Advent of Code 2020 - BaseA Style (Day 15)

[Jean-Balteryx]

Discussion thread for day 15 of the Advent of Code : https://adventofcode.com/2020/day/15

[Greg_Murray]

I only completed part 1. Interested to see if there are baseA solutions for part 2. My macro is pretty slow (~3.5 minute for the input). I am sure I overcomplicated it.

Saqueador

workflow

Marco

[jdunkerley79]

Part 1 in BaseA, part 2 resorted to Abacus functions (VarNum)

Saqueador

iif([RowCount]<=[Init],
  GetWord([Field1], [RowCount]-1)
  + iif([RowCount]>1 AND VarNum([Row-1:Value], RowCount - 1),"",""),
  ToString(
    iif(VarNumExists([Row-1:Value]),
      RowCount-VarNum([Row-1:Value])-1,
      0)
    + VarNum([Row-1:Value], RowCount - 1) * 0
   )
)​

[Steph_Maddrell]

My part 1 solution based on iterative macro. Runs in 3.5 mins so I can't see that I'll be solving part 2 any time soon!

Workflow:

Saqueador

Macro:

Saqueador

[dsmdavid]

On hindsight, this was not the right day to play golf...

Saqueador

Part I, the browse is needed to be able to scroll down all the rows

Part II... the logic would hold, but it's not really efficient. I've switched to an iterative macro that runs ~ 300 iterations per second, so maybe I'll have some results in one day...😅 EDIT: running 100k took almost 20 min, so the timeframe for solutions has moved to 1 week... I think I'm giving up on BaseA for this one.
I do not give up that there must be a pattern to be found. I've just not seen it yet.

[AkimasaKajitani]

Day15!

My Part 1 Macro takes 4 minutes 30 seconds, but I enabled AMP Engine it is 10.2 seconds.

Saqueador

Macro:

After that I made non Macro version at Part 1.

This takes 0.7sec(no AMP). When enabled AMP, 0.5sec.

Saqueador

But at Part 2, when I enabled AMP Engine, Multi-Row Formula tool throw the error "String variable switched type". 

So I have tried another macro, but it takes longer(Now Running).

[danilang]

Part 1 with no macro runs in 0.6 seconds.  It generates really long strings though so it bogs down when trying part 2

Saqueador

Dan

[Balders]

My part 1 macro ran in 4s for 2020 rows, gonna resort to python again for part 2

[Balders]

Here's me for Day 15 with the Part 2 python inside an Alteryx workflow.

The Alteryx Macro takes 4s for 2020 iterations with AMP on inside the macro.

The Python for p2 takes 21s for 30m iterations.

Saqueador

In the Macro I use negative record IDs because they helped with summarise sorts but then I managed to get rid of summarise tools and filtered out old rows more efficiently with the sample tools. However I've left the negative record IDs in because why not. That's why you see record ID - 1 in the calcs. 

Main Workflow


Macro

Python 

#################################
from ayx import Package
from ayx import Alteryx
import pandas as pd

#################################
li = Alteryx.read("#1")["data"].tolist()

next = 0
dict = {n: i+1 for i, n in enumerate(li)}
target = 30000000

for i in range(len(li)+1,target):
    val = next
    next = i - dict.get(val,i)
    dict[val] = i

df = pd.DataFrame([next],columns=["output"])

Alteryx.write(df,1)

[dsmdavid]

I ended doing the second part in Python.
Thanks, @Balders, for the link to the video about this sequence https://youtu.be/etMJxB-igrc