Capturing Computational Experiments requires:
- Domain knowledge
-
- What structure? What calculation type? etc
- Technical knowledge
-
- Which computer? How much CPU/memory? What batch system? etc
We minimise such distraction by wrapping the technical details into a lite electronic notebook
 |
 |
 |
| Listing jobs |
Starting a new jobs |
Configured programs |
An imposed structured workflow captures all input and output data to provide a full record context and future reproducibility for the calculation. The golden moment ensures no additional burden for the user by capture:
- Naturally-arising from the workflow
- User name, datestamps, free-text descriptions, etc
- Silently created as part of the computation
- CML (XML) representations of structures
- Portable application-specific data formats (eg Gaussian FCHK)
- Metadata tags abstracted from input/output files (InCHI, SMILES, calculation type, etc )