Currently, when loading optimization results from an HDF5 file, the complete results are read into memory. For large result sets, this takes pretty long. It would be convenient if the data was only loaded as needed. E.g., I might only want to access objective function values, without loading some thousand Hessians.
Currently, when loading optimization results from an HDF5 file, the complete results are read into memory. For large result sets, this takes pretty long. It would be convenient if the data was only loaded as needed. E.g., I might only want to access objective function values, without loading some thousand Hessians.