--- title: Port-Hamiltonian Descriptor Systems url: https://www.emergentmind.com/topics/port-hamiltonian-descriptor-systems type: topic --- # Port-Hamiltonian Descriptor Systems Port-Hamiltonian descriptor systems are differential-algebraic systems that combine the energy-based modeling paradigm of port-Hamiltonian systems with the algebraic constraints of descriptor systems. In linear time-invariant form they are typically written as \[ E\dot x=(J-R)Qx+(B-P)u,\qquad y=(B+P)^{T}Qx+(S+N)u, \] or, in a simplified variant, \(E\dot x=(J-R)Qx+Bu,\ y=B^{T}Qx\), with structure matrices satisfying skew-symmetry, semidefinite dissipation, and the compatibility condition \(Q^{T}E=E^{T}Q\ge 0\). The framework encodes interconnection, energy storage, dissipation, and ports in a single descriptor representation, and it is designed to retain power-balance and passivity under constraints, transformations, and interconnections [1705.09081], [2105.08934], [2201.06590]. ## 1. Algebraic definition and power balance A linear time-invariant descriptor system has the form \[ E\dot x = Ax+Bu,\qquad y=Cx+Du, \] with \((E,A)\) regular if \(\det(\lambda E-A)\not\equiv 0\) for at least one \(\lambda\in\mathbb C\) [2408.14115], [2204.04990]. In the port-Hamiltonian descriptor setting, the coefficients are factorized as \[ \begin{pmatrix}A & B\\ C & D\end{pmatrix} = \begin{pmatrix} (J-R)Q & G-P\\ (G+P)^TQ & S+N \end{pmatrix}, \] subject to \[ Q^TE=E^TQ\ge 0,\qquad \Gamma= \begin{pmatrix} J & G\\ -G^T & N \end{pmatrix} =-\Gamma^T, \qquad W= \begin{pmatrix} Q^TRQ & Q^TP\\ P^TQ & S \end{pmatrix}\ge 0, \] with \(J=-J^T\), \(R=R^T\ge 0\), \(N=-N^T\), \(S=S^T\) [2204.04990], [2408.14115]. A commonly used homogeneous specialization is \[ E\dot x=(J-R)Qx, \] and when \(Q^TE\succ 0\) this is called a dissipative-Hamiltonian descriptor system [2105.08934]. The Hamiltonian is the quadratic stored-energy functional \[ H(x)=\tfrac12 x^TE^TQx \] or, in the complex notation used in some sources, \[ H(x)=\tfrac12 x^HE^HQx, \] and \(Q^TE=E^TQ\ge 0\) ensures nonnegativity [2509.02715], [2204.04990]. Along sufficiently regular trajectories, the power balance reads \[ \frac{d}{dt}H(x(t)) = -\,x^TQ^TRQx -\,u^TP^TQx -\,x^TQ^TPu +y^Tu \le y^Tu, \] while in the simplified form it reduces to \[ \dot H = -x^TQ^TRQx + y^Tu \] or, with \(D\ge 0\), \[ \dot H(x)= -x^TRQx-u^TDu+y^Tu \] [2509.02715], [2408.14115], [1705.09081]. This identity is the central reason pH descriptor systems are naturally passive and Lyapunov stable in the zero-input case [1903.10451], [2201.06590]. The descriptor matrix \(E\) may be singular. This is not an accessory feature but part of the modeling framework: pHDAEs extend port-Hamiltonian ODEs to constrained systems and to high-index descriptor formulations while preserving the energetic interpretation [1705.09081], [2201.06590]. ## 2. Geometric structure, invariance, and equivalent representations The descriptor formulation has a geometric counterpart in terms of Dirac, Lagrange, and maximal resistive structures. In the geometric picture, a trajectory is specified by storage variables, efforts, resistive variables, and external ports satisfying \[ (-\dot x,f_R,y,e_L,e_R,u)\in\mathcal D,\qquad (x,e_L)\in\mathcal L,\qquad (f_R,e_R)\in\mathcal R, \] and the Dirac orthogonality together with the resistive inequality yields \[ \dot x^Te_L=f_R^Te_R+y^Tu\le y^Tu \] [2305.08270]. A one-to-one correspondence between this geometric formulation and the descriptor realization by pHDAEs is established by explicit constructions in both directions [2305.08270]. The pHDAE class is invariant under congruence and basis changes. For time-varying invertible \(U(t)\) and \(V(t)\), the transformed coefficients \[ \widetilde E=U^TEV,\quad \widetilde Q=U^{-1}QV,\quad \widetilde J=U^TJU,\quad \widetilde R=U^TRU,\quad \widetilde B=U^TB,\quad \widetilde P=U^TP \] again define a pHDAE, and the transformed Hamiltonian satisfies \(\widetilde H(\tilde x)=H(x)\) [1705.09081]. The same invariance principle appears in nonlinear pHDAEs under diffeomorphic coordinate changes, where \(\tilde H(\tilde x)=H(\varphi(\tilde x))\) and the transformed system preserves the pH form [1903.10451]. Power-conserving interconnection is equally intrinsic. Two subsystems \[ E_i\dot x_i=(J_i-R_i)Q_ix_i+G_iu_i,\qquad y_i=G_i^TQ_ix_i \] can be interconnected by static power-preserving relations such as \(u_1=-y_2,\ u_2=y_1\), and the aggregate system remains port-Hamiltonian with total Hamiltonian \(H_1+H_2\) [2201.06590], [1903.10451]. This invariance under interconnection is one reason the framework is described as ideal for automated network-based modeling [2201.06590]. The same structural ideas extend beyond finite-dimensional ODE/DAE settings. For PDEs, Schöberl and Siuka formulate infinite-dimensional pH systems directly from the power-balance relation, allowing the Hamiltonian density \(\mathcal H(x,\partial x)\) to depend on derivative variables. In the non-differential-operator case one recovers \[ E\dot x=(J-R)Qx+Gu,\qquad y=G^TQx, \] while in the differential-operator case one writes \[ \partial_t x=[J(\partial)-R(\partial)]\,\delta H+G(\partial)u,\qquad y=G^*\delta H, \] and the energy balance acquires boundary-port terms from integration by parts [1207.4732]. For 1D distributed systems, explicit descriptor formulations and implicit Stokes-Lagrange subspace representations are linked by bijective transformations that commute with flow-constraint projections [2402.07628]. ## 3. Regularity, index, system space, and stability Regularity and differentiation index are basic analytical notions for pH descriptor systems. The pencil \(\alpha E-\beta A\) is regular if \(\det(\alpha E-\beta A)\neq 0\) for some \((\alpha,\beta)\in\mathbb C^2\setminus\{0\}\); equivalently it has exactly \(\operatorname{rank}(E)\) finite eigenvalues [2509.02715]. In Weierstraß or strict Kronecker form, the nilpotent blocks at infinity determine the differentiation index. In port-Hamiltonian DAEs one always has \(\operatorname{ind}\le 2\), but index two is troublesome because solutions may not exist for arbitrary inputs and impulses or non-smooth phenomena may appear [2509.02715], [2403.18967]. The system space \[ V^{[E,A]}=\{x(0):x\in\mathcal B_{[E,A]}\}\subseteq\mathbb R^n \] is the natural invariant space on which descriptor trajectories evolve [2105.08934]. In quasi-Kronecker coordinates, it is obtained by deleting components associated with inconsistent or purely algebraic blocks [2105.08934]. This viewpoint is essential in stability theory because a singular descriptor pair need not define dynamics on the whole ambient space. A generalized Lyapunov inequality characterizes behavioral stability. For a homogeneous DAE \(E\dot x=Ax\), stability is equivalent to regularity of \(sE-A\), spectral inclusion \(\sigma(E,A)\subset\overline{\mathbb C_-}\) with semisimple imaginary-axis eigenvalues, and the existence of a symmetric matrix \(X\) satisfying \[ X\succ_V 0,\qquad X(EV)=EV,\qquad A^TXE+E^TXA\preceq_V 0 \] on the system space \(V=V^{[E,A]}\) [2105.08934]. If such an \(X\) exists, then on \(V\) the DAE can be rewritten as a dissipative-Hamiltonian descriptor system via \[ Q_H=XE,\qquad J_H=\tfrac12(AQ_H^\dagger-(AQ_H^\dagger)^T),\qquad R_H=-\tfrac12(AQ_H^\dagger+(AQ_H^\dagger)^T), \] with \(J_H=-J_H^T\), \(R_H\succeq 0\), and \(Q_H^TE\succ 0\) on \(V\) [2105.08934]. This gives a converse-to-structure statement: every behaviorally stable DAE admits a dH representation on its system space. For homogeneous pH DAEs \(E\dot x=(J-R)Qx\), sufficient and necessary conditions for stability are available under \(\ker Q\subseteq \ker E\). In that case, when \(Q\) is invertible, \[ (E,J,R,Q)\ \text{is stable}\iff \ker J\cap\ker R\cap(Q\,\ker E)=\{0\} \] [2105.08934]. A related geometric criterion states that, for dH systems with \(A=(J-R)Q\), regularity and stability are implied by \[ \ker E\cap\ker(Q^TJQ)\cap\ker(Q^TRQ)=\{0\} \] under the same nondegeneracy assumption [2105.08934]. A common misconception is that passivity alone implies asymptotic stability. The 2024 state-feedback results explicitly distinguish these notions: pH descriptor systems are known to be stable and passive, but they may not be asymptotically stable or strictly passive [2406.08994]. This distinction drives much of the later feedback theory. ## 4. Regularization and stabilization by feedback A central control problem for pH descriptor systems is to make a possibly non-regular or index-two closed loop regular, index at most one, asymptotically stable, and still port-Hamiltonian. For output feedback, the general proportional-derivative law is \[ u=Fy-K\dot y+v, \] which leads to the closed-loop system \[ (E+BKC)\dot x=(A+BFC)x+Bv,\qquad y=Cx \] [2509.02715]. Three special cases are pure proportional feedback, pure derivative feedback, and mixed feedback [2509.02715]. For proportional feedback, there exists \(F\) such that \((E,A+BFC)\) is regular, \(\operatorname{ind}(E,A+BFC)\le 1\), and the loop remains pH if and only if \[ \operatorname{rank} \begin{bmatrix} E & A\,S_\infty(E)\\ 0 & C\,S_\infty(E) \end{bmatrix} =n, \] where \(S_\infty(E)\) is any basis of the right nullspace of \(E\) [2509.02715]. For derivative feedback, there exists \(K\) such that \((E+BKC,A)\) is regular, \(\operatorname{ind}(E+BKC,A)\le 1\), \(\operatorname{rank}(E+BKC)\) is maximal among matrices of the form \(E+B\widehat WC\), and the closed loop is pH if and only if \[ \operatorname{rank} \begin{bmatrix} E & A\,S_\infty([E;C])\\ C & 0 \end{bmatrix} =n, \] equivalently \(\operatorname{rank}[E;C]=\operatorname{rank}[E;B]\) [2509.02715]. Under complete observability and additional rank data, mixed proportional-derivative feedback allows prescribed rank \(\operatorname{rank}(E+BKC)=r\) together with regularity, index reduction, and pH preservation [2509.02715]. The 2024 stabilization results treat the standard form \[ E\dot x=(J-R)x+Bu,\qquad y=B^Hx, \] obtained after a standard reformulation so that \(Q=I\) and \(S-N=0\) [2403.18967]. With static output feedback \[ u(t)=-K_p y(t)-K_d\dot y(t), \] the closed-loop matrices are \[ E_{\mathrm{cl}}=E+BKB^H,\qquad A_{\mathrm{cl}}=J+BF_sB^H-(R+BF_HB^H), \] where \(K=K^H\succeq 0\) and \(F=F_s-F_H\) with \(F_s=-F_s^H\), \(F_H=F_H^H\succeq 0\) [2403.18967]. Necessary and sufficient conditions are given for regularization, index reduction to \(\le 1\), and asymptotic stabilization. In particular, proportional feedback achieves regularity and index \(\le 1\) exactly when \[ \operatorname{rank}[\,E,\ (J-R)S_0(E),\ B\,]=n, \] and asymptotic stabilization further requires \[ \operatorname{rank}[\,J-R-sE,\ B\,]=n \quad\text{for every } s\in i\mathbb R \] [2403.18967]. Combined proportional and derivative feedback can enforce \(E_{\mathrm{cl}}\succ 0\), \(R_{\mathrm{cl}}\succ 0\), regularity, index \(\le 1\), and asymptotic stability under the corresponding rank conditions [2403.18967]. State-feedback results provide the analogous structure-preserving picture. For \[ E\dot x=(J-R)Qx+Bu,\qquad y=B^TQx+Du, \] the proportional state feedback \(u=Fx+v\) yields \[ A_{\rm cl}=(J-R)Q+BF,\qquad J_{\rm cl}=J+\tfrac12(BF-F^TB^T),\qquad R_{\rm cl}=R-\tfrac12(BF+F^TB^T), \] so the closed loop remains pH [2406.08994]. There exists \(F\) such that \(sE-A_{\rm cl}\) is regular, of index \(\le 1\), and all finite eigenvalues lie in \(\mathbb C_-\) if and only if \[ \operatorname{rank}[\,sE-(J-R)Q,\ B\,]=n\quad \forall s\in\mathbb C, \] and \[ \operatorname{rank}[\,E,\ (J-R)Q\,S_\infty(E),\ B\,]=n \] [2406.08994]. Strict passivity is characterized separately by \(D>0\) and an explicit positivity condition involving \(R\) and \(B D^{-1}B^T\) [2406.08994]. The 2025 paper on regularization isolates the rank conditions for output-feedback regularization without requiring full asymptotic stabilization, and emphasizes that all constructions proceed through condensed forms computed using orthogonal transformations, so the resulting algorithms are numerically reliable [2509.02715]. The 2025 output-feedback stabilization paper extends the unknown-\(Q\) case and states that, under its two rank conditions, any positive definite feedback matrix \(K>0\) achieves a regular, impulse-free, asymptotically stable, structure-preserving closed loop [2512.23203]. A plausible implication is that the pH structure makes a classically difficult descriptor feedback problem much more tractable than in the unstructured setting; that contrast is stated explicitly in the stabilization papers [2403.18967], [2512.23203]. ## 5. Passivity, positive realness, controllability, and realizations The relation among port-Hamiltonian, passive, positive-real, and KYP-based system classes is highly structured but not fully reversible without additional hypotheses. For any regular descriptor system \(\Sigma=(E,A,B,C,D)\), \[ \text{(pH)}\Longrightarrow \text{(KYP)}\Longrightarrow \text{(Pa)}, \qquad \text{(KYP)}\Longrightarrow \text{(PR)} \] [2204.04990]. Here positive realness means that \[ \mathcal T(s)=C(sE-A)^{-1}B+D \] is analytic for \(\Re s>0\) and satisfies \(\mathcal T(s)+\mathcal T(s)^H\ge 0\) there [2408.14115], [2204.04990]. The generalized KYP inequality is \[ \begin{pmatrix} -\,A^HQ-Q^HA & C^H-Q^HB\\ C-B^HQ & D+D^H \end{pmatrix}\succeq 0,\qquad Q^HE=E^HQ\succeq 0 \] [2204.04990]. The converses require hypotheses. If in the KYP inequality the solution \(Q\) is invertible, then one can reconstruct a pH representation explicitly [2204.04990]. If \((E,A,C)\) is observable, then any KYP solution \(Q\) is invertible; if \(\ker Q\subseteq\ker A\cap\ker C\), then KYP implies pH [2204.04990]. Passivity implies pH under behavioral observability and \(\operatorname{index}(E,A)\le 1\) [2204.04990]. Positive realness implies passivity under behavioral controllability [2204.04990]. The converse from positive real to port-Hamiltonian becomes exact for minimal descriptor realizations. If \((E,A,B,C,D)\) is regular, completely controllable, completely observable, and positive real, then it is already a pH descriptor system if and only if \[ D+D^T\ge 0 \] [2408.14115]. If \(D+D^T\) fails to be nonnegative, an equivalent pH realization can still be constructed for descriptor systems by a feedthrough shift: find \(W_\Delta\) such that \[ E^TW_\Delta=0,\qquad D+D^T-BW_\Delta^T-W_\Delta B^T\ge 0, \] then define \[ \widetilde C=C-W_\Delta^TA,\qquad \widetilde D=D-W_\Delta^TB. \] The new system \((E,A,B,\widetilde C,\widetilde D)\) is port-Hamiltonian and has the same transfer function [2408.14115]. In the standard case \(E=I\), by contrast, \(D\) is invariant under similarity, so such feedthrough shifting is not available [2408.14115]. Controllability and stabilizability also admit structural dimension criteria in the pH descriptor class. For systems \[ E\dot x=(J-R)Qx+Bu,\qquad y=B^\top Qx \] with \(E^\top Q=Q^\top E\ge 0\), five pH controllability concepts are relative generic in \(\Sigma^H_{\ell,n,m}\), \(\Sigma^{sdH}_{\ell,n,m}\), and \(\Sigma^{dH}_{\ell,n,m}\) precisely under explicit inequalities in \(\ell,n,m\) [2302.05156]. In particular, freely initializable and impulse-controllable are relative generic iff \(\ell\le n+m\); behavioral-controllable iff \(\ell\neq n+m\); and completely- and strongly-controllable iff \(\ell