MIP Heuristics log is not detailed or explanatory enough

[bbozkaya]

When running MIP Solver with Heuristics only mode (or together with B&B), more information is needed regarding the iterative process of the algorithm

Specifically, it would be ideal to report best feasible MIP solution found every N iterations, similar to how PDLP algorithm logs its progress. This is critical for tracking the algorithm progress, especially to see when it finds new feasible solution, and what it finds at certain time intervals such as 1 min, 3 min, 5 min, etc.

Furthermore, there are ambiguous or incomplete log messages such as:
"Enough solutions couldn't be generated,exiting heuristics!"

When this happens, there is no additional information provided. Did the Heuristic algorithm find any solution at all? If it did, why not report it? If not, be more specific and say that no feasible solution could be generated.

[anandhkb]

@bbozkaya Please try with latest 25.10 and report if the logs are explanatory or not.

[anandhkb]

Advise from @bbozkaya

TLDR - The issue is NOT addressed in 25.10.

Burcin ran MIP solver in both modes (heuristics only and concurrent). I still do not see any informative log entries for when feasible/incumbent solutions are encountered. For instance, for the heuristics only mode, I get this:
Setting CUDA_MODULE_LOADING to EAGER
Setting parameter log_file to log2.txt
Setting parameter mip_heuristics_only to true
Setting parameter time_limit to 9.000000e+02
Reading file dcap_model.mps
cuOpt version: 25.10.0, git hash: 6d22826, host arch: x86_64, device archs: 90a-real
CPU: Intel(R) Xeon(R) Silver 4314 CPU @ 2.40GHz, threads (physical/logical): 16/32, RAM: 233.15 GiB
CUDA 13.0, device: NVIDIA H100 80GB HBM3 (ID 0), VRAM: 95.07 GiB
CUDA device UUID: ffffffafffffffc3ffffffc15a-48ffffffa

Solving a problem with 57989 constraints, 1134589 variables (1134589 integers), and 11313038 nonzeros
Problem scaling:
Objective coefficents range: [8e-03, 6e+00]
Constraint matrix coefficients range: [1e+00, 1e+00]
Constraint rhs / bounds range: [1e+00, 1e+00]
Variable bounds range: [0e+00, 1e+00]

Original problem: 57989 constraints, 1134589 variables, 11313038 nonzeros
Calling Papilo presolver
Presolve status: reduced the problem
Presolve removed: 11 constraints, 242562 variables, 2198937 nonzeros
Presolved problem: 57978 constraints, 892027 variables, 9114101 nonzeros
Papilo presolve time: 61.712676
Objective offset 5.000000 scaling_factor 1.000000
Running presolve!
After trivial presolve: 57978 constraints, 892027 variables, objective offset 5.000000.
Post-solve status: succeeded
Solution objective: 15850.324791 , relative_mip_gap 0.000000 solution_bound 15850.324788 presolve_time 61.716148 total_solve_time 900.116982 max constraint violation 0.000000 max int violation 0.000000 max var bounds violation 0.000000 nodes 0 simplex_iterations 0
which actually reported nothing after the line "After trivial presolve: 57978 constraints, 892027 variables, objective offset 5.000000." until it was finished. When was this solution (iteration, time)? Were there other incumbent solutions found along the way? What does solver find at certain time intervals, eg. 1, 3, 5 min, etc.?