H-Matrix approximation for the operator exponential with applications
H-Matrix approximation for the operator exponential with applications
H-Matrix approximation for the operator exponential with applications
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
102 I. P. Gavrilyuk et al.<br />
<strong>with</strong>respect to j = −N,...,N, to compute <strong>the</strong> <strong>approximation</strong> u H (t).<br />
The second level of parallelisation appears if we are interested to calculate<br />
<strong>the</strong> right-hand side of (4.2) <strong>for</strong> different time values.<br />
4.2 Dynamical systems and control <strong>the</strong>ory<br />
In <strong>the</strong> second example, we consider <strong>the</strong> linear dynamical system of equations<br />
dX(t)<br />
= AX(t)+X(t)B + C(t), X(0) = X 0 ,<br />
dt<br />
where X, A, B, C ∈ R n×n . The solution is given by<br />
∫ t<br />
X(t) =e tA X 0 e tB +<br />
0<br />
e (t−s)A C(s)e (t−s)B ds.<br />
Suppose that we can construct <strong>the</strong> H-matrix <strong>approximation</strong>s of <strong>the</strong> corresponding<br />
matrix exponents<br />
exp H (tA) =<br />
2N∑<br />
0 −1<br />
l=1<br />
γ al e −a lt A l , exp H (tB) =<br />
A l , B j ∈M H,k (I × I,P 2 ).<br />
2N∑<br />
0 −1<br />
j=1<br />
γ bj e −b jt B j ,<br />
Then <strong>the</strong> approximate solution X H (t) may be computed in parallel as in <strong>the</strong><br />
first example,<br />
[<br />
N∑<br />
X H (t) = γ alJ0 γ bjJ0 e −(a lJ 0<br />
+b jJ0 )t A lJ0 X 0 B jJ0<br />
l,j=−N<br />
∑J 0<br />
∫<br />
+ γ alα γ bjα e −(a lα+b jα )t A lα<br />
α=1<br />
δ α<br />
e (a lα+b jα )s C(s)dsB jα<br />
]<br />
Let C be constant and <strong>the</strong> eigenvalues of A, B have negative real parts, <strong>the</strong>n<br />
X(t) → X ∞ as t →∞, where<br />
X ∞ =<br />
∫ ∞<br />
satisfies <strong>the</strong> Lyapunov-Sylvester equation<br />
0<br />
e tA Ce tB dt<br />
AX ∞ + X ∞ B + C =0.<br />
.