This issue is related to maxkratz/dissertation-ihtp#44.
- In some cases, GIPS is only used to determine the optimal state of a model without actually changing the model, i.e., the selected GT rules will never be applied.
- For some larger problems, GIPS frequently produced OOM errors, e.g., because the initial model + the matches + the generated MILP problem are too large to fit in the system's memory during the MILP solving process in combination.
If both of the above combinations are valid, one could argue that the initial EMF model and all of the resulting lookup data structures (e.g., type indexer, mapping indexer, ...) could be removed after constructing the MILP problem but before solving it. This could (potentially) enable GIPS to solve large problems that cannot fit within the system's memory right now.
This issue is related to maxkratz/dissertation-ihtp#44.
If both of the above combinations are valid, one could argue that the initial EMF model and all of the resulting lookup data structures (e.g., type indexer, mapping indexer, ...) could be removed after constructing the MILP problem but before solving it. This could (potentially) enable GIPS to solve large problems that cannot fit within the system's memory right now.