-
Evolutionary Algorithms for Solving Unconstrained, Constrained and Multi-objective Noisy Combinatorial Optimisation Problems
Authors:
Aishwaryaprajna,
Jonathan E. Rowe
Abstract:
We present an empirical study of a range of evolutionary algorithms applied to various noisy combinatorial optimisation problems. There are three sets of experiments. The first looks at several toy problems, such as OneMax and other linear problems. We find that UMDA and the Paired-Crossover Evolutionary Algorithm (PCEA) are the only ones able to cope robustly with noise, within a reasonable fixed…
▽ More
We present an empirical study of a range of evolutionary algorithms applied to various noisy combinatorial optimisation problems. There are three sets of experiments. The first looks at several toy problems, such as OneMax and other linear problems. We find that UMDA and the Paired-Crossover Evolutionary Algorithm (PCEA) are the only ones able to cope robustly with noise, within a reasonable fixed time budget. In the second stage, UMDA and PCEA are then tested on more complex noisy problems: SubsetSum, Knapsack and SetCover. Both perform well under increasing levels of noise, with UMDA being the better of the two. In the third stage, we consider two noisy multi-objective problems (CountingOnesCountingZeros and a multi-objective formulation of SetCover). We compare several adaptations of UMDA for multi-objective problems with the Simple Evolutionary Multi-objective Optimiser (SEMO) and NSGA-II. We conclude that UMDA, and its variants, can be highly effective on a variety of noisy combinatorial optimisation, outperforming many other evolutionary algorithms.
△ Less
Submitted 5 October, 2021;
originally announced October 2021.
-
Ordinal Optimisation and the Offline Multiple Noisy Secretary Problem
Authors:
Robert Chin,
Jonathan E. Rowe,
Iman Shames,
Chris Manzie,
Dragan Nešić
Abstract:
We study the success probability for a variant of the secretary problem, with noisy observations and multiple offline selection. Our formulation emulates, and is motivated by, problems involving noisy selection arising in the disciplines of stochastic simulation and simulation-based optimisation. In addition, we employ the philosophy of ordinal optimisation - involving an ordinal selection rule, a…
▽ More
We study the success probability for a variant of the secretary problem, with noisy observations and multiple offline selection. Our formulation emulates, and is motivated by, problems involving noisy selection arising in the disciplines of stochastic simulation and simulation-based optimisation. In addition, we employ the philosophy of ordinal optimisation - involving an ordinal selection rule, and a percentile notion of goal softening for the success probability. As a result, it is shown that the success probability only depends on the underlying copula of the problem. Other general properties for the success probability are also presented. Specialising to the case of Gaussian copulas, we also derive an analytic lower bound for the success probability, which may then be inverted to find sufficiently large sample sizes that guarantee a high success probability arbitrarily close to one.
△ Less
Submitted 10 June, 2021; v1 submitted 2 June, 2021;
originally announced June 2021.
-
A Sequential Learning Algorithm for Probabilistically Robust Controller Tuning
Authors:
Robert Chin,
Chris Manzie,
Iman Shames,
Dragan Nešić,
Jonathan E. Rowe
Abstract:
We introduce a sequential learning algorithm to address a robust controller tuning problem, which in effect, finds (with high probability) a candidate solution satisfying the internal performance constraint to a chance-constrained program which has black-box functions. The algorithm leverages ideas from the areas of randomised algorithms and ordinal optimisation, and also draws comparisons with th…
▽ More
We introduce a sequential learning algorithm to address a robust controller tuning problem, which in effect, finds (with high probability) a candidate solution satisfying the internal performance constraint to a chance-constrained program which has black-box functions. The algorithm leverages ideas from the areas of randomised algorithms and ordinal optimisation, and also draws comparisons with the scenario approach; these have all been previously applied to finding approximate solutions for difficult design problems. By exploiting statistical correlations through black-box sampling, we formally prove that our algorithm yields a controller meeting the prescribed probabilistic performance specification. Additionally, we characterise the computational requirement of the algorithm with a probabilistic lower bound on the algorithm's stop** time. To validate our work, the algorithm is then demonstrated for tuning model predictive controllers on a diesel engine air-path across a fleet of vehicles. The algorithm successfully tuned a single controller to meet a desired tracking error performance, even in the presence of the plant uncertainty inherent across the fleet. Moreover, the algorithm was shown to exhibit a sample complexity comparable to the scenario approach.
△ Less
Submitted 17 October, 2021; v1 submitted 18 February, 2021;
originally announced February 2021.
-
Active Learning for Linear Parameter-Varying System Identification
Authors:
Robert Chin,
Alejandro I. Maass,
Nalika Ulapane,
Chris Manzie,
Iman Shames,
Dragan Nešić,
Jonathan E. Rowe,
Hayato Nakada
Abstract:
Active learning is proposed for selection of the next operating points in the design of experiments, for identifying linear parameter-varying systems. We extend existing approaches found in literature to multiple-input multiple-output systems with a multivariate scheduling parameter. Our approach is based on exploiting the probabilistic features of Gaussian process regression to quantify the overa…
▽ More
Active learning is proposed for selection of the next operating points in the design of experiments, for identifying linear parameter-varying systems. We extend existing approaches found in literature to multiple-input multiple-output systems with a multivariate scheduling parameter. Our approach is based on exploiting the probabilistic features of Gaussian process regression to quantify the overall model uncertainty across locally identified models. This results in a flexible framework which accommodates for various techniques to be applied for estimation of local linear models and their corresponding uncertainty. We perform active learning in application to the identification of a diesel engine air-path model, and demonstrate that measures of model uncertainty can be successfully reduced using the proposed framework.
△ Less
Submitted 2 May, 2020;
originally announced May 2020.
-
Ordinal Optimisation for the Gaussian Copula Model
Authors:
Robert Chin,
Jonathan E. Rowe,
Iman Shames,
Chris Manzie,
Dragan Nešić
Abstract:
We present results on the estimation and evaluation of success probabilities for ordinal optimisation over uncountable sets (such as subsets of $\mathbb{R}^{d}$). Our formulation invokes an assumption of a Gaussian copula model, and we show that the success probability can be equivalently computed by assuming a special case of additive noise. We formally prove a lower bound on the success probabil…
▽ More
We present results on the estimation and evaluation of success probabilities for ordinal optimisation over uncountable sets (such as subsets of $\mathbb{R}^{d}$). Our formulation invokes an assumption of a Gaussian copula model, and we show that the success probability can be equivalently computed by assuming a special case of additive noise. We formally prove a lower bound on the success probability under the Gaussian copula model, and numerical experiments demonstrate that the lower bound yields a reasonable approximation to the actual success probability. Lastly, we showcase the utility of our results by guaranteeing high success probabilities with ordinal optimisation.
△ Less
Submitted 13 May, 2021; v1 submitted 5 November, 2019;
originally announced November 2019.
-
A Polynomial Time Bounded-error Quantum Algorithm for Boolean Satisfiability
Authors:
Ahmed Younes,
Jonathan E. Rowe
Abstract:
The aim of the paper is to answer a long-standing open problem on the relationship between NP and BQP. The paper shows that BQP contains NP by proposing a BQP quantum algorithm for the MAX-E3-SAT problem which is a fundamental NP-hard problem. Given an E3-CNF Boolean formula, the aim of the MAX-E3-SAT problem is to find the variable assignment that maximizes the number of satisfied clauses. The pr…
▽ More
The aim of the paper is to answer a long-standing open problem on the relationship between NP and BQP. The paper shows that BQP contains NP by proposing a BQP quantum algorithm for the MAX-E3-SAT problem which is a fundamental NP-hard problem. Given an E3-CNF Boolean formula, the aim of the MAX-E3-SAT problem is to find the variable assignment that maximizes the number of satisfied clauses. The proposed algorithm runs in $O(m^2)$ for an E3-CNF Boolean formula with $m$ clauses and in the worst case runs in $O(n^6)$ for an E3-CNF Boolean formula with $n$ inputs. The proposed algorithm maximizes the set of satisfied clauses using a novel iterative partial negation and partial measurement technique. The algorithm is shown to achieve an arbitrary high probability of success of $1-ε$ for small $ε>0$ using a polynomial resources. In addition to solving the MAX-E3-SAT problem, the proposed algorithm can also be used to decide if an E3-CNF Boolean formula is satisfiable or not, which is an NP-complete problem, based on the maximum number of satisfied clauses.
△ Less
Submitted 26 July, 2015; v1 submitted 16 July, 2015;
originally announced July 2015.
-
Conducting Atomic Force Microscopy Studies of Nanoscale Cobalt Silicide Schottky Barriers on Si(111) and Si(100)
Authors:
Joseph L. Tedesco,
J. E. Rowe,
Robert J. Nemanich
Abstract:
Cobalt silicide (CoSi2) islands have been formed by the deposition of thin films (~0.1 to 0.3 nm) of cobalt on clean Si(111) and Si(100) substrates in ultrahigh vacuum (UHV) followed by annealing to ~880 degrees C. Conducting atomic force microscopy has been performed on these islands to characterize and measure their current-voltage (I-V) characteristics. Current-voltage curves were analyzed usin…
▽ More
Cobalt silicide (CoSi2) islands have been formed by the deposition of thin films (~0.1 to 0.3 nm) of cobalt on clean Si(111) and Si(100) substrates in ultrahigh vacuum (UHV) followed by annealing to ~880 degrees C. Conducting atomic force microscopy has been performed on these islands to characterize and measure their current-voltage (I-V) characteristics. Current-voltage curves were analyzed using thermionic emission theory to obtain the Schottky barrier heights and ideality factors between the silicide islands and the silicon substrates. Current-voltage measurements were performed ex situ for samples ("passivated surfaces") where the silicon surface surrounding the islands was passivated with a native oxide. Other samples ("clean surfaces") remained in UHV while I-V curves were recorded. By comparing barrier heights and ideality factors for islands on both surfaces, the effects of the non-passivated surfaces on conduction were studied. The clean surface barrier heights were found to be ~0.2 to 0.3 eV below barrier heights measured from similar islands on passivated surfaces. The reduced Schottky barrier is mainly attributed to Fermi level pinning by non-passivated surface states of the clean silicon surface. However, the measured barrier heights of the clean surface islands are equivalent on both Si(111) and Si(100), suggesting that the non-passivated surfaces are influenced by cobalt impurities. Furthermore, the barrier heights of islands on the clean surfaces are lower than due to Fermi level pinning alone and the additional barrier height lowering is primarily attributed to spreading resistance effects. Schottky barrier inhomogeneity may also cause additional barrier height lowering and non-ideality in the contacts. The clean surface barrier heights decreased and the ideality factors increased with decreasing temperature, which is attributed to the temperature variation of the Fermi level.
△ Less
Submitted 28 July, 2010;
originally announced July 2010.
-
Titanium Silicide Islands on Atomically Clean Si(100): Identifying Single Electron Tunneling Effects
Authors:
Joseph L. Tedesco,
J. E. Rowe,
Robert J. Nemanich
Abstract:
Titanium silicide islands have been formed by the ultrahigh vacuum deposition of thin films of titanium (< 2 nm) on atomically clean Si(100) substrates followed by annealing to ~800 degrees C. Scanning tunneling microscopy (STM) and scanning tunneling spectroscopy have been performed on these islands to record current-voltage (I-V) curves. Because each island forms a double barrier tunnel junction…
▽ More
Titanium silicide islands have been formed by the ultrahigh vacuum deposition of thin films of titanium (< 2 nm) on atomically clean Si(100) substrates followed by annealing to ~800 degrees C. Scanning tunneling microscopy (STM) and scanning tunneling spectroscopy have been performed on these islands to record current-voltage (I-V) curves. Because each island forms a double barrier tunnel junction (DBTJ) structure with the STM tip and the substrate, they would be expected to exhibit single electron tunneling (SET) according to the orthodox model of SET. Some of the islands formed are small enough (diameter < 10 nm) to exhibit SET at room temperature and evidence of SET has been identified in some of the I-V curves recorded from these small islands. Those curves are analyzed within the framework of the orthodox model and are found to be consistent with that model, except for slight discrepancies of the shape of the I-V curves at current steps. However, most islands that were expected to exhibit SET did not do so, and the reasons for the absence of observable SET are evaluated. The most likely reasons for the absence of SET are determined to be a wide depletion region in the substrate and Schottky barrier lowering due to Fermi level pinning by surface states of the clean silicon near the islands. The results establish that although the Schottky barrier can act as an effective tunnel junction in a DBTJ structure, the islands may be unreliable in future nanoelectronic devices. Therefore, methods are discussed to improve the reliability of future devices.
△ Less
Submitted 28 July, 2010;
originally announced July 2010.
-
Surface Reconstruction of Hexagonal Y-doped HoMnO3 and LuMnO3 studied using low-energy electron diffraction
Authors:
R. Vasic,
J. T. Sadowski,
Y. J. Choi,
H. D. Zhou,
C. R. Wiebe,
S. W. Cheong,
J. E. Rowe,
M. D. Ulrich
Abstract:
We have investigated the (0001) surfaces of several hexagonal manganite perovskites by low-energy electron diffraction (LEED) in order to determine if the surface periodicity is different from that of the bulk materials. These LEED studies were conducted using near-normal incidence geometry with a low energy electron microscope (LEEM)/LEED apparatus from room temperature to 1200 degrees Celsius an…
▽ More
We have investigated the (0001) surfaces of several hexagonal manganite perovskites by low-energy electron diffraction (LEED) in order to determine if the surface periodicity is different from that of the bulk materials. These LEED studies were conducted using near-normal incidence geometry with a low energy electron microscope (LEEM)/LEED apparatus from room temperature to 1200 degrees Celsius and with an electron energy in the range of 15-50 eV. Diffraction patterns showed features of bulk-terminated periodicity as well as a 2\times2 surface reconstruction. Possible origins for this surface reconstruction structure are discussed and comparisons are made with surface studies of other complex oxides.
△ Less
Submitted 7 April, 2010; v1 submitted 16 December, 2009;
originally announced December 2009.
-
Tight Bounds for Blind Search on the Integers
Authors:
Martin Dietzfelbinger,
Jonathan E. Rowe,
Ingo Wegener,
Philipp Woelfel
Abstract:
We analyze a simple random process in which a token is moved in the interval $A=\{0,...,n\$: Fix a probability distribution $μ$ over $\{1,...,n\$. Initially, the token is placed in a random position in $A$. In round $t$, a random value $d$ is chosen according to $μ$. If the token is in position $a\geq d$, then it is moved to position $a-d$. Otherwise it stays put. Let $T$ be the number of rounds…
▽ More
We analyze a simple random process in which a token is moved in the interval $A=\{0,...,n\$: Fix a probability distribution $μ$ over $\{1,...,n\$. Initially, the token is placed in a random position in $A$. In round $t$, a random value $d$ is chosen according to $μ$. If the token is in position $a\geq d$, then it is moved to position $a-d$. Otherwise it stays put. Let $T$ be the number of rounds until the token reaches position 0. We show tight bounds for the expectation of $T$ for the optimal distribution $μ$. More precisely, we show that $\min_μ\{E_μ(T)\=Θ((\log n)^2)$. For the proof, a novel potential function argument is introduced. The research is motivated by the problem of approximating the minimum of a continuous function over $[0,1]$ with a ``blind'' optimization strategy.
△ Less
Submitted 20 February, 2008;
originally announced February 2008.
