# Timeline for Multi-product distribution modeling : too many constraints

### Current License: CC BY-SA 4.0

35 events
when toggle format what by license comment
Sep 14 '21 at 15:37 comment Indeed, thank's for the tips !
Sep 12 '21 at 18:31 comment For n5, omit the explicit constraints and instead use a sparse index set for admissible $(i,j)$ pairs. For n6, you are missing a $j$ index.
Sep 12 '21 at 18:01 comment Here are the other variables (python format) pastebin.com/q14a1jMU
Sep 12 '21 at 17:57 comment You could try with this. $I = \{0,1,2\}$, $J = \{3,4,5,6,7\}$, $K = \{0,1,2\}$, $T = \{0,...,30\}$, $S = \{0,..4\}$.
Sep 10 '21 at 23:47 comment Yes exactly: all the data required to solve the problem ($T,I,J,K,S,V,c,...$)
Sep 10 '21 at 12:55 comment what kind of data would be useful ? you mean a sample of algo inputs for example ?
Sep 10 '21 at 9:17 comment Any chance you could share some anonymized data ? (ideally one small set, and one large set)
Sep 10 '21 at 8:51 history edited
edited tags
Sep 10 '21 at 8:35 comment I am still blocked on this problem of number of constraints, I have rewrite the problem from zero to make it a lot clearer and updated to my actuel code !
Sep 10 '21 at 8:34 history edited
deleted 510 characters in body; edited title
Aug 24 '21 at 20:37 comment Sure ! pierre.fontaine@student-cs.fr
Aug 24 '21 at 20:35 history edited
deleted 175 characters in body
Aug 24 '21 at 16:39 answer timeline score: 2
Aug 24 '21 at 14:59 comment @PierreFtn may I contact you by mail ?
Aug 24 '21 at 11:45 comment A time step is a day, it is typically the time of a delivery since the real problem takes place in France (which is app a square of 800km x 800km). This is custom products so it is very difficult to define a typical day. For confidentiality reason I can't reveal what type of product it is, but it is something you buy once in your life and which is custom made. So each client is unique, and the flow of products is never the same from one day to another.
Aug 24 '21 at 11:43 history edited
Aug 24 '21 at 9:30 comment What is the considered time span ? And what is the production profile ? Would it make sense to define a "typical day", or a "typical week" ? Or is every day very different during the entire time span ?
Aug 10 '21 at 7:54 comment You are right, this is something we don't want to be allowed ! I also have this problem since I code it "thinking" in matter of inventory. I must find a constraint that guarantees that vehicles stop making "teleportation"
Aug 9 '21 at 20:49 comment I'm not sure your "coherence of routing" constraints are adequate. Suppose that $\textrm{road}_{i,j,1,1}=1$ for $(i,j)=(1,2), (2,1), (3,4), (4,3)$ and 0 otherwise, with $\textrm{visit}_{j,1,1}=1$ for $j=1,\dots,4$ and 0 otherwise. It seems to satisfy the constraints, but the vehicle is traveling two disjoint loops. Is that allowable?
Aug 8 '21 at 22:51 history edited
Aug 8 '21 at 17:12 comment Thank's I will check that as well :)
Aug 8 '21 at 10:20 comment I have worked on a very similar problem and solved it as a multi commodity flow problem with transshipment constraints (and other side constraints). I generated a set of « good » routes and used a decomposition approach to select the best routes.
Aug 7 '21 at 18:59 history edited
added 58 characters in body; edited tags
Aug 7 '21 at 18:22 answer timeline score: 2
Aug 7 '21 at 18:07 comment I think you confused factories 1 and 2 at one point in the previous comment, but I believe I get the general idea.
Aug 7 '21 at 0:34 comment It can sit in a factory indeed. Let's imagine we must deliver two products next to factory 2 but the first one can only be produce in the factory 1. Let's now imagine the second product is produced 5 days after the first one. It is possible to transfer the second product from factory 1 to factory 2, then wait 5 days to make a grouped delivery with the two products from factory 2 (which will be cheaper than two separate deliveries)
Aug 7 '21 at 0:32 comment Thanks for the edit !
Aug 6 '21 at 19:41 comment You mention "stores" and "factories" (which I assume are the same), and you mention transferring product between factories. If a product is transferred from factory 1 to factory 2, must it immediately go on another vehicle, or can it sit at factory 2 until the optimization gods decide it is time to move it ahead?
Aug 6 '21 at 19:38 comment I edited your objective function -- you did not have the vehicle index in it.
Aug 6 '21 at 19:38 history edited
Added missing index in the objective.
Aug 6 '21 at 16:44 comment thank's for the tips ! It must be way easier to read now :)
Aug 6 '21 at 16:44 history edited