A python library to compute resource-constrained task schedules. Express your scheduling problem in terms of tasks, resources and constraints, ProcessScheduler suggests one/the best schedule that satisfies the requirements.
The computation is based on a set of constraints expressed under the form of first-order logic assertions. Problem solving is performed by the SAT/SMT Z3 Theorem Prover.
User-end documentation available at https://processscheduler.readthedocs.io/
- tasks: zero duration task, fixed duration task, variable duration task, work amount, optional task,
- resources: worker, cumulative workers, workers selection, productivity attribute,
- advanced cost functions,
- buffer: NonConcurrentBuffer,
- task constraints: precedence, start synced, end synced, start at, end at, start after, end before,
- resource constraints: AllSameSelected, AllDifferentSelected,
- everything can be set as optional (tasks, resources, constraints),
- first-order-logic operations (not, or, xor, and, implies, if/then/else) between task or resource constraints,
- builtin and customized indicators (resource utilization, resource cost),
- single and multiobjective optimization (makespan, flowtime, earliest, latest, resource cost, etc.),
- exporters: smtlib2.0, json
- Gantt chart rendering using matplotlib or plotly
Install with pip.
pip install ProcessSchedulerThe Z3 theorem prover is the only required dependency.
Optional dependencies (install either with pip or conda):
- matplotlib (Gantt chart rendering),
- plotly (Gantt chart rendering).
There are some Jupypter notebooks. They can be executed online at myBinder.org
import processscheduler as ps
# a simple problem, without horizon (solver will find it)
pb = ps.SchedulingProblem('HelloWorldProcessScheduler')
# add two tasks
task_hello = ps.FixedDurationTask('Process', duration=2)
task_world = ps.FixedDurationTask('Scheduler', duration=2)
# precedence constraint: task_world must be scheduled
# after task_hello
c1 = ps.TaskPrecedence(task_hello, task_world, offset=0)
pb.add_constraint(c1) # explicitly add this constraint to the problem
# solve
solver = ps.SchedulingSolver(pb)
solution = solver.solve()
# display solution, ascii or matplotlib gantt diagram
solution.render_gantt_matplotlib()
ProcessScheduler uses the following tools to ensure code quality:
- unittests,
- code coverage (coverage.py, codecov.io),
- continuous-integration at MS azure,
- static code analysis (codacy),
- spelling mistakes tracking (codespell)
ProcessScheduler is distributed under the terms of the GNU General Public License v3 or (at your option) any later version. It is currently developed and maintained by Thomas Paviot (tpaviot@gmail.com).