Changeset 51 for liacs/nc/low-correlation/report.tex
- Timestamp:
- Dec 19, 2009, 6:03:12 PM (15 years ago)
- File:
-
- 1 edited
Legend:
- Unmodified
- Added
- Removed
-
liacs/nc/low-correlation/report.tex
r46 r51 18 18 \usepackage{color} 19 19 \usepackage{subfig} 20 \usepackage{marvosym} 20 21 \floatstyle{ruled} 21 22 \newfloat{result}{thp}{lop} … … 56 57 57 58 \section{Problem description} 58 Due to the huge 'exploding' pos ibilities it is not possible to walk trough the59 whole list of pos ibities, so we need alternative approches to tackle this60 problem. First we will try the \emph{Monte Carlo Search algorit m [MCS]} and next try59 Due to the huge 'exploding' possibilities it is not possible to walk trough the 60 whole list of possibilities, so we need alternative approaches to tackle this 61 problem. First we will try the \emph{Monte Carlo Search algorithm [MCS]} and next try 61 62 \emph{Simulated Annealing [SA]}. 62 63 63 64 \emph{MCS} is all about random sequence generation and trying to find a good 64 solution, do a small a justment on the solution and compare the new solution65 solution, do a small adjustment on the solution and compare the new solution 65 66 again. 66 67 67 \emph{SA} takes a more and less the same ap roch, but it also accept small68 \emph{SA} takes a more and less the same approach, but it also accept small 68 69 'losses' from time-to-time. But as time passes it get less likely to accept 69 70 'bad solutions'. … … 71 72 \section{Statistics} 72 73 Of course many people have run numerous computer hours on finding the best 73 possible fitness as shown in table~\ref{tab:best}. The algorit ms used to find74 possible fitness as shown in table~\ref{tab:best}. The algorithms used to find 74 75 those numbers are not found. 75 76 … … 92 93 93 94 \section{Approach} 94 The \emph{MCS} is implemented str eight-forward from the wikipedia95 The \emph{MCS} is implemented straight-forward from the Wikipedia 95 96 page\footnote{http://en.wikipedia.org/wiki/Monte\_Carlo\_method}. The mutation 96 97 choosing is flipping one bit in the array. 97 98 98 For the \emph{SA} implemented also comes str eight from 'the book'99 For the \emph{SA} implemented also comes straight from 'the book' 99 100 \footnote{http://en.wikipedia.org/wiki/Simulated\_annealing}, the process 100 choosing for the cooling-down sequence is taken from the dutch wikipedia page101 choosing for the cooling-down sequence is taken from the dutch Wikipedia page 101 102 'Simulated annealing', which is nice indicator when to choose something when 102 the solution is worse (logically, bet er solutions will always be accepted).103 the solution is worse (logically, better solutions will always be accepted). 103 104 \footnote{http://nl.wikipedia.org/wiki/Simulated\_annealing}: 104 105 \begin{math} … … 119 120 120 121 \section{Results} 121 All experiments are run 5 times the best solution is cho osen and will be122 All experiments are run 5 times the best solution is chosen and will be 122 123 resented at table~\ref{tab:result}. Iteration size is set to 1000. For $n=20$ 123 124 the best fitness history is shown at figure~\ref{fig:fitness}. … … 156 157 temperature function is not working as expected. 157 158 158 Both algorit ms are preforming much worse then the best found solutions.159 Both algorithms are preforming much worse then the best found solutions. 159 160 \section{Appendix 1} 160 161 \label{app:code}
Note:
See TracChangeset
for help on using the changeset viewer.