From 3f0d2e1c86c455646e073cf80df4635d20a79fcb Mon Sep 17 00:00:00 2001 From: Alexander Fedotov Date: Mon, 14 Mar 2016 20:38:28 +0100 Subject: [PATCH] report update --- report1/eval.tex | 11 +++++++++++ report1/intro.tex | 2 ++ report1/report1.tex | 5 +++++ 3 files changed, 18 insertions(+) create mode 100644 report1/eval.tex create mode 100644 report1/intro.tex diff --git a/report1/eval.tex b/report1/eval.tex new file mode 100644 index 0000000..131029e --- /dev/null +++ b/report1/eval.tex @@ -0,0 +1,11 @@ +\section{Conclusion} +Detailed benchmarking and thorough comparison of the results still has to be done. +%thoughts on the performance of the object-centered approach are needed here +The coordinate-based approach revealed that the following factors affect the feasibility of solving a particular sokoban field: +\begin{itemize} + \item Overall size of a field + \item Proximity of boxes to the man + \item Number of boxes (to a lesser extent) +\end{itemize} +A possible improvement for the coordinate-based approach is the elimination of outer walls, which can reduce overal field sizes.\\ +In total, the material obtained by implementing two different ways to encode a sokoban solver might give the us the ability to approach the next assignment in a more elaborate way. diff --git a/report1/intro.tex b/report1/intro.tex new file mode 100644 index 0000000..141416a --- /dev/null +++ b/report1/intro.tex @@ -0,0 +1,2 @@ +\section{Introduction} +During the initial discussion, we came to two ideas of possible NuSMV encoding of a sokoban solver: an object-centered approach and a coordinate-based one. According to our hypothesis, neither of them is universally superior. To be more precise, the object-centered solution might be advantageous in case of sokoban fields with fewer boxes, but with bigger empty spaces, whereas the coordinate-based encoding might not be that limited with amounts of boxes, being more limited on the sizes of fields. The following hypothesis motivated us to try both approaches and conduct benchmarks wich can help us building a more versatile hybrid solution later. For both approaches, fully automated model generators were made. diff --git a/report1/report1.tex b/report1/report1.tex index 0898bc2..8ee5620 100644 --- a/report1/report1.tex +++ b/report1/report1.tex @@ -3,9 +3,14 @@ \maketitle \section{Introduction} +\input{intro.tex} + %Object centered approach \input{mart.tex} %Coordinate centered approach \input{alex.tex} + +\input{eval.tex} + \end{document} -- 2.20.1