7,063 research outputs found
Strategy complexity of lim sup and lim inf payoff objectives in countable MDPs
No full textWe study countably infinite Markov decision processes (MDPs) with real-valued tran-
sition rewards. A strategy is a function which decides how plays proceed within the
MDP. Every strategy induces a set of infinite runs in the MDP and each infinite run
induces the following sequences of payoffs:
1. Point payoff (the sequence of directly seen transition rewards),
2. Mean payoff (the sequence of the sums of all rewards so far, divided by the number
of steps), and
3. Total payoff (the sequence of the sums of all rewards so far).
For each payoff type, the threshold objective is to maximise the probability that the
lim sup/lim inf is non-negative. We are interested in the strategy complexity of the
above objectives, i.e. the amount of memory and/or randomisation that a strategy
needs access to in order to play well (optimally resp. ε-optimally). Our results seek
not only to decide whether an objective requires finite or infinite memory, but in the
case of infinite memory, what kind of infinite memory is necessary and sufficient. For
example, a step counter which acts as a clock, or a reward counter which sums up the
seen rewards may be sufficient.
We compare the lim sup/lim inf point payoff objectives to the Büchi/co-Büchi ob-
jectives which, given a set of states or transitions, seek to maximise the probability
that this set is visited infinitely/finitely often. Convergence effects are what differen-
tiate lim sup/lim inf point payoff objectives from Büchi/co-Büchi. For example, the
sequence −1/2, −1/3, −1/4 . . . does satisfy lim sup ≥ 0 and lim inf ≥ 0 despite all of
the rewards being negative. It is in dealing with these effects which we make our main
technical contributions. We establish a complete picture of the strategy complexity for
both the lim sup and lim inf point payoff objectives. In particular we show that opti-
mal lim sup requires either randomisation or access to a step counter and that lim inf
of point payoff requires a step counter (but not more) when the underlying MDP is
infinitely branching.
We also comprehensively pin down the strategy complexity for the lim inf total
and mean payoff objectives. This result requires a novel counterexample involving
unboundedly growing rewards as well as finely tuned transition probabilities which
force the player to use memory in order to mimic what occurred in past random events.
This allows us to show that both of these objectives require the use of both a step
counter as well as a reward counter.
We apply our results to solve two open problems from Sudderth [35] about the
strategy complexity of optimal strategies for the expected lim sup/lim inf point pay-
off. We achieve this by reducing each objective to its respective optimal threshold
lim sup/lim inf point payoff counterpart. Thus we are able to conclude that they share
the same optimal strategy complexity
Yves-Heng Lim
No full textYves-Heng Lim est enseignant-chercheur au Département d’Etudes de Sécurité et de Criminologie de l’Université Macquarie, Sydney. Il est l’auteur de China’s Naval Power: An Offensive Realist Approach (Ashgate, 2014). Yves-Heng Lim is a lecturer at the Department of Security Studies and Criminology, Macquarie University. He is the author of China’s Naval Power: An Offensive Realist Approach (Ashgate, 2014)
Surgery and Advances in Competing Technology
No full textEric Lim of the Royal Brompton Hospital in London, UK, discusses the future of general thoracic surgery in the context of a rapid rise of competing radiotherapy technology. He demonstrates the rapid growth of radiotherapy but also presents ways that general thoracic surgeons can have an advantage over this competing specialty.This presentation was originally given during the SCTS Ionescu University program at the 2017 Annual Meeting of the Society for Cardiothoracic Surgery in Great Britain and Ireland. This content is published with the permission of SCTS. Please click here for more information on SCTS educational programs.</div
Impact of start-up and shut-down losses on the economic benefit of an integrated hybrid solar cavity receiver and combustor
No full textAbstract not availableJin Han Lim, Eric Hu, Graham J. Natha
Analytical assessment of a novel hybrid solar tubular receiver and combustor
No full textAbstract not availableJin Han Lim, Graham J. Nathan, Eric Hu, Bassam B. Dall
Professional attachment report [with] Chio Lim & Associates.
No full textThis report serves as a summary of the professional attachment. Besides touching on author experiences working with Chio Lim & Associates (CLA), it wil also touch on other issues before, during and after the program
Supplemental Material - Unveiling the formation of conspiracy theory on social media: A discourse analysis
No full textSupplemental Material for Unveiling the formation of conspiracy theory on social media: A discourse analysis by Boying Li, David Ji, Mengyao Fu, Chee-Wee Tan, Alain Chong and Eric TK Lim in Journal of Information Technology</p
RETRACTED: Nuclear Translocation of LIM Kinase Mediates Rho-Rho Kinase Regulation of Cyclin D1 Expression
No full textAbstractThis article has been retracted at the request of the authors.Reason: We have recently reviewed the primary source data for our paper entitled “Nuclear Translocation of LIM Kinase Mediates Rho-Rho Kinase Regulation of Cyclin D1 Expression” [Dev. Cell, 5: 273–284 (2003)]. We have found that portions of the figures assembled by the lead author are not fully supported by the primary data. We are therefore retracting this paper. We express our deep regret to the scientific community
Lim-inf convergence and its compactness
No full textAbstract- We describe the Mizar formalization of the proof of compactness of lim-inf convergence given in [W33] according to [CCL]. Lim-inf convergence formalized in [W28] is a Moore-Smith convergence investigated in [Y6] and involves the concept of nets. The proof is based on the equivalence of two approaches to convergence in topological spaces: filter convergence and Moore-Smith (net) convergence. The equivalence is worked out in [Y19] and different characterizations of compactness are also given there. These efforts are a continuation of the international project of formalizing the theory of continuous lattices headed by the first author
Senior Recital: Andrew Lim
No full textSenior recital featuring Andrew Lim.https://digitalcommons.kennesaw.edu/musicprograms/2383/thumbnail.jp
- …
