2,706 research outputs found
Fitness Landscape-Based Characterisation of Nature-Inspired Algorithms
A significant challenge in nature-inspired algorithmics is the identification
of specific characteristics of problems that make them harder (or easier) to
solve using specific methods. The hope is that, by identifying these
characteristics, we may more easily predict which algorithms are best-suited to
problems sharing certain features. Here, we approach this problem using fitness
landscape analysis. Techniques already exist for measuring the "difficulty" of
specific landscapes, but these are often designed solely with evolutionary
algorithms in mind, and are generally specific to discrete optimisation. In
this paper we develop an approach for comparing a wide range of continuous
optimisation algorithms. Using a fitness landscape generation technique, we
compare six different nature-inspired algorithms and identify which methods
perform best on landscapes exhibiting specific features.Comment: 10 pages, 1 figure, submitted to the 11th International Conference on
Adaptive and Natural Computing Algorithm
Quantifying the Impact of Parameter Tuning on Nature-Inspired Algorithms
The problem of parameterization is often central to the effective deployment
of nature-inspired algorithms. However, finding the optimal set of parameter
values for a combination of problem instance and solution method is highly
challenging, and few concrete guidelines exist on how and when such tuning may
be performed. Previous work tends to either focus on a specific algorithm or
use benchmark problems, and both of these restrictions limit the applicability
of any findings. Here, we examine a number of different algorithms, and study
them in a "problem agnostic" fashion (i.e., one that is not tied to specific
instances) by considering their performance on fitness landscapes with varying
characteristics. Using this approach, we make a number of observations on which
algorithms may (or may not) benefit from tuning, and in which specific
circumstances.Comment: 8 pages, 7 figures. Accepted at the European Conference on Artificial
Life (ECAL) 2013, Taormina, Ital
Evaluation of GEOS Total Cloud Fraction with GLOBE Citizen Science Observations and Co-Located Satellite Data
Here we evaluate the total cloud fraction in cycled forecast experiments with the NASA Global Earth Observing System (GEOS) model. Forecasts were run for summer and winter periods of 2017/2018 and compared with ground-based and satellite observations. Citizen science observations from the Global Learning and Observations to Benefit the Environment (GLOBE) Program were matched with MODIS satellite overpasses and geostationary cloud fraction, yielding a dataset of coincident observations for comparison with hourly model output. The observations indicate a model bias toward overcast and clear conditions, with underestimation of intermediate cloud fractions. We investigate underestimation of variance in the sub-grid total water probability density function (PDF) as a possible cause. The PDF determines large-scale cloud fraction as the sub-grid fraction with total water exceeding the saturation specific humidity. We conduct model experiments in which the shape, width, and spatial dependence of the PDF are varied, and consider the impacts on cloud fraction relative to the combined observations
Recommended from our members
Quantifying the latitudinal representivity of in situ solar wind observations
Advanced space-weather forecasting relies on the ability to accurately predict near-Earth solar wind conditions. For this purpose, physics-based, global numerical models of the solar wind are initialized with photospheric magnetic field and coronagraph observations, but no further observation constraints are imposed between the upper corona and Earth orbit. Data assimilation (DA) of the available in situ solar wind observations into the models could potentially provide additional constraints, improving solar wind reconstructions, and forecasts. However, in order to effectively combine the model and observations, it is necessary to quantify the error introduced by assuming point measurements are representative of the model state. In particular, the range of heliographic latitudes over which in situ solar wind speed measurements are representative is of primary importance, but particularly difficult to assess from observations alone. In this study we use 40+ years of observation-driven solar wind model results to assess two related properties: the latitudinal representivity error introduced by assuming the solar wind speed measured at a given latitude is the same as that at the heliographic equator, and the range of latitudes over which a solar wind measurement should influence the model state, referred to as the observational localisation. These values are quantified for future use in solar wind DA schemes as a function of solar cycle phase, measurement latitude, and error tolerance. In general, we find that in situ solar wind speed measurements near the ecliptic plane at solar minimum are extremely localised, being similar over only 1° or 2° of latitude. In the uniform polar fast wind above approximately 40° latitude at solar minimum, the latitudinal representivity error drops. At solar maximum, the increased variability of the solar wind speed at high latitudes means that the latitudinal representivity error increases at the poles, though becomes greater in the ecliptic, as long as moderate speed errors can be tolerated. The heliospheric magnetic field and solar wind density and temperature show very similar behaviour
Peri-urban wild dogs : diet and movements in north-eastern Australia
Knowledge of peri-urban dingo or wild dog ecology can assist management agencies in developing management approaches that alleviate human-wildlife conflicts. Here we summarise (1) the food and dietary items identified in wild dog scats and (2) wild dog movement ecology in urban areas. Individual prey species commonly observed in scats included agile wallabies, northern brown bandicoots and swamp wallabies. Dietary overlap analyses indicated that wild dogs ate the same types or sizes of prey in different regions. In general, wild dogs occupied small fragments of bushland within an urban matrix, were active at all times of the day, and lived within a few hundred meters of houses and humans at all times. These data suggest that urban wild dog management strategies should focus on the mitigation of impacts at the individual or group level, and not population-level reductions in numbers
Computational Calculation of Nitrobenzene and Its Derivatives
Nitrobenzene is one of benzene derivatives. Nitrobenzene can be found naturally and also from the synthesis process. Nitrobenzene is used as a raw material to synthesize aniline, textile dyes, pesticides, and drugs. Nitrobenzene is a solvent in the paint industry. The computational calculation was performed for nitrobenzene and its derivatives. Nitrobenzene and its four nitrobenzene derivatives have been optimized using density functional theory/B3LYP functional. The basis set is 3-21G(d). The optimized structure from geometry optimization of the nitrobenzene and its derivatives are in one plane (planar). The parameter structure is changed when substituents change. The bond length increases, and the bond angle decreases when substituents are present
Population estimation methods, home range and habitat use for wild red deer (Cervus elaphus) at Cressbrook Dam, south-eastern Queensland
Computational Calculation of Nitrobenzene and Its Derivatives
Nitrobenzene is one of benzene derivatives. Nitrobenzene can be found naturally and also from the synthesis process. Nitrobenzene is used as a raw material to synthesize aniline, textile dyes, pesticides, and drugs. Nitrobenzene is a solvent in the paint industry. The computational calculation was performed for nitrobenzene and its derivatives. Nitrobenzene and its four nitrobenzene derivatives have been optimized using density functional theory/B3LYP functional. The basis set is 3-21G(d). The optimized structure from geometry optimization of the nitrobenzene and its derivatives are in one plane (planar). The parameter structure is changed when substituents change. The bond length increases, and the bond angle decreases when substituents are present
Quasigeoid modelling in New Zealand to unify multiple local vertical datums
One goal of modern geodesy is the global unification of vertical datums so that height data from them can be properly integrated. This thesis studies the unification of the 13 disparate levelling- and tide-gauge-based vertical datums in New Zealand (NZ). It proposes a new NZ-wide single vertical datum based on a gravimetric quasigeoid model to unify the existing local vertical datums. This will also include methods to transform height data in terms of the existing datums to the new datum and vice versa. After defining and comparing the main types of height system and vertical datum used around to world, the system of heights used in NZ was shown to be normal-orthometric. Consequently, datum unification was achieved using a quasigeoid model, as opposed to a geoid model. The quasigeoid was computed by combining the GRACE-based GGM02 and EGM96 global geopotential models with land gravity data (40,737 observations) and a 56-m resolution digital elevation model (DEM). Marine gravity data came from a least-squares collocation combination of 1,300,266 crossover-adjusted ship track observations and gravity anomalies derived from multi-mission satellite altimetry.To ensure that the best quasigeoid was computed for the NZ datasets, a number of computation processes were compared and contrasted. The Hammer chart, fast Fourier transform (FFT) and prism integration methods of computing terrain corrections (TCs) were compared. This showed that the prism integration TC is the most realistic in NZ. The mean Helmert gravity anomalies, required for numerical integration of Stokes’s formula, were computed via refined Bouguer anomalies with the prism TCs, and reconstruction with heights from the DEM used to ‘reconstruct’ more representative mean anomalies. In addition, five deterministic modifications to Stokes’s formula were compared. There was little difference between three of them, so the Featherstone et al. (1998) modification ( 0 y = 1.5°, M = 40) was chosen because it is theoretically better than its predecessors. The global geopotential, gravimetric geoid, sea surface topography and geodetic boundary-value problem approaches to vertical datum unification were then contrasted. As none was likely to be effective in NZ, a new iterative quasigeoid approach was adopted. This procedure computes an initial quasigeoid from existing data on the various local vertical datums to estimate the vertical datum offsets from co-located GPS-levelling data. These offsets were then subsequently applied to the gravity observations by way of additional reductions to the initially computed quasigeoid to reduce gravity anomaly biases caused by the vertically offset datums. These adjusted gravity anomalies were then used to compute a new quasigeoid model, and the process repeated until the computed offsets between the local vertical datums (or equivalently two quasigeoid solutions) converged, which took only two iterations.The computed offsets were then compared with spirit-levelled height differences among adjoining datums, where these were available, giving an average agreement of 7 cm. Since the local vertical datums are effectively unified, the new national vertical datum for NZ will comprise the iteratively computed gravimetric quasigeoid model, accompanied by local vertical datums. This approach is implemented to give a new national vertical datum for NZ. When used with the appropriate offset, this enables the transformation of heights in terms of the national vertical datum to the 13 precise-levelling datums and the ellipsoidal national geodetic datum, NZ Geodetic Datum 2000
Management Of Wild Dogs and Deer in Peri-Urban Landscapes: Strategies for Safe Communities
Peri-urban areas are rapidly growing landscapes in many countries worldwide, including Australia. In these areas, where urban areas transition into bush or farmland, traditional pest management methods are largely unsuitable or need refining. There is also an increasing awareness of the impacts of pest animals in these unique landscapes; and the need to better manage natural resources, animal and plant production, and risks to human and animal health. Managing pest animals in the rapidly expanding peri-urban regions of eastern Australia requires tools and strategies markedly different to management in the rural landscape. The two priority pests for many peri-urban councils are wild dogs and deer. Wild dogs are widespread but are becoming increasingly visible in peri-urban areas, raising public awareness and concern. Their impacts range from killing livestock, injuring pets, harassing and injuring humans, being vectors of livestock and zoonotic disease, to preying on threatened wildlife. Peri-urban local governments have identified the need for better tools and strategies for control of wild dogs, red foxes and deer. Managers have had some initial successes in controlling wild dogs, but these approaches should be refined, expanded and tailored to other areas
- …