That being said:
I fully agree that the solver industry as a whole has focused for far too long on global optima for academic requirements, instead of real-world use for the actual business requirements, and how to deal with business objective changes each quarter.
Comments
I beg to differ. Timefold is used daily in real-world operations.
Neither our clusters in the US or EU can afford to go down for a minute, or business operations in logistics, retail and healthcare are impacted.
- Traffic: supported
- Driver preferences: the APIs support
-- Area affinity (soft) and geo fencing (hard)
-- Fairness and load balancing of work (soft)
-- Overtime (soft and hard)
-- Seniors not doing boring junior tasks (soft)
- Continuous feedback: the APIs support
-- Real-time rescheduling (warm starts) as actual execution data comes in.
-- Recommend assignment: scheduling in survival mode
-- Pinning: user stays in control through overriding assignments
-- Explainability: why a certain decision was made
-- Insights: what are the bottlenecks in my schedule - what type of employees should I hire or upskill.
That being said: I fully agree that the solver industry as a whole has focused for far too long on global optima for academic requirements, instead of real-world use for the actual business requirements, and how to deal with business objective changes each quarter.