A large MIP model that should be solved as LP: the root node

In this post, I want to discuss an observation about the root node when solving a MIP model.

Problem Description

The model I want to tackle here is from [1]:

We have a large matrix $\color{darkblue}A=\color{darkblue}a_{i,j}$ with values 0 and 1. In addition, we have a minimum on the row and column totals. These are called $\color{darkblue}r_i$ and $\color{darkblue}c_j$. The goal is to remove as many 1's in the matrix $\color{darkblue}A$ subject to these minimum row and column totals.

Some GAMS embedded Python notes

Here are two issues you may want to be aware of. The discussion below is relevant for Windows and not so much for Unix variants.

Using Python Raw strings for directories

Here we use some Python inside an otherwise empty GAMS model:

$onEmbeddedCode Python: dir = r"%system.fp%" print(dir) $offEmbeddedCode

In-process, in-memory databases

There are a few database systems that are a bit different. They are libraries that can be linked directly to your application. Linking can be done statically (during the compilation/linking step) or dynamically (using a shared library or DLL). Here I want to show two cases:

• SQLite [1] used from R on data frames
• DuckDB [2] used from Python, again on data frames

So these databases don't only run inside R or Python but also can operate directly on data frames.

Compression and large tables (Excel and CSV)

Here is an interesting little experiment: load a large CSV file into Excel. My original Powerpoint slide from a recent presentation was not as complete as it should be. Here is more size info: