Search for: All records

Abstract Trigonal planar M₃(O/OH) trimers are among the most important clusters in inorganic chemistry and are the foundational features of multiple high‐impact MOF platforms. Here we introduce a concept called isoreticular cluster series and demonstrate that M₃(O/OH), as the first member of a supertrimer series, can be combined with a higher hierarchical member (double‐deck trimer here) to advance isoreticular chemistry. We report here an isoreticular series of pore‐space‐partitioned MOFs called M₃M₆pacsmade from co‐assembly between M₃single‐deck trimer and M_3x2double‐deck trimer. Important factors were identified on this multi‐modular MOF platform to guide optimization of each module, which enables the phase selection of M₃M₆pacsby overcoming the formation of previously‐always‐observed same‐cluster phases. The newpacsmaterials exhibit high surface area and high uptake capacity for CO₂and small hydrocarbons, as well as selective adsorption properties relevant to separation of industrially important mixtures such as C₂H₂/CO₂and C₂H₂/C₂H₄. Furthermore, new M₃M₆pacsmaterials show electrocatalytic properties with high activity. 

Abstract Although metal–organic frameworks are coordination‐driven assemblies, the structural prediction and design using metal‐ligand interactions can be unreliable due to other competing interactions. Leveraging non‐coordination interactions to develop porous assemblies could enable new materials and applications. Here, we use a multi‐module MOF system to explore important and pervasive impact of ligand‐ligand interactions on metal‐ligand as well as ligand‐ligand co‐assembly process. It is found that ligand‐ligand interactions play critical roles on the scope or breakdown of isoreticular chemistry. With cooperative di‐ and tri‐topic ligands, a family of Ni‐MOFs has been synthesized in various structure types including partitioned MIL‐88‐acs (pacs), interruptedpacs(i‐pacs), and UMCM‐1‐muo. A new type of isoreticular chemistry on the muo platform is established between two drastically different chemical systems. The gas sorption and electrocatalytic studies were performed that reveal excellent performance such as high C₂H₂/CO₂selectivity of 21.8 and high C₂H₂uptake capacity of 114.5 cm³/g at 298 K and 1 bar. 

Centralized Training for Decentralized Execution, where agents are trained offline in a centralized fashion and execute online in a decentralized manner, has become a popular approach in Multi-Agent Reinforcement Learning (MARL). In particular, it has become popular to develop actor-critic methods that train decentralized actors with a centralized critic where the centralized critic is allowed access to global information of the entire system, including the true system state. Such centralized critics are possible given offline information and are not used for online execution. While these methods perform well in a number of domains and have become a de facto standard in MARL, using a centralized critic in this context has yet to be sufficiently analyzed theoretically or empirically. In this paper, we therefore formally analyze centralized and decentralized critic approaches, and analyze the effect of using state-based critics in partially observable environments. We derive theories contrary to the common intuition: critic centralization is not strictly beneficial, and using state values can be harmful. We further prove that, in particular, state-based critics can introduce unexpected bias and variance compared to history-based critics. Finally, we demonstrate how the theory applies in practice by comparing different forms of critics on a wide range of common multi-agent benchmarks. The experiments show practical issues such as the difficulty of representation learning with partial observability, which highlights why the theoretical problems are often overlooked in the literature. 

Synchronizing decisions across multiple agents in realistic settings is problematic since it requires agents to wait for other agents to terminate and communicate about termination reliably. Ideally, agents should learn and execute asynchronously instead. Such asynchronous methods also allow temporally extended actions that can take different amounts of time based on the situation and action executed. Unfortunately, current policy gradient methods are not applicable in asynchronous settings, as they assume that agents synchronously reason about action selection at every time step. To allow asynchronous learning and decision-making, we formulate a set of asynchronous multi-agent actor-critic methods that allow agents to directly optimize asynchronous policies in three standard training paradigms: decentralized learning, centralized learning, and centralized training for decentralized execution. Empirical results (in simulation and hardware) in a variety of realistic domains demonstrate the superiority of our approaches in large multi-agent problems and validate the effectiveness of our algorithms for learning high-quality and asynchronous solutions. 

Abstract Compared to exploratory development of new structure types, pushing the limits of isoreticular synthesis on a high‐performance MOF platform may have higher probability of achieving targeted properties. Multi‐modular MOF platforms could offer even more opportunities by expanding the scope of isoreticular chemistry. However, navigating isoreticular chemistry towards best properties on a multi‐modular platform is challenging due to multiple interconnected pathways. Here on the multi‐modular pacs (partitioned acs) platform, we demonstrate accessibility to a new regime of pore geometry using two independently adjustable modules (framework‐forming module 1 and pore‐partitioning module 2). A series of new pacs materials have been made. Benzene/cyclohexane selectivity is tuned, progressively, from 4.5 to 15.6 to 195.4 and to 482.5 by pushing the boundary of the pacs platform towards the smallest modules known so far. The exceptional stability of these materials in retaining both porosity and single crystallinity enables single‐crystal diffraction studies of different crystal forms (as‐synthesized, activated, guest‐loaded) that help reveal the mechanistic aspects of adsorption in pacs materials. 

In real-world multi-robot systems, performing high-quality, collaborative behaviors requires robots to asynchronously reason about high-level action selection at varying time durations. Macro-Action Decentralized Partially Observable Markov Decision Processes (MacDec-POMDPs) provide a general framework for asynchronous decision making under uncertainty in fully cooperative multi-agent tasks. However, multi-agent deep reinforcement learning methods have only been developed for (synchronous) primitive-action problems. This paper proposes two Deep Q-Network (DQN) based methods for learning decentralized and centralized macro-action-value functions with novel macro-action trajectory replay buffers introduced for each case. Evaluations on benchmark problems and a larger domain demonstrate the advantage of learning with macro-actions over primitive-actions and the scalability of our approaches.