a feasible model turns out to be infeasible randomly

[disonvon]

What version of OR-Tools and what language are you using?
Version: 9.8
Language: python

Which solver are you using (e.g. CP-SAT, Routing Solver, GLOP, BOP, Gurobi)
CP-SAT
What operating system (Linux, Windows, ...) and version?
Mac Apple M2 Mac Os 13.0
What did you do?
Steps to reproduce the behavior:
a text model file return different solution status, sometimes it is feasible, sometimes not

What did you expect to see
the feasible status should be consistent, which is feasible in my case
What did you see instead?
as above mentioned
Make sure you include information that can help us debug (full error message, model Proto).

#coding=utf-8
from ortools.sat.python import cp_model
from google.protobuf import text_format
import pickle
import os

def solve(file):
    model = cp_model.CpModel()
    with open(file, "r") as file:
        text_format.Parse(file.read(), model.Proto())

    # Create a solver and solve
    solver = cp_model.CpSolver()
    solver.parameters.max_time_in_seconds = 120
    # solver.parameters.log_search_progress = True
    status = solver.Solve(model)

    # Check the solution
    if status == cp_model.OPTIMAL:
        print('Objective value =', solver.ObjectiveValue())
    elif status == cp_model.FEASIBLE:
        print('A potentially suboptimal solution was found.')
    elif status == cp_model.INFEASIBLE:
        print('The problem is infeasible.')
    else:
        print('ohter status')


if __name__ == '__main__':
    file = './../../output/test.pd.txt'
    for i in range(1, 10):
        print(f'----------iteration: {i}----------')
        solve(file)

    print('done')

----------iteration: 1----------
Objective value = 283621783174.0
----------iteration: 2----------
The problem is infeasible.
----------iteration: 3----------
The problem is infeasible.
----------iteration: 4----------
The problem is infeasible.
----------iteration: 5----------
The problem is infeasible.
----------iteration: 6----------
The problem is infeasible.
----------iteration: 7----------
The problem is infeasible.
----------iteration: 8----------
The problem is infeasible.
----------iteration: 9----------
The problem is infeasible.
done

Anything else we should know about your project / environment

[disonvon]

test.pd.txt.zip

[disonvon]

model file attached above

[lperron]

Can you try with 9.9 ?

[disonvon]

Can you try with 9.9 ?

9.9 is always feasible(optimal) for this model.
BTW, i also encountered the similar cases months ago with 9.4

[lperron]

Yes, bugs + nondeterminism happen. Le jeu. 25 avr. 2024, 07:33, Zhiwei Feng ***@***.***> a écrit :

…

Can you try with 9.9 ? 9.9 is always feasible(optimal) for this model. BTW, i also encountered the similar cases months ago with 9.4 — Reply to this email directly, view it on GitHub <#4199 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACUPL3KKTUSGNGYGVU7UCVTY7CITNAVCNFSM6AAAAABGYEJCP6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDANZWGQYDGNBWGU> . You are receiving this because you commented.Message ID: ***@***.***>