Advent of Code 2023 Day 11 (BaseA Style)

Discussion thread for day 11 of the Advent of Code - https://adventofcode.com/2023/day/11

Accepted answers

All comments

I'm just wondering if we upload WF here right after we finish, somebody can download, and use them to get answers immediately for just collecting stars...That might be not fair for everybody living in different time zone so I'll upload screenshot meanwhile😅

Spoiler

CoG

I am quite satisfied with the solution I came up with today, it feels very intuitive and simple (although I have little doubt someone will post a much denser workflow at some point). It took me a while to work out the kinks and bugs, but it works now and that's what counts. I am inclined to agree with @gawa, so I will include only a screenshot of my solution for now.

Edit: Adding my workflow now that top 15 have solved at least part 1.

Spoiler

Solution2.yxmd

Tokimatsu

When I created Part 1, I thought this logic might not work for Part 2, but I still had to create a new logic for Part 2.

Spoiler

AkimasaKajitani

My solution.

Spoiler

There is no special techniques.

スクリーンショット 2023-12-11 153327.png

kelsey_kincaid

I built out an entire solution for Part 1 that worked on my sample data and not my real data. Instead of spending time troubleshooting, I decided I didn't like my approach and started from scratch. While this slowed me down substantially for Part 1, it paid off for Part 2.

ETA: I personally don't have a problem with people uploading their workflows and choose to do so. If someone wants to compete in bad faith, they will. Even if we didn't upload any workflows here, Reddit is full of scripts people could run to get answers and then work through the Alteryx later. I'd rather just post my workflow in the spirit of learning and fun competition, but that's my two cents

Spoiler

Part 1: My original approach for expanding the universe was to generate rows, then reshape the data and generate columns. At some point I realized that Generating Rows has historically not served me well in AOC, and that I didn't actually need to have a row in the dataset to get the distances I needed. So instead, I calculated the number of empty rows and columns upstream from each Galaxy (#) and used that to recalculate the coordinates of the galaxy. For the distance calculations, I used a simple macro to generate pairs and subtract the coordinates. Honestly, the macro is overkill, but it was the easiest way for me to figure out how to generate every pair without having duplicate pairs. Though, thinking about it now, I have a few ideas.

Part 2: My rework in Part 1 paid off. All I had to do to answer this question was to change 2 formula tools. Originally whether a row/col needed to expand was a binary flag of 0 or 1. Expanding 1 million times was just a matter of changing my data type and making the expanded value 999999 instead of 1. It's rare that adjusting for part 2 is this straightforward, so I'm pretty happy with this even if I didn't get to part 1 quickly.

Workflow:

Macro:

AOC2023_Day11_KelseyKincaid.yxzp

ApekshaSharma

Spoiler

I will make more effort to clean up my workflows & post them here!

As the days have progressed, I have become quite wary of using Joins & multi-row formula tools in my workflow. Was relieved to see that P1 solution required very few tweaks to work on P2. I have kept Expansion multiple as a manual input which can be updated to solve for both P1 & P2. Also, the max RowID & max ColID of the initial grid is an input in my flow which I kept manual as I was lazy!

Day 10 .png

Qiu

Had to reconstrct the logic for Part 2...

Three things I learned from the AoC of 2023

1. Brutal Force does not work, so dont tempt.

2. One day without AoC is actually making me feel life is so much easier..

3. I tends to be stupid when doing AoC. Was trying to develop a macro for the combination of only 2 lists of numbers..

Spoiler

DaisukeTsuchiya

Intersting problem.

Spoiler

P1 WF was complicated then I realized that WF need to be recreated for P2.
WF is modified so that it can be used both for P1 and P2.

スクリーンショット 2023-12-11 183021.png

AkimasaKajitani

Designer Cloud Solution.

Spoiler

Also I tidied up my Designer Desktop solution of Part 2.

Spoiler

DataNath

Day 11 completed! Despite claiming that I would brute force part 2 at any cost in the WhatsApp group, about 90 minutes into generating over 2billion rows, I realised that this could actually be done in a much simpler way and decided to implement that instead... smart, eh?

Spoiler

Day 11.yxmd

patrick_digan

Spoiler

Part 1 I used Generate Rows, but part 2 I used a multi-row formula. I started to brute force part 2 but 5M x 8M was going to be a bit large...

Day 11.yxmd

SuguruYoshinaga

Day 11 finished.

Spoiler

day11.yxzp

phottovy

My solution:

Spoiler

aoc_ph_2023_11.yxzp

mmontgomery

No macro/generate rows! 0.9 sec for p1/p2

Spoiler

2023_Day11.yxzp

ccanterman

Solved!! Time to head back to D10

Spoiler

Part 2 took way longer than it should have, first because I was doing the "10x larger" to mean "add 10 in ADDITION to the 1 already there". And then I ran into 2 PEBKAC problems

Yoshiro_Fujimori

My Solution

Spoiler

Part 1 can be solved by setting "expansion ratio = 2".
It took 2.0 seconds to run.

caitlynmcintyre

I liked this one, Part 1 and Part 2 just need to adjust the multiplier in the circled tool

Spoiler

11.yxmd

alisonpitt

It's so hard to resist the urge to go back and "fix" part 1 when you find a better solution in part 2!

Spoiler

Day 11 - AP.yxmd

AncientPandaman

no macro day is a happy day.

Day 11.yxmd

Gem

A nice, tiny break from making iterative macros!

In other news, my efforts/desire to limit the number of tools I slap on the canvas is going the same way as my efforts to reduce the number of cookies I'm consuming this season: not particularly well. I'll pick up on both of these things in the new year 😅

Spoiler

Day 11.yxmd

DaisukeTsuchiya

I post again with solution in yxzp.

Spoiler

Day11.yxmd

SeanAdams

Hello Manhattan distance my old friend!

This runs in a few seconds on my machine - so I didn't spend too much time optimizing

Spoiler

Solution:

A few things help here:

You only need to keep the galaxies – the # characters – the empty space can be discarded as long as you know the address of the galaxy
By knowing which columns (or Rows) are empty – you can just shuffle everything along
To measure the distance between galaxies – just use Manhattan distance

I was a bit lazy – and built 2 macros for the bulk-out, one to move rows, and one to move columns – but it was less brain tax.

The prep macro is a little more complex – it splits out rows; columns to make downstream processing easier

The G output is the normal row/col format grid but only for galaxies
The R output is blank rows
The C output is blank columns
And the M output is all cells unaltered for debugging

The Bulk-out macros:

I suspect I could have done this better – but this goes through each row which is empty from the largest to the smallest, and shuffles them downwards. This way you can iterate from the end. If I had spent more time on this – there is probably a non-iterative solution, but this runs lightning quick already.

2023_Day11-Part2.yxzp

estherb47

Very much enjoyed this challenge. In years past, the "shortest distance" challenges didn't make sense. This one did, and I'm thrilled that my part 2 runs in .3 seconds!

Straight math, no macros, no spatial. I do wonder if you could spatially match points to a line, maybe with a buffer or trade area, and count the number of points to take a spatial approach. Don't have the energy to poke through workflows to see.

Spoiler

Using the test data, my initial approach of using the full number, e.g., 100 as the number to add, was slightly high. Decided to subtract one, and it worked. I chalk this up to a lucky guess.

AOC11.yxmd

Laurap1228

Here is my Day 11 solution

Spoiler

Advent of Code 2023.yxmd

Quick Links

This months top contributors

atcodedog05 19598

Qiu 15867

binu_acs 15708

MarqueeCrew 13708

apathetichell 13703