Conjecture: There exists a positive definite matrix P
i
∈
R
˜n
i
טn
i
that satisfies
P
i
˜
A
i
j
+ (
˜
A
i
j
)
T
P
i
⪯ 0 for all j = 1, . . . , N
if and only if there exists a polynomial G(x) of degree 2i
with G(0) = 0 and G(x) > 0, x ̸= 0, satisfying
∂G
∂x
.A
j
x < 0 for all j = 1, . . . , N
proving that conjecture is left for future work.
VIII. CONCLUSION AND FUTURE WORK
This work presents a procedure to compute non-
homogeneous polynomial Lyapunov functions and invariant
sets for LTV systems. This procedure is based on build-
ing a hierarchy of LTV systems such that the quadratic
Lyapunov function for any system in the hierarchy is a
non-homogeneous polynomial Lyapunov function for the
base level system. These non-homogeneous polynomials are
shown to be effective in LTV systems performance analysis
by computing outer approximation for the reachable sets,
bounds for the impulse response and invariant sets that are
not Lyapunov level sets. In future work, the theoretical re-
lation between the generated non-homogeneous polynomials
and other polynomials computed by means of Sum-of-Square
relaxations will be considered.
REFERENCES
[1] S. Boyd, L. El Ghaoui, E. Feron, and V. Balakrishnan, Linear matrix
inequalities in system and control theory. SIAM, 1994.
[2] A. A. Ahmadi and P. A. Parrilo, “Sum of squares certificates for stabil-
ity of planar, homogeneous, and switched systems,” IEEE Transactions
on Automatic Control, vol. 62, no. 10, pp. 5269–5274, 2017.
[3] P. Mason, U. Boscain, and Y. Chitour, “Common polynomial Lya-
punov functions for linear switched systems,” SIAM journal on control
and optimization, vol. 45, no. 1, pp. 226–245, 2006.
[4] D. Liberzon, Switching in systems and control, vol. 190. Springer,
2003.
[5] A. Zelentsovsky, “Nonquadratic Lyapunov functions for robust sta-
bility analysis of linear uncertain systems,” IEEE Transactions on
Automatic control, vol. 39, no. 1, pp. 135–138, 1994.
[6] M. Abate, C. Klett, S. Coogan, and E. Feron, “Lyapunov differ-
ential equation hierarchy and polynomial Lyapunov functions for
switched linear systems,” in 2020 American Control Conference
(ACC), pp. 5322–5327, IEEE, 2020.
[7] P. A. Parrilo, “Semidefinite programming relaxations for semialgebraic
problems,” Mathematical programming, vol. 96, pp. 293–320, 2003.
[8] Z. Jarvis-Wloszek, R. Feeley, W. Tan, K. Sun, and A. Packard,
“Control applications of sum of squares programming,” Positive poly-
nomials in control, pp. 3–22, 2005.
[9] P. A. Parrilo, Structured semidefinite programs and semialgebraic
geometry methods in robustness and optimization. California Institute
of Technology, 2000.
[10] J. Abedor, K. Nagpal, and K. Poolla, “A linear matrix inequality
approach to peak-to-peak gain minimization,” International Journal of
Robust and Nonlinear Control, vol. 6, no. 9-10, pp. 899–927, 1996.
[11] F. Blanchini, “Nonquadratic Lyapunov functions for robust control,”
Automatica, vol. 31, no. 3, pp. 451–461, 1995.
[12] M. Abate, C. Klett, S. Coogan, and E. Feron, “Pointwise-in-time
analysis and non-quadratic Lyapunov functions for linear time-varying
systems,” in 2021 American Control Conference (ACC), pp. 3550–
3555, IEEE, 2021.
[13] H. Abdelraouf, G.-Y. Immanuel, and E. Feron, “Computing bounds
on l
∞
-induced norm for linear time-invariant systems using homoge-
neous Lyapunov functions,” arXiv preprint arXiv:2203.00716, 2022.
[14] H. Lin and P. J. Antsaklis, “Stability and stabilizability of switched
linear systems: a survey of recent results,” IEEE Transactions on
Automatic control, vol. 54, no. 2, pp. 308–322, 2009.
[15] G. Chesi, A. Garulli, A. Tesi, and A. Vicino, Homogeneous polynomial
forms for robustness analysis of uncertain systems, vol. 390. Springer
Science & Business Media, 2009.
[16] L. Vandenberghe and S. Boyd, “A polynomial-time algorithm for
determining quadratic Lyapunov functions for nonlinear systems,” in
Eur. Conf. Circuit Th. and Design, pp. 1065–1068, 1993.
[17] M. Grant and S. Boyd, “Cvx: Matlab software for disciplined convex
programming, version 2.1,” 2014.
[18] E. Feron, “Linear matrix inequalities for the problem of absolute
stability of control systems,” Ph.D. Thesis, Stanford University, 1994.
[19] C. Klett, M. Abate, S. Coogan, and E. Feron, “A numerical method
to compute stability margins of switching linear systems,” in 2021
American Control Conference (ACC), pp. 864–869, IEEE, 2021.
[20] H. Abdelraouf, E. Feron, and J. Shamma, “Algebraic Lyapunov
functions for homogeneous dynamic systems,” arXiv preprint
arXiv:2303.02185, 2023.
APPENDIX
A. Dimensionality reduction
Using Kronecker product produces redundant states that
significantly increase the dimension of the lifted system. For
example, if x ∈ R
2
, then ξ
2
∈ R
4
such that ξ
2
= x ⊗ x =
x
2
1
x
1
x
2
x
1
x
2
x
2
2
T
. To get rid of the redundant states,
a simple procedure introduced in [6]. This procedure is based
on introducing a new variable η
2
∈ R
3
such that ξ
2
= W
2
η
2
where
W
2
=
1 0 0
0 1 0
0 1 0
0 0 1
.
For second order systems, i.e. n = 2, this method can be
generalized for k ≥ 2 as ξ
k
= W
k
η
k
such that
W
k
=
W
k−1
0
2
k−1
0
2
k−1
W
k−1
and W
1
= I
2
. So, the dimension of the system H
k
in the
hierarchy (9) is reduced from n
k
to k + 1. Without loss
of generality, if M = {A}, then every system
˙
ξ
k
= A
k
ξ
k
in the hierarchy (9) can be written as ˙η
k
= W
+
k
A
k
W
k
η
k
where W
+
k
is the pseudo-inverse of W
k
for k ≥ 1. We can
then use the reduced dimension state vector η
k
to rebuild
the new hierarchy (10). Hence, ˜η
i
=
η
T
1
, η
T
2
, . . . , η
T
i
T
which reduces the dimension of the system
˜
H
i
in the
hierarchy (10) from 2(2
i
− 1) to i(i + 3)/2. Thus, the
reduced dimension system
˜
H
i
will be ˜η
i
=
˜
W
+
i
˜
A
i
˜
W
i
˜η
where
˜
W
i
= diag(W
1
, W
2
, . . . , W
i
).