Welcome to the monthly release roundup. We had a little hiatus and a lot has shipped, so let’s hop right to the good stuff! And if you want to see a live demo of it all and more, join us for “Decision model, meet production” with our CTO Ryan O’Neil.
The Nextmv CLI and SDK
The Nextmv platform is changing the way optimization and operations teams build, test, and deploy decision models. This release featured a dev-friendly CLI and customizable SDK, making it easier to go from zero to customizable decision model in minutes.
The CLI provides an opinionated workflow for the decision optimization lifecycle. Its menu of commands guide you to your next move for all the scaffolding needed to create, manage, test, and maintain decision optimization as a service.
The refreshed Nextmv SDK doubles down on making the process of decision modeling more accessible to more users with a “decisions as code” approach. This developer-based mindset allows users to build robust decision models more like operators rather than mathematicians and reduces translation between business logic and mathematical representations.
You can get started with the CLI and SDK in five minutes with a free Nextmv account and a few commands.
Learn more about this release with the announcement blog, in a 30-minute techtalk (a nice watch for a mid-day brain break), and in the documentation.
Interface for MIP solving
Nextmv provides an extensible interface for using different optimization solver paradigms with Nextmv. We’ve added MIP solving capabilities to Nextmv (to accompany decision diagrams and heuristics via Nextmv’s Hop solver) to give even more flexibility to model and solve many types of optimization problems.
This first MIP-based solving paradigm we’ve exposed uses HiGHS, an MIT-licensed open source linear optimization solver. With a few commands, you can build and run customizable MIP models (for meal allocation and knapsack) in minutes.
Learn more about this release in this blog post, in this 6-minute video featuring the engineer behind the work, and in the documentation.
More pre-built decision models
With decision optimization, we think it’s nice to have a helpful starting point rather than diving head first into the deep end. That’s why we provide fully functional, pre-built decision models (we call them templates) that help users orient themselves to a familiar problem domain in the context of Nextmv.
Some of the templates we’ve recently released include:
- Routing: For a range of vehicle routing problems (VRPs) such as food delivery and mobility
- Shift scheduling: For assigning workers to shifts across healthcare, call centers, sports leagues
- Knapsack (with MIP or Hop): For solving packing problems such as warehouse binning
- PagerDuty: For on-call rotation scheduling designed to send an output to the PagerDuty API
You can see a full list of templates in our documentation.
Templates provide a way to get started without beginning from scratch (though you can if you want, and we even provide a “fill in the blanks” template for that). Each template is fully customizable to adapt to unique business logic and rules. Simply initialize from the CLI to run, experiment, learn, and then make it your own.
We recommend getting started with this approachable walkthrough about optimizing on-call PagerDuty rotation schedules.
The end(dive) for now…
This section header will make more sense if you check out the MIP model featuring meal allocation. (Plus, you’ll learn about binkies!) Over the last few weeks, we also made smaller updates to Go versions and added cost measures. See the full list in our release notes.
Lastly, we also had a chance to participate and present at INFORMS 2022 in Indianapolis in October. It was great to host our interactive workshop (an online version is coming soon!), attend sessions, and meet with people in the community.
Stay current with the latest by subscribing to updates (form’s in the footer below) and follow us on social (LinkedIn, Twitter, YouTube). There’s more good stuff on the way.
.jpg)
