[FIBERALLOC-13] Allow modification of the optimization problem after creation Created: 03/Jun/19 Updated: 28/Nov/19 Resolved: 28/Nov/19 |
|
| Status: | Done |
| Project: | Target to fiber allocation and configuration |
| Component/s: | None |
| Affects Version/s: | None |
| Fix Version/s: | None |
| Type: | Task | Priority: | Normal |
| Reporter: | Martin Reinecke | Assignee: | Martin Reinecke |
| Resolution: | Done | Votes: | 0 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Description |
|
Currently the ETS code builds an integer linear programming problem from the given inputs and determines an optimized solution. This solution can then be analyzed whether it produces fiber collisions during the movement stage. If such events are detected, it is desirable to be able to alter the problem configuration and re-run the optimization, hopefully without a complete restart. I'm preparing a change that exposes the LP variables and constraints in the problem to the user so that they may change their parameters even after optimization and then re-run the optimizer. |
| Comments |
| Comment by Martin Reinecke [ 03/Jun/19 ] |
|
Implementation is taking place on the branch "tickets/ |
| Comment by Martin Reinecke [ 01/Aug/19 ] |
|
I have restructured the code so that an object describing the problem (instead of the full solution) is returned from the netflow module. This object supports writing to disk as MPS file (via`dump()`) and solution (via `solve()`). The problem's variables and constraints are accessible via their names and I plan to make them modifiable after the problem has been built. However I'd be happy about feedback on naming conventions. Please have a look at the code! |
| Comment by Kiyoto Yabe [ 02/Aug/19 ] |
|
Thank you for this update! The idea and naming convention are all fine to me. I have reflected the change to my survey simulator. I didn't check the output in very detail, but the new implementation also works. |
| Comment by Martin Reinecke [ 27/Nov/19 ] |
|
Merged to master. |