Don't tell the editors, but this feels like the weirdest paper I've ever written.

🌱 See how Nextmv helps KaleAI streamline urban deliveries for a greener, more efficient future.

"To add to that, a cargo bike basically has no parking issues. You can just park it right outside the door. So the whole process of finding parking and walking to the door is basically eliminated." ☝️I cannot stress enough the impact of park and embark times on last mile.

I've seen urban areas where nobody with a car wants to pick up an order because there's nowhere to park. 🤷 It makes sense. Drivers have to park illegally. Overeager parking enforcement slaps them with a $30 ticket on a $10 payout.

Or consider delivering a pizza to a suburban hospital or corporate office park. You have to drive around a big parking lot to find a spot, walk to the front door, talk to security, ... 🚲 Much better just to leave your bike outside the front door.

You can take this lens to any on-demand system. Identify anything that isn't directly serving the end user as overhead, and cut it out. ⏲️ In meal delivery, for example, trim any time spent that doesn't involve either making the food or taking it directly to the diner.

Our book "Combinatorial Models for Scheduling Sports Tournaments" (Ribeiro, Urrutia, de Werra) is now published in the EURO Advanced Tutorials on Operational Research series. Thanks to Michael Trick for writing the foreword and @SperanzaGrazia as its editor link.springer.com/book/10.1007/9…

In September we introduced a whirlwind of new features for #DecisionOps in the Nextmv console. Want to run a custom decision app directly from the console? Want to perform an acceptance test on a prebuilt decision app? Check and check. Dive in >> hubs.ly/Q024sZgq0

FOMO

Josh Bongard the concept of "pulling on a maze to solve it" is a consequence of duality theory applied to the shortest path problem: cs.ucr.edu/~neal/1998/cos…

At Nextmv we do #optimization speed runs. In a couple weeks, I'll up the ante and do 3 in 30 minutes. Come see the 3 main models you need to operate an #ondemand #lastmile service using 3 solvers: #ORTools, @HighsOpt, and our own #nextroute. nextmv.io/videos/forecas…

I'll be showing how to build and deploy the most important models for on-demand delivery services using #ORTools at #PyData Global next month! Get your tickets here. #orms ti.to/pydata/pydata-…

The other night my 2 year old wanted to bake. We got partway into a batch of banana bread before I realized we were out of brown sugar. Gopuff brought me some quickly enough that we popped it into the oven before she went to bed. What was life like before ultrafast delivery? 🤔

🧟 #ORMS tweeps: is anyone out there using one of these packages? I'd love to pick your 🧠. - Gurobi ML - PySCIPOptML - another ML embedding library I don't know about...

Fun but kinda true: listen to the latest FiveThirtyEight podcast and every time you hear "poll" think "benchmark." open.spotify.com/episode/2flJUP…

That feeling when #GenAI shows you the #Python #ML world you wish you lived in, rather than the world you actually live in.

😱 The formulation? Truly scary, 👹 And the code was rough and hairy. 🤷 Still, I gave it to the solver, 😨 Hoping it would soon be over. 😡 But the solver sent it back, 👻 And "Infeasiboo!" it spat. 🎃 Happy Halloween!

A simple example of Gurobi Optimization's hierarchical optimization features in action, which let you make trade-offs between different objectives in a sensible way. Spoiler alert, it's pretty nice! ryanjoneil.github.io/posts/2024-11-… #orms #datascience

The "Emmy Winner" bit is so cool.

This, TBH.