This repository contains the source code for the paper:
A Study of Solving Life-and-Death Problems in Go
Using Relevance-Zone Based Solvers*
Accepted by IEEE Transactions on Games
git clone git@github.com:rockmanray/rzone.git
cd rzone/database
unzip radius3.zip
cd dcnn/ELF
unzip elf_v2_weight.zippodman run --rm -it \
-v $HOME:/workspace \
--runtime runc \
--security-opt seccomp=unconfined \
-w /workspace/rzone \
--network="host" \
rockmanray/gorzoneThe Life-and-Death (L&D) problems used in the experiments are provided in the tsumego directory.
For evaluation, each problem is encapsulated in JSON format, which specifies the problem settings such as ko rules, player to move, crucial stones, and other configurations required by the solver.
For 7×7 problems:
./run_7x7.shFor 19×19 problems:
./run_19x19.sh-
All results are saved in JSON format under the
resultdirectory. -
Example:
- For the 7×7 problem
1.json - Using AlphaZero with Relevance-Zone Search (RZS)
- The output file will be:
rzone/result/7x7/AZ_RZ/result_1.json
- For the 7×7 problem
-
The field
NumSimulationsin each result file indicates the number of search nodes required to solve the corresponding problem.