What is an appropriate value for Jsprit.Parameter.FIXED_COST_PARAM?


I am having trouble in determining an appropriate value for Jsprit.Parameter.FIXED_COST_PARAM, which is used in this equation as weight_deltaFixCost.

As stated in the paper, the impact of FIXED_COST_PARAM (alpha in the paper) on the solution is very great. They tested with a few values between 0 and 2, but they did not indicate what an appropriate value would be.

I guess it would differ from case to case, but I wonder if it has been tested against benchmark cases and if there would be a general “good” value for it? Or what do you think might be a “good” value based on your experiences?

I have been using 1.0, but have been wondering if there could be a more appropriate value.


Best regards,

This seems related to this raised by Philip.

I printed an output of solution_completeness_ratio and it was being set to zero, which gives too much bias on the first vehicle deployed. To my understanding, the idea is that SCR goes from 0.5 --> 1. I think there is something unintended going on in JobInsertionConsideringFixCostsCalculator.java, which could also explain why you can’t pin down the impact properly.

1 Like

Just opened an issue for this. Will investigate this asap.

1 Like