibility experiments, and two reproducibility results included in the initial set. This requires reconciliation against the original SI figures; Figure S10's six values and conditions are not available in extracted text. Also the main manuscript and SI characterization state 1.5 volumes before sampling, while SI optimisation procedure states 1.7; Tables S4/S5 concentration column is not to be assumed an initial feed concentration. Retained all 104 reported numeric table rows and these caveats locally. Full faithful coverage cannot yet be asserted, so no dataset will be submitted from incomplete evidence.

Wishlist item 126570057 is Available in the public GitHub project embedded data (status f75ad846). Read the pinned curation Skill, reference card and shared verifier/review requirements. Only physical methyl phenyl sulfide/H2O2 flow experiments are in scope; simulator benchmarks are excluded. Seeking the actual publisher ESI section 4 and author manuscripts; the article landing page returned HTTP 403 to web retrieval, while the indexed page advertises one 1919 KB supplementary PDF. No bot-check or paywall bypass will be attempted. Agent-assisted extraction will retain a rebuild script, source references, ORD 0.8.3 validator output, and independent review before submission.