28.06.2014 Views

sparse grid method in the libor market model. option valuation and the

sparse grid method in the libor market model. option valuation and the

sparse grid method in the libor market model. option valuation and the

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

6.1.1 Convergence of MC, Full <strong>and</strong> Sparse Grid solutions.<br />

MC convergence <strong>and</strong> its st<strong>and</strong>ard error are shown <strong>in</strong> Figure (6.1). Simulation was<br />

performed for <strong>the</strong> maximum of 4, 000, 000 trials, reach<strong>in</strong>g m<strong>in</strong>imum st<strong>and</strong>ard error of<br />

0.022. FD solution for full <strong>grid</strong> discretization was computed for up to discretization<br />

level L = 9. Figure (6.1) shows conv<strong>in</strong>c<strong>in</strong>g convergence behavior of simulated solution<br />

Level of Discretization<br />

6 6.5 7 7.5 8 8.5 9<br />

10.2<br />

MC Simulation Value+Std.Error<br />

Full-Grid FD value<br />

10.15<br />

10.1<br />

Option price<br />

10.05<br />

10<br />

9.95<br />

9.9<br />

9.85<br />

500000 1e+006 1.5e+006 2e+006 2.5e+006 3e+006 3.5e+006 4e+006<br />

Number of MC simulations<br />

Figure 6.1: 2D Chooser Option. MC vs. FD Convergence.<br />

<strong>and</strong> full-<strong>grid</strong> FD, provid<strong>in</strong>g accurate reference values for fur<strong>the</strong>r tests. For our <strong>sparse</strong><br />

approximation experiments we have used a classical <strong>sparse</strong> <strong>grid</strong> (a.k.a ”maximumnorm-based”<br />

type) <strong>and</strong> a basic comb<strong>in</strong>ation technique, both described <strong>in</strong> <strong>the</strong> previous<br />

chapter. Table (6.3) <strong>and</strong> Figure (6.2) give idea of FD convergence on full <strong>and</strong> <strong>sparse</strong><br />

<strong>grid</strong>s 2 :<br />

Level<br />

Full <strong>grid</strong> Execution Sparse Grid Execution Full <strong>grid</strong> Sparse <strong>grid</strong><br />

solution time(sec.) solution time(sec.) po<strong>in</strong>ts count po<strong>in</strong>ts count<br />

5 10.5969 1.5 11.351 0.36 1089 257<br />

6 10.1755 6.6 11.477 0.88 4225 577<br />

7 10.0964 30.1 9.8731 2.16 16641 1281<br />

8 10.0762 176.4 10.005 5.18 66049 2817<br />

9 10.0762 651.4 10.165 12.89 263169 6145<br />

10 10.153 31.64 1050625 13313<br />

11 9.998 74.86 4198401 28673<br />

Table 6.3: 2D Chooser. Convergence values for full <strong>and</strong> <strong>sparse</strong> <strong>grid</strong> solutions.<br />

2 The execution time for <strong>sparse</strong> <strong>grid</strong> technique does not <strong>in</strong>clude <strong>in</strong>terpolation procedure<br />

45

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!