Pub Date : 2021-12-14DOI: 10.1109/CDC45484.2021.9683283
A. Dautt-Silva, R. A. Callafon
A reference signal shaping problem formulated as a convex optimization problem is presented for the design of the reference signal in a closed-loop discrete-time lineartime-invariant system, with the purpose that internal control signals and system output are bounded within constraints. A causal solution endures the reference profiles changes only after a system output is required to change. The proposed solution allows us to compute a causal or noncausal reference profile, by adding a time-dependent signal constraint. Feasibility and existence of a reference profile is verified with a linear programming (LP) problem, while an optimal reference profile for the closed-loop system is obtained via a quadratic program (QP) problem. A mass-spring-damper system paired with a PID controller is the illustrative example for closed-loop reference shaping. To evaluate the proposed design, the closed-system response for both causal and noncausal reference profiles are compared.
{"title":"Reference signal shaping for closed-loop systems with causality constraints","authors":"A. Dautt-Silva, R. A. Callafon","doi":"10.1109/CDC45484.2021.9683283","DOIUrl":"https://doi.org/10.1109/CDC45484.2021.9683283","url":null,"abstract":"A reference signal shaping problem formulated as a convex optimization problem is presented for the design of the reference signal in a closed-loop discrete-time lineartime-invariant system, with the purpose that internal control signals and system output are bounded within constraints. A causal solution endures the reference profiles changes only after a system output is required to change. The proposed solution allows us to compute a causal or noncausal reference profile, by adding a time-dependent signal constraint. Feasibility and existence of a reference profile is verified with a linear programming (LP) problem, while an optimal reference profile for the closed-loop system is obtained via a quadratic program (QP) problem. A mass-spring-damper system paired with a PID controller is the illustrative example for closed-loop reference shaping. To evaluate the proposed design, the closed-system response for both causal and noncausal reference profiles are compared.","PeriodicalId":229089,"journal":{"name":"2021 60th IEEE Conference on Decision and Control (CDC)","volume":"47 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115001052","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-12-14DOI: 10.1109/CDC45484.2021.9683268
Ryan S. Johnson, Adnane Saoud, R. Sanfelice
We consider the problem of estimating a constant or piecewise constant vector of unknown parameters for a linear dynamical system. Using a hybrid systems framework, a hybrid algorithm that achieves finite-time convergence of the parameter estimate to the true value is proposed. Sufficient conditions that guarantee convergence of the parameter estimate are provided. Robustness of the proposed algorithm with respect to measurements noise is analyzed, and examples are provided showing the merits of the proposed approach.
{"title":"Robust Finite-Time Parameter Estimation for Linear Dynamical Systems","authors":"Ryan S. Johnson, Adnane Saoud, R. Sanfelice","doi":"10.1109/CDC45484.2021.9683268","DOIUrl":"https://doi.org/10.1109/CDC45484.2021.9683268","url":null,"abstract":"We consider the problem of estimating a constant or piecewise constant vector of unknown parameters for a linear dynamical system. Using a hybrid systems framework, a hybrid algorithm that achieves finite-time convergence of the parameter estimate to the true value is proposed. Sufficient conditions that guarantee convergence of the parameter estimate are provided. Robustness of the proposed algorithm with respect to measurements noise is analyzed, and examples are provided showing the merits of the proposed approach.","PeriodicalId":229089,"journal":{"name":"2021 60th IEEE Conference on Decision and Control (CDC)","volume":"159 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115170336","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-12-14DOI: 10.1109/CDC45484.2021.9683366
Hamideh Saadabadi, H. Werner
This paper proposes a novel approach to event-triggered formation control for homogeneous, non-holonomic multi-agent systems with undirected interaction topology, where the non-holonomic vehicle dynamics are represented by poly-topic linear-parameter-varying (LPV) models. The proposed event-triggered strategy is able to reduce the communication cost by transmitting information only when needed. To maintain a formation, each agent is equipped with an inner state-feedback loop that is time-triggered, while an outer position loop is closed by each agent individually through the communication network whenever a local trigger condition is satisfied. The control strategy can be implemented in a distributed manner; the trigger condition is based only on locally available information. The proposed method allows to simultaneously design a controller and a trigger level that guarantee stability and a bound on the overall ℓ2 performance of the network. The synthesis problem is formulated as an LMI problem. Under the additional assumption that the agents are homogeneously scheduled, the synthesis problem can be decomposed to reduce its complexity to the size of a single agent, regardless of the number of agents, without degrading the performance. The effectiveness of the results is illustrated in a simulation scenario with non-holonomic agents modeled as dynamic unicycles.
{"title":"Event-Triggered ℓ2-Optimal Formation Control for Agents Modeled as LPV Systems","authors":"Hamideh Saadabadi, H. Werner","doi":"10.1109/CDC45484.2021.9683366","DOIUrl":"https://doi.org/10.1109/CDC45484.2021.9683366","url":null,"abstract":"This paper proposes a novel approach to event-triggered formation control for homogeneous, non-holonomic multi-agent systems with undirected interaction topology, where the non-holonomic vehicle dynamics are represented by poly-topic linear-parameter-varying (LPV) models. The proposed event-triggered strategy is able to reduce the communication cost by transmitting information only when needed. To maintain a formation, each agent is equipped with an inner state-feedback loop that is time-triggered, while an outer position loop is closed by each agent individually through the communication network whenever a local trigger condition is satisfied. The control strategy can be implemented in a distributed manner; the trigger condition is based only on locally available information. The proposed method allows to simultaneously design a controller and a trigger level that guarantee stability and a bound on the overall ℓ2 performance of the network. The synthesis problem is formulated as an LMI problem. Under the additional assumption that the agents are homogeneously scheduled, the synthesis problem can be decomposed to reduce its complexity to the size of a single agent, regardless of the number of agents, without degrading the performance. The effectiveness of the results is illustrated in a simulation scenario with non-holonomic agents modeled as dynamic unicycles.","PeriodicalId":229089,"journal":{"name":"2021 60th IEEE Conference on Decision and Control (CDC)","volume":"2630 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115228837","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-12-14DOI: 10.1109/CDC45484.2021.9683585
A. Kasis, S. Timotheou, M. Polycarpou
A major transition in modern power systems is the replacement of conventional generation units with renewable sources of energy. The latter results in lower rotational inertia which compromises the stability of the power system, as testified by the growing number of frequency incidents. To resolve this problem, numerous studies have proposed the use of virtual inertia to improve the stability properties of the power grid. In this study, we consider how inertia variations, resulting from the application of control action associated with virtual inertia and fluctuations in renewable generation, may affect the stability properties of the power network within the primary frequency control timeframe. We consider the interaction between the frequency dynamics and a broad class of non-linear power supply dynamics at the presence of time-varying virtual inertia and provide suitable conditions such that stability is guaranteed. In particular, we impose two conditions; a decentralized passivity-related condition on the power supply dynamics and a condition that associates the maximum rate of growth of virtual inertia with the local power supply dynamics. The presented conditions are locally verifiable and applicable to arbitrary network configurations. In addition, in case of linear power supply dynamics, they can be efficiently verified by solving suitable linear matrix inequalities. Our analytic results are validated with simulations on the Northeast Power Coordinating Council (NPCC) 140-bus system, where we demonstrate how varying virtual inertia may induce large frequency oscillations and show that the application of the proposed conditions yields a stable response.
{"title":"Stability of power networks with time-varying inertia","authors":"A. Kasis, S. Timotheou, M. Polycarpou","doi":"10.1109/CDC45484.2021.9683585","DOIUrl":"https://doi.org/10.1109/CDC45484.2021.9683585","url":null,"abstract":"A major transition in modern power systems is the replacement of conventional generation units with renewable sources of energy. The latter results in lower rotational inertia which compromises the stability of the power system, as testified by the growing number of frequency incidents. To resolve this problem, numerous studies have proposed the use of virtual inertia to improve the stability properties of the power grid. In this study, we consider how inertia variations, resulting from the application of control action associated with virtual inertia and fluctuations in renewable generation, may affect the stability properties of the power network within the primary frequency control timeframe. We consider the interaction between the frequency dynamics and a broad class of non-linear power supply dynamics at the presence of time-varying virtual inertia and provide suitable conditions such that stability is guaranteed. In particular, we impose two conditions; a decentralized passivity-related condition on the power supply dynamics and a condition that associates the maximum rate of growth of virtual inertia with the local power supply dynamics. The presented conditions are locally verifiable and applicable to arbitrary network configurations. In addition, in case of linear power supply dynamics, they can be efficiently verified by solving suitable linear matrix inequalities. Our analytic results are validated with simulations on the Northeast Power Coordinating Council (NPCC) 140-bus system, where we demonstrate how varying virtual inertia may induce large frequency oscillations and show that the application of the proposed conditions yields a stable response.","PeriodicalId":229089,"journal":{"name":"2021 60th IEEE Conference on Decision and Control (CDC)","volume":"79 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115423850","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-12-14DOI: 10.1109/CDC45484.2021.9683077
Sayan Samanta, N. Chaudhuri, C. Lagoa
A decentralized supplementary control for quantifiable primary frequency support from renewable generation interfaced with class-A grid-forming converters (GFCs) under dc-side current limitation is proposed. GFCs regulated by droop, dispatchable virtual oscillator control (dVOC), and virtual synchronous machine (VSM) strategies belong to this class. The approach requires communication of frequency measurements of GFCs from adjacent buses. The proposed controller guarantees asymptotic stability of power grids with generic configurations that include multiple synchronous generators (SGs) and GFCs under dc power flow approximation and a mild assumption on center-of-inertia based frequency dynamics model. Simulations on a simplified model of a 4-bus system and a detailed phasor model of IEEE 16-machine system show the effectiveness of the proposed approach.
{"title":"Quantifiable Frequency Support from Grid-Forming Converters with DC-side Current Limits in Grids with Synchronous Generators","authors":"Sayan Samanta, N. Chaudhuri, C. Lagoa","doi":"10.1109/CDC45484.2021.9683077","DOIUrl":"https://doi.org/10.1109/CDC45484.2021.9683077","url":null,"abstract":"A decentralized supplementary control for quantifiable primary frequency support from renewable generation interfaced with class-A grid-forming converters (GFCs) under dc-side current limitation is proposed. GFCs regulated by droop, dispatchable virtual oscillator control (dVOC), and virtual synchronous machine (VSM) strategies belong to this class. The approach requires communication of frequency measurements of GFCs from adjacent buses. The proposed controller guarantees asymptotic stability of power grids with generic configurations that include multiple synchronous generators (SGs) and GFCs under dc power flow approximation and a mild assumption on center-of-inertia based frequency dynamics model. Simulations on a simplified model of a 4-bus system and a detailed phasor model of IEEE 16-machine system show the effectiveness of the proposed approach.","PeriodicalId":229089,"journal":{"name":"2021 60th IEEE Conference on Decision and Control (CDC)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115429640","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-12-14DOI: 10.1109/CDC45484.2021.9683660
Elvina Gindullina, M. Zorzi, A. Bertoldo, A. Chiuso
One of the main outstanding issues in the neuroscience is estimation of effective connectivity in brain networks, which models the causal interactions among neuronal populations. Estimation of effective connectivity embraces two types of the challenges, such as estimation accuracy and computational complexity. In this paper, we consider resting-state (rs) fMRI data serving as an input for a stochastic linear DCM model. The model parameters are estimated through an EM (expectation maximization) iterative procedure. In this work, we propose the alternative scheme for the hyperparameters estimation aiming in reduction of computational burden of the original EM-algorithm. The simulation results demonstrate the viability of the proposed block-reweighting scheme and represents a promising research direction to be further investigated.
{"title":"Estimating Effective Connectivity using Brain Partitioning","authors":"Elvina Gindullina, M. Zorzi, A. Bertoldo, A. Chiuso","doi":"10.1109/CDC45484.2021.9683660","DOIUrl":"https://doi.org/10.1109/CDC45484.2021.9683660","url":null,"abstract":"One of the main outstanding issues in the neuroscience is estimation of effective connectivity in brain networks, which models the causal interactions among neuronal populations. Estimation of effective connectivity embraces two types of the challenges, such as estimation accuracy and computational complexity. In this paper, we consider resting-state (rs) fMRI data serving as an input for a stochastic linear DCM model. The model parameters are estimated through an EM (expectation maximization) iterative procedure. In this work, we propose the alternative scheme for the hyperparameters estimation aiming in reduction of computational burden of the original EM-algorithm. The simulation results demonstrate the viability of the proposed block-reweighting scheme and represents a promising research direction to be further investigated.","PeriodicalId":229089,"journal":{"name":"2021 60th IEEE Conference on Decision and Control (CDC)","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115451847","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-12-14DOI: 10.1109/CDC45484.2021.9682890
Vyoma Singh, Birupaksha Pal, Tushar Jain
To ensure maximum efficiency, low emissions, and lower fuel consumption in the vehicles, advanced control schemes are required. Due to the engine operation, the sensors cannot be installed to measure all the variables that are needed for an effective control. While addressing this issue, a new adaptive Unscented Kalman filter (UKF) algorithm is proposed in this paper to estimate the intake manifold pressure, engine speed, and fuel flow rate. New adaptive laws are designed to update the process noise and measurement noise covariance matrices within the constrained augmented state-based UKF (CASUKF). Another contribution lies in the new combination of the novel adaptive laws, and CASUKF, unlike other variants of the UKF that either adapt the process noise and measurement noise covariance matrices on the standard UKF or implement CASUKF with constant values of the process noise and measurement noise matrices. Simulation results are provided for the nonlinear mean value spark-ignition engine model, and the effectiveness of the algorithm is also compared with other variants of the UKF.
{"title":"State Estimation for Spark-Ignition Engines Using New Noise Adaptive Laws In Unscented Kalman Filter","authors":"Vyoma Singh, Birupaksha Pal, Tushar Jain","doi":"10.1109/CDC45484.2021.9682890","DOIUrl":"https://doi.org/10.1109/CDC45484.2021.9682890","url":null,"abstract":"To ensure maximum efficiency, low emissions, and lower fuel consumption in the vehicles, advanced control schemes are required. Due to the engine operation, the sensors cannot be installed to measure all the variables that are needed for an effective control. While addressing this issue, a new adaptive Unscented Kalman filter (UKF) algorithm is proposed in this paper to estimate the intake manifold pressure, engine speed, and fuel flow rate. New adaptive laws are designed to update the process noise and measurement noise covariance matrices within the constrained augmented state-based UKF (CASUKF). Another contribution lies in the new combination of the novel adaptive laws, and CASUKF, unlike other variants of the UKF that either adapt the process noise and measurement noise covariance matrices on the standard UKF or implement CASUKF with constant values of the process noise and measurement noise matrices. Simulation results are provided for the nonlinear mean value spark-ignition engine model, and the effectiveness of the algorithm is also compared with other variants of the UKF.","PeriodicalId":229089,"journal":{"name":"2021 60th IEEE Conference on Decision and Control (CDC)","volume":"71 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115636253","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-12-14DOI: 10.1109/CDC45484.2021.9683503
Jean-Sébastien Brouillon, E. Fabbiani, P. Nahata, F. Dörfler, G. Ferrari-Trecate
The increasing integration of intermittent renewable generation, especially at the distribution level, necessitates advanced planning and optimisation methodologies contingent on the knowledge of the admittance matrix, capturing the topology and line parameters of an electric network. However, a reliable estimate of the admittance matrix may either be missing or quickly become obsolete for temporally varying grids. In this work, we propose a data-driven identification method utilising voltage and current measurements collected from micro-PMUs. More precisely, we first present a maximum likelihood approach and then move towards a Bayesian framework, leveraging the principles of maximum a posteriori estimation. In contrast with most existing contributions, our approach not only factors in measurement noise on both voltage and current data, but is also capable of exploiting available a priori information such as sparsity patterns and known line admittances. Simulations conducted on benchmark cases demonstrate that, compared to other algorithms, our method can achieve greater accuracy.
{"title":"Bayesian Methods for the Identification of Distribution Networks","authors":"Jean-Sébastien Brouillon, E. Fabbiani, P. Nahata, F. Dörfler, G. Ferrari-Trecate","doi":"10.1109/CDC45484.2021.9683503","DOIUrl":"https://doi.org/10.1109/CDC45484.2021.9683503","url":null,"abstract":"The increasing integration of intermittent renewable generation, especially at the distribution level, necessitates advanced planning and optimisation methodologies contingent on the knowledge of the admittance matrix, capturing the topology and line parameters of an electric network. However, a reliable estimate of the admittance matrix may either be missing or quickly become obsolete for temporally varying grids. In this work, we propose a data-driven identification method utilising voltage and current measurements collected from micro-PMUs. More precisely, we first present a maximum likelihood approach and then move towards a Bayesian framework, leveraging the principles of maximum a posteriori estimation. In contrast with most existing contributions, our approach not only factors in measurement noise on both voltage and current data, but is also capable of exploiting available a priori information such as sparsity patterns and known line admittances. Simulations conducted on benchmark cases demonstrate that, compared to other algorithms, our method can achieve greater accuracy.","PeriodicalId":229089,"journal":{"name":"2021 60th IEEE Conference on Decision and Control (CDC)","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115743701","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-12-14DOI: 10.1109/CDC45484.2021.9683615
Shantanu Singh, G. Weiss, M. Tucsnak
We investigate abstract nonlinear infinite dimensional systems of the form: $dot x(t) in Ax(t) - {mathcal{M}}(x(t)) + Bu(t)$ . These are obtained by subtracting a nonlinear maximal monotone (possibly multi-valued) operator ${mathcal{M}}$ from the semigroup generator A of a linear system. While the linear system may have un-bounded linear damping (for instance, boundary damping), the operator ${mathcal{M}}$ is "bounded" in the sense that it is defined on the whole state space. We show that under some assumptions, such nonlinear infinite dimensional systems have unique classical and generalized solutions. Moreover, these solutions are Lipschitz continuous on any finite time interval and right differentiable. Our approach uses the theory of maximal monotone operators and the Crandall-Pazy theorem about nonlinear contraction semigroups, which we apply to a Lax-Phillips type nonlinear semigroup that represents the entire system, with states and input signals. We illustrate the theory with Maxwell’s equations in a bounded domain with a nonlinear conductor.
我们研究了抽象的非线性无限维系统的形式:$dot x(t) in Ax(t) - {mathcal{M}}(x(t)) + Bu(t)$。这些是通过从线性系统的半群生成器a中减去非线性极大单调(可能是多值)算子${mathcal{M}}$得到的。虽然线性系统可能具有无界线性阻尼(例如，边界阻尼)，但算子${mathcal{M}}$是“有界的”，因为它是在整个状态空间上定义的。在一定的假设条件下，证明了这类非线性无穷维系统具有唯一的经典解和广义解。而且，这些解在任意有限时间区间上是Lipschitz连续的，并且是右可微的。我们的方法使用极大单调算子理论和关于非线性收缩半群的Crandall-Pazy定理，我们将其应用于具有状态和输入信号的整个系统的Lax-Phillips型非线性半群。我们用具有非线性导体的有界区域中的麦克斯韦方程组来说明这一理论。
{"title":"Abstract nonlinear control systems","authors":"Shantanu Singh, G. Weiss, M. Tucsnak","doi":"10.1109/CDC45484.2021.9683615","DOIUrl":"https://doi.org/10.1109/CDC45484.2021.9683615","url":null,"abstract":"We investigate abstract nonlinear infinite dimensional systems of the form: $dot x(t) in Ax(t) - {mathcal{M}}(x(t)) + Bu(t)$ . These are obtained by subtracting a nonlinear maximal monotone (possibly multi-valued) operator ${mathcal{M}}$ from the semigroup generator A of a linear system. While the linear system may have un-bounded linear damping (for instance, boundary damping), the operator ${mathcal{M}}$ is \"bounded\" in the sense that it is defined on the whole state space. We show that under some assumptions, such nonlinear infinite dimensional systems have unique classical and generalized solutions. Moreover, these solutions are Lipschitz continuous on any finite time interval and right differentiable. Our approach uses the theory of maximal monotone operators and the Crandall-Pazy theorem about nonlinear contraction semigroups, which we apply to a Lax-Phillips type nonlinear semigroup that represents the entire system, with states and input signals. We illustrate the theory with Maxwell’s equations in a bounded domain with a nonlinear conductor.","PeriodicalId":229089,"journal":{"name":"2021 60th IEEE Conference on Decision and Control (CDC)","volume":"104 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116777999","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2021-12-14DOI: 10.1109/CDC45484.2021.9683142
Yuhao Ding, J. Lavaei
We consider structured online convex optimization (OCO) with bandit feedback, where either the loss function is smooth or the constraint set is strongly convex. Projection-free methods are among the most popular and computationally efficient algorithms for solving this problem, mainly due to their ability to handle convex constraints appearing in machine learning for which computing projections is often impractical in high-dimensional settings. Despite the improved regret bound results for the full-information setting where the gradients of the functions are readily available, it remains unclear whether simple projection-free zero-order algorithms become more efficient for structured OCO problems in the case when multiple function values can be sampled at each time instance. In this paper, we develop some simple projection-free algorithms and prove that they indeed achieve the same improved regret bounds as the full-information case under various additional problem structures. This implies that leveraging the structural properties of the problem compensates for the lack of access to the gradients. Experiments on the online matrix completion reveal several attractive advantages of the proposed algorithms, including their simplicity, easy implementation, and effectiveness, as they outperform other competing algorithms.
{"title":"Structured Projection-free Online Convex Optimization with Multi-point Bandit Feedback","authors":"Yuhao Ding, J. Lavaei","doi":"10.1109/CDC45484.2021.9683142","DOIUrl":"https://doi.org/10.1109/CDC45484.2021.9683142","url":null,"abstract":"We consider structured online convex optimization (OCO) with bandit feedback, where either the loss function is smooth or the constraint set is strongly convex. Projection-free methods are among the most popular and computationally efficient algorithms for solving this problem, mainly due to their ability to handle convex constraints appearing in machine learning for which computing projections is often impractical in high-dimensional settings. Despite the improved regret bound results for the full-information setting where the gradients of the functions are readily available, it remains unclear whether simple projection-free zero-order algorithms become more efficient for structured OCO problems in the case when multiple function values can be sampled at each time instance. In this paper, we develop some simple projection-free algorithms and prove that they indeed achieve the same improved regret bounds as the full-information case under various additional problem structures. This implies that leveraging the structural properties of the problem compensates for the lack of access to the gradients. Experiments on the online matrix completion reveal several attractive advantages of the proposed algorithms, including their simplicity, easy implementation, and effectiveness, as they outperform other competing algorithms.","PeriodicalId":229089,"journal":{"name":"2021 60th IEEE Conference on Decision and Control (CDC)","volume":"229 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125805635","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：