puzzle-and-dragons-optimal-boards

An optimized brute-force searcher for the game, Puzzle & Dragons, to search for all high-combo layouts when there are 15+ orbs in the same color on a 5 x 6 board.

Results

See all the optimal boards here

Features

Finding Optimal Boards

• multi-processing
• speed-up tricks
  • only compute on one of the two symmetric boards
  • only compute on one of the equivalent boards
  • estimate the upper limit of number of combos before calculating the true number of combos, so we can prevent computing on obviously low-combo boards.
• output results in JSON format

Visualizing Boards

• visualize the results with a webpage
  • written with React.js + Webpack
• orb matching animation
  • using @powder0326's website

How to Use

Requirements

• For finding optimal boards, you need python3 and pypy3
• For visualizing the boards, you need node.js and npm.

Step 1. Find Optimal Boards Given an Orb Combination

• For example, I want to find all boards with 21+5+4 orbs with more than or equal to 8 combos. And I want to multi-process it with 4 threads.

  $ cd find_optimal_boards
  $ pypy3 find_optimal_boards.py 21 5 4 8 -t 4
  

  It will output a report.json file in find_optimal_boards/output/21-5-4/.

• The program runs with 4 threads by default, so you can also go with

  $ pypy3 find_optimal_boards.py 21 5 4 8
  

• We recommend using pypy3 because it's about 10x faster than python3 in our case. But if you don't have pypy3 installed, you can still run with python3. For example,

  $ python3 find_optimal_boards.py 21 5 4 8
  

• While the program is running, it outputs some temporary results, like fixed-0-2.json, in find_optimal_boards/output/21-5-4/. If the program stops running for some reasons, the next time you run the program, it will automatically resume from the previous breakpoint.

Step 2. Simulate Orb Matching and Find the Best Board

• After find_optimal_boards.py finishes running, you get a report.json file in a folder named with the orb combination, like find_optimal_boards/output/21-5-4/report.json

• First, rename the folder to 21-5-4/ done_21-5-4/ to let our programs know that it's a finished one.

• Run

  $ pypy3 add_properties.py
  

  to simulate orb matching and get some statistics like average number of combos and average damage of each board. It will read the boards in find_optimal_boards/output/done_21-5-4/report.json, do the simulations, add some properties to it, and output another json file find_optimal_boards/output/compact/done_21-5-4/report.json.

Step 3. Collect the Boards with Highest Damages

• After you have run Step 1 and 2 for multiple orb combinations, you get many report.json. It's time to get the best boards together, so the PAD players can easily find the boards with highest damages.
• Run

  $ pypy3 collect_optimal_boards.py
  

  and it will read all the report.json in find_optimal_boards/output/compact/done_xxx/ and extract the boards with highest damages and output them to find_optimal_boards/output/compact/optimal_boards.json