General Discussions
Discuss any topics that are not product-specific here.
I don't think my WF is not so optimized that I admit there is a room for improvement, but I'm satisfied with Macro-less solution at the end.
For Part2, all of connection ID are memorized as string by using the Multi-row formula tool, and check first record that contains all of junction boxes.
Even using my powerful computer, it takes 2 minutes. I wanna see other solution for learning wise algorithm.
While total 499,500 connections are there, target result is hit at very early examination, that is 4,765th. My WF should have some waste calculation, though I don't figure out how to get rid of them yet.
Day 8 was a problem I really enjoyed.
- For P1, I generated all possible pairs, calculated their distances, and then took the 1,000 shortest pairs. I used the Make Group Tool to group them into circuits.
- For P2, I needed to find how many pairs it would take to form a single group. I first used trial and error to narrow down the range to 4,600-4,700 connections. After that, I used a Batch Macro to identify the exact number of pairs needed.
- Since I ran the workflow on this narrowed range, the processing time was 4 seconds.
Wow! Seeing non-macro solutions is incredible! I built a macro today to solve via Binary Search algorithm!
For Part 2, I built a binary search algorithm macro. For speed I solved with manual binary search.
Happy Solving!
It's been resolved, but I expect a smarter solution for the Day 8.
In part 2, I thought that by evaluating at the midpoint and discarding the branch with no solution, we could explore half the cases at a time and likely find the answer in far fewer than 50 loops. I also confirmed that processing all combinations in the grouping tool took only seconds.
After that, it took time to implement the desired logic and correctly set the completion condition for the iterration macro.
I briefly considered whether manually narrowing down the likely solution range and then checking that range might be faster. However, since it seemed possible to manage without building a highly data-dependent workflow, I decided against that approach.
Back on track today
Basically ran the same logic through an iterative macro.
I dropped the coordinates early in part 1, so rather than change part 1 I just join them back on to my phase 2 macro outputs.
I could probably have made the iterative macro run less iterations, but this works.
