Mathematica 教學：Policy function iterations for deterministic cake eating problem

 

Policy function iterations方法是通過計算出最佳化問題的每一步所實現的最佳結果，並將這些結果用於計算下一步的最佳結果。此過程在一定數量的迭代後會得到一個最佳策略，並且這個策略也是最終的最佳解。因此它對於解決複雜的最佳化問題非常有用。它可以被用於解決各種經濟學、工程學和其他領域的最佳化問題。

ClearAll["Global`*"]
\[Alpha]=0.3;
\[Beta]=0.95;
\[Delta]=0.1;
\[Sigma]=1.5;

size=99;

K=Range[0.1,5,4.9/(size)];
V=Table[0,{size}];


astHoward=With[{n=size},
obj=ParallelTable[If[i^\[Alpha]+(1-\[Delta])*i-j<0,-10^9,
((i^\[Alpha]+(1-\[Delta])*i-j)^(1-\[Sigma])-1)/(1-\[Sigma])],
{i,0.1,5,4.9/n},{j,0.1,5,4.9/n}];
loops=0;
\[Epsilon]=99999;

Print[Grid[{
{"loop number ",
Dynamic[loops],
" with distance ",
Dynamic[NumberForm[\[Epsilon],
ExponentFunction->(If[-10<#<10,Null,#]&)]]}}]];

NestWhile[
NestWhile[
Transpose@
Table[{obj[[i,#[[2,i]]]]+\[Beta]*#[[1,#[[2,i]]]],#[[2,i]]},{i,Length[#[[1]]]}]&,
Transpose[SortBy[#,First][[-1]]&/@
Table[{obj[[i,j]]+\[Beta]*#[[1,j]],j},{i,n+1},{j,n+1}]//Parallelize],
Norm[#1[[1]]-#2[[1]]]>10^-6&,2]&,
{Table[0,n+1],Table[0,n+1]},
(loops++;\[Epsilon]=Norm[#1[[1]]-#2[[1]]];
Norm[#1[[1]]-#2[[1]]]>10^-6)&,2]
]


astHowardMatrix=With[{n=size},
obj=ParallelTable[If[i^\[Alpha]+(1-\[Delta])*i-j<
0,-10^9,((i^\[Alpha]+(1-\[Delta])*i-j)^(1-\[Sigma])-1)/(1-\[Sigma])],
{i,0.1,5,4.9/n},{j,0.1,5,4.9/n}];
loops=0;
\[Epsilon]=99999;

Id=IdentityMatrix[size+1];

Print[Grid[{
{"loop number ",
Dynamic[loops],
" with distance ",
Dynamic[NumberForm[\[Epsilon],
ExponentFunction->(If[-10<#<10,Null,#]&)]]}}]];


NestWhile[
(
myValue=Transpose[SortBy[#,First][[-1]]&/@
Table[{obj[[i,j]]+\[Beta]*#[[1,j]],j},{i,n+1},{j,n+1}]//Parallelize];
{Inverse[(Id-\[Beta]*Normal[SparseArray[
Rule[#,1]&/@Transpose[
{Range[size+1],myValue[[2]]}],size+1]])].(obj[[#[[1]],#[[2]]]]&/@
Transpose[{Range[size+1],myValue[[2]]}]),myValue[[2]]}
)&,
{Table[0,n+1],Table[0,n+1]},
(loops++;\[Epsilon]=Norm[#1[[1]]-#2[[1]]];
Norm[#1[[1]]-#2[[1]]]>10^-6)&,2]
]


ListLinePlot[astHowardMatrix[[1]],PlotRange->All,AspectRatio->1,Frame->True]
ListLinePlot[astHowardMatrix[[2]],PlotRange->All,AspectRatio->1,Frame->True]

Tags: Deterministic cake eating problem, Dynamic programming, Mathematica 教學, Mathematica 教學 程式, Polic function iterations

• 編輯文章
• 此文章網址
• 轉寄此文章
• 目前有0 意見