pyomo VS FLAML

Pyomo: We tried pyomo still using the same matrix representation as above (5-minutes timeslot interval), but still encountered the same difficulty of expressing program durations as constraint. I seem to not able to make a condition inside the constraint declaration such that if this matrix entry is 1, then do this operation.

Alternative: pyomo

I’m looking for something similar to Pyomo for Python. Something that connects on the backend to something like GLPK, CBC, IPOPT. Using Google, I’ve only been able to find a few linear programming libraries. If anyone could point me the right direction, it would be greatly appreciated!

Pyomo, it even has its own book. Additionally, CVXOPT focuses on convex optimization, PuLP on linear programming (it has lots of interfaces for other solvers).

There are some optimization modeling tools, Pulp andScipy are known for linear programming (LP) modeling, CVXOPT and Pyomo for quadratic programming (QP).

To rephrase this in math terms, you're looking for the least expensive possible path that covers every node in your yard. As for tools, if you don't mind programming in python, maybe try this: http://www.pyomo.org/.

For modelling libraries in general-purpose languages, Gurobi's python bindings have the best reputation. But of course Gurobi is very expensive (I have heard about $50k for a fully unrestricted license, plus $10k yearly for support). On the open-source side, besides Google's OR-Tools, there is Pyomo [1] and PuLP [2] in Python (as the article mentions). In Julia, there is JuMP [3], whose development community is extremely enthusiastic.

Traditionally, however, mathematical models were encoded in domain-specific languages. The most prominent one is AMPL [4] which is proprietary. The glpk [5] people have developed a very neat open source clone of AMPL: the GNU MathProg language. For a more modern take on AMPL-type modelling DSLs, look at ZIMPL [6], which is open source as well.

[1] http://www.pyomo.org/

[2] https://coin-or.github.io/pulp/

[3] https://jump.dev/JuMP.jl/stable/

[4] https://ampl.com

[5] https://www.gnu.org/software/glpk/

[6] https://zimpl.zib.de/

I really like the simplicity of this framework, and they hit on a lot of common problems found in other agent-based frameworks. Most intrigued by the RAG improvements.

Seems like Microsoft was frustrated with the pace of movement in this space and the shitty results of agents (which admittedly kept my interest turned away from agents for the last few months). I'm interested again because it makes practical sense, and from looking at the example notebooks, seems fairly easy to integrate into existing applications.

Maybe this is the 'low code' approach that might actually work, and bridge together engineering and non-engineering resources.

This example was what caught my eye: https://github.com/microsoft/FLAML/blob/main/notebook/autoge...

4. FLAML

Improved AutoML - Again, with collaboration from Microsoft Research, we used FLAML to update our existing AutoML solutions. What does this mean for you? You're using the latest techniques but all you need is a problem to solve and some data to get started.

For a complete tutorial, navigate to this Jupyter Notebook: https://github.com/microsoft/FLAML/blob/main/notebook/flaml_automl.ipynb

Microsoft Researchers have developed FLAML (Fast Lightweight AutoML) which can now utilize Ray Tune for distributed hyperparameter tuning to scale up FLAML’s resource-efficient & seamlessly parallelizable algorithms across a cluster.

Looks nice but I wonder if this is practical for non-tiny problems. The papers are a bit hard to follow but it looks like training is restarted with every new architecture choice. As for the library itself, the only large neural net example is a finetune of an NLP model that only searches over ADAM's optimizer params - which could be useful but it's a stretch to call that AutoML.