Experiment Index
Every lab route should point to a dataset and limitation.
Open view ->EVIDENCE BROWSER
Do not trust the plot.
Trace the run.
SemiAgora already has a distinctive edge: reproducible experiments and explicit limitations. This page makes that edge easier to inspect by separating experiment routes, source method, corrections, and service boundaries.
TRACE VIEWS
Five ways to inspect a claim.
The evidence layer is the answer to low-trust content. It should be obvious which page is teaching, which page executed a model, which page only links out, and which page receives corrections.
Source Method
Official link, license boundary, and reuse rule are separate.
Open view ->Correction Gate
Metric changes must be logged rather than silently overwritten.
Open view ->Service Boundary
Support means simulation, audit, figure, and language feedback, not fabricated data.
Open view ->Factory Log
Daily automation writes evidence first; publication remains manual.
Open view ->RECENT EXPERIMENT ROUTES
Every row should answer: what was run?
These are pulled from the public research catalog. The next improvement is a filterable evidence table with domain, engine, fidelity, checksum, and route.
DomainExperimentEvidence stateRoute
Cross-domain
Researcher source and citation workspace
client-side public metadata filtering and selection / independent generated-data and rendered-route contracts
Open ->Cross-domain
Public evidence provenance and route-aware search
precomputed verified metadata / independent contract verification
Open ->Packaging / Test / Reliability / Metrology
Packaging and test evidence atlas
explicit one-family-at-a-time browser deterministic gzip CSV evidence replay / four independent no-import packet verifiers, four source-scope locks, deterministic replay parity, aggregate input audit, exact public-v115 hold, preintegration hashes and semantics, gzip replay and canonical restoration checks, CSV and compact-manifest contracts, raw CSV and private-row exclusion, and SHA-256 public-copy parity
Open ->Equipment / Facilities / Factory / Sustainability
Factory, facilities, and sustainability evidence atlas
explicit one-family-at-a-time browser deterministic gzip CSV evidence replay / three independent no-import packet verifiers, three source-scope locks, deterministic replay parity, aggregate input audit, exact public-v114 hold, preintegration hashes and semantics, gzip replay and canonical restoration checks, CSV and compact-manifest contracts, raw CSV and private-row exclusion, and SHA-256 public-copy parity
Open ->Equipment / Metrology / Process Control / Yield
Sensor, metrology, and control evidence atlas
explicit one-family-at-a-time browser deterministic gzip CSV evidence replay / four independent no-import packet verifiers, four source-scope locks, deterministic replay parity, aggregate input audit, preintegration hashes and semantics, gzip replay and canonical restoration checks, CSV and compact-manifest contracts, raw CSV and private-row exclusion, and SHA-256 public-copy parity
Open ->Process / Patterning / Implant / Thermal / Clean / CMP / Mask / Substrate
Patterning and module flow evidence atlas
explicit one-family-at-a-time browser deterministic gzip CSV evidence replay / seven independent no-import packet verifiers, seven source-scope locks, deterministic replay parity, aggregate input audit, preintegration hashes and semantics, gzip replay and canonical restoration checks, CSV and compact-manifest contracts, raw CSV and private-row exclusion, and SHA-256 public-copy parity
Open ->SERVICE BOUNDARY
Evidence audit is allowed. Fake evidence is blocked.
For graduate researchers, this is the operating rule: improve the simulation record, figure logic, citation clarity, and technical explanation. Do not fabricate results, hide failed runs, or rewrite claims beyond the supportable evidence.