Save Minimals

• Guide
• Minimals
• Solution finder

Save minimals are strict minimals, but saved pieces are prioritized. That means the program would rather settle for a 3-solution set that has 100% Save T over a 2-solution set that has 100% Save O (if you tell it to do so).

The program currently being used can prioritize multiple saves. If you ask for save T then O minimals, the solution set will maximize save T before finding solutions with the next specified save, and so on.

Save minimals are currently being found using Marfung37’s PC-Saves-Get program. It requires Python3 and NodeJS and will work off of sfinder’s Path Command the same way strict minimals do.

Go to the Python3 download page. It should be a straight-forward installation.

• When installing, make sure that the option to Add python.exe to PATH is enabled during installation.

Go to the NodeJS download page. It should be a straight-forward installation.

• When installing, make sure that NodeJS and npm are added to PATH. This appears in the Custom Setup step of installation.

In a terminal, enter the following separately

• pip install argparse for parsing input parameters.
• npm install tetris-fumen for working with fumens.
• npm install -g sfinder-strict-minimal for generating minimals.

Solution finder’s Path Command is used to produce a .csv file that is used to find minimals. The sfinder command is identical to a normal path file, but you must specify --format csv (-f csv) and --key pattern (-k p).

The commands for the program is structured similar to sfinder. The following explanations are excerpts of important parameters from the official documentation.

1

py sfinder-saves.py filter -w [Wanted Saves] -p [Pieces Used]

• The --wanted-save or -w parameter details the saves you wish to get from the program. More on how to use this parameter later.
• The --pieces or -p parameter details the pieces being used. It should always be the same as what you used in the sfinder command when generating the path.csv file.
• The --best-save or -bs option prioritizes the first defined save before moving onto the next. For making minimals, always include this in the command. It doesn’t require a value, simply having it in the command is already enough.
• The --solve or -s parameter specifies the type of output. This guide uses its default value which is --solve minimal but it’s notable that you can use --solve unique to get all saves for a certain save if you want.

For most use cases, all you need for specifying wanted saves are:

• -w "I,LS,LSZ" a plain list of saves you wish to get. These are prioritized according to what’s written first.
  • If the path.csv file saves more pieces than specified in the -w parameter, it’ll find all saves that include those pieces: -w "I" may match TILJ, ISZO, ILSZ etc. while -w "LSZ" may match TLSZ,LSZO, etc.
• -w "!O" the “NOT” operator (!). It negates the given specification.
  • As an example, using -w "!O" will match TISZ, LJSZ, etc.– any save that does not contain O.
• -w "S||Z" the “OR” operator (||). It joins two or more saves into a single item, removing prioritization between them.
  • As an example, using -w "T,O" may find a 3-solution set that saves 100% Save T, while -w "T||O" may find a 2-solution set that saves 50% T and 50% O.

Don’t forget -f csv and -k p. Take note of the -p parameter as this will be also used in the filter command. This should create a file in the /output/ folder named path.csv.

1
2
3
4
5
6
7

Input:
java -jar sfinder.jar path -t v115@LhA8GeC8FeD8AeD8JeAgH -p T,*! -f csv -k p

Output:
...
perfect clear percent
  -> success = 100.00% (5040/5040)

From this you can already find minimals. Don’t forget -bs and to re-use the -p parameter that was used in the path command.

1

py sfinder-saves.py filter -w 'O' -p T,*! -bs

The resulting minimals should be the following:

The setup also has a 42.46% chance for saving T, so here’s a second example filter command:

1

py sfinder-saves.py filter -w 'T,O' -p T,*! -bs

The resulting minimals should be the following. If you double-check with cover, this set does max out the setup’s potential 42.46% save T before filling in the rest of the missing cover with save O.