ترغب بنشر مسار تعليمي؟ اضغط هنا

Analysing Large Scale Structure: I. Weighted Scaling Indices and Constrained Randomisation

71   0   0.0 ( 0 )
 نشر من قبل Christoph Raeth
 تاريخ النشر 2002
  مجال البحث فيزياء
والبحث باللغة English




اسأل ChatGPT حول البحث

The method of constrained randomisation is applied to three-dimensional simulated galaxy distributions. With this technique we generate for a given data set surrogate data sets which have the same linear properties as the original data whereas higher order or nonlinear correlations are not preserved. The analysis of the original and surrogate data sets with measures, which are sensitive to nonlinearities, yields information about the existence of nonlinear correlations in the data. We demonstrate how to generate surrogate data sets from a given point distribution, which have the same linear properties (power spectrum) as well as the same density amplitude distribution. We propose weighted scaling indices as a nonlinear statistical measure to quantify local morphological elements in large scale structure. Using surrogates is is shown that the data sets with the same 2-point correlation functions have slightly different void probability functions and especially a different set of weighted scaling indices. Thus a refined analysis of the large scale structure becomes possible by calculating local scaling properties whereby the method of constrained randomisation yields a vital tool for testing the performance of statistical measures in terms of sensitivity to different topological features and discriminative power.



قيم البحث

اقرأ أيضاً

Deep reinforcement learning has the potential to train robots to perform complex tasks in the real world without requiring accurate models of the robot or its environment. A practical approach is to train agents in simulation, and then transfer them to the real world. One popular method for achieving transferability is to use domain randomisation, which involves randomly perturbing various aspects of a simulated environment in order to make trained agents robust to the reality gap. However, less work has gone into understanding such agents - which are deployed in the real world - beyond task performance. In this work we examine such agents, through qualitative and quantitative comparisons between agents trained with and without visual domain randomisation. We train agents for Fetch and Jaco robots on a visuomotor control task and evaluate how well they generalise using different testing conditions. Finally, we investigate the internals of the trained agents by using a suite of interpretability techniques. Our results show that the primary outcome of domain randomisation is more robust, entangled representations, accompanied with larger weights with greater spatial structure; moreover, the types of changes are heavily influenced by the task setup and presence of additional proprioceptive inputs. Additionally, we demonstrate that our domain randomised agents require higher sample complexity, can overfit and more heavily rely on recurrent processing. Furthermore, even with an improved saliency method introduced in this work, we show that qualitative studies may not always correspond with quantitative measures, necessitating the combination of inspection tools in order to provide sufficient insights into the behaviour of trained agents.
113 - C. Raeth , P. Schuecker 2003
The identification of non-Gaussian signatures in cosmic microwave background (CMB) temperature maps is one of the main cosmological challenges today. We propose and investigate altenative methods to analyse CMB maps. Using the technique of constraine d randomisation we construct surrogate maps which mimic both the power spectrum and the amplitude distribution of simulated CMB maps containing non-Gaussian signals. Analysing the maps with weighted scaling indices and Minkowski functionals yield in both cases statistically significant identification of the primordial non-Gaussianities. We demonstrate that the method is very robust with respect to noise. We also show that Minkowski functionals are able to account for non-linearities at higher noise level when applied in combination with surrogates than when only applied to noise added CMB maps and phase randomis
We have extended our evolutionary synthesis code, GALEV, to include Lick/IDS absorption-line indices for both simple and composite stellar population models (star clusters and galaxies), using the polynomial fitting functions of Worthey et al. (1994) and Worthey & Ottaviani (1997). We present a mathematically advanced Lick Index Analysis Tool (LINO) for the determination of ages and metallicities of globular clusters (CGs): An extensive grid of GALEV models for the evolution of star clusters at various metallicities over a Hubble time is compared to observed sets of Lick indices of varying completeness and precision. A dedicated chi^2 - minimisation procedure selects the best model including 1-sigma uncertainties on age and metallicity. We discuss the age and metallicity sensitivities of individual indices and show that these sensitivities themselves depend on age and metallicity; thus, we extend Wortheys (1994) concept of a metallicity sensitivity parameter for an old stellar population at solar metallicity to younger clusters of different metallicities. We find that indices at low metallicity are generally more age sensitive than at high metallicity. Our aim is to provide a robust and reliable tool for the interpretation of star cluster spectra becoming available from 10m class telescopes in a large variety of galaxies: metal-rich & metal-poor, starburst, post-burst, and dynamically young. We test our analysis tool using observations from various authors for Galactic and M31 GCs, for which reliable age and metallicity determinations are available in the literature, and discuss in how far the observational availability of various subsets of Lick indices affects the results. For M31 GCs, we discuss the influence of non-solar abundance ratios on our results.
We consider that no mean magnetic field exists during this epoch, but that there is a mean magnetic energy associated with large-scale magnetic inhomogeneities. We study the evolution of these inhomogeneities and their influence on the large scale de nsity structure, by introducing linear perturbations in Maxwell equations, the conservation of momentum-energy equation, and in Einstein field equations. The primordial magnetic field structure is time independent in the linear approximation, only being diluted by the general expansion, so that $vec{B}R^2$ is conserved in comoving coordinates. Magnetic fields have a strong influence on the formation of large-scale structure. Firstly, relatively low fields are able to generate density structures even if they were inexistent at earlier times. Second, magnetic fields act anisotropically more recently, modifying the evolution of individual density clouds. Magnetic flux tubes have a tendency to concentrate photons in filamentary patterns.
We use the presently observed number density of large X-ray clusters and linear mass power spectra to constrain the shape parameter ($Gamma$), the spectral index ($n$), the amplitude of matter density perturbations on the scale of $8 h^{-1}$Mpc ($sig ma_8$), and the redshift distortion parameter ($beta$). The non-spherical-collapse model as an improvement to the Press-Schechter formula is used and yields significantly lower $sigma_8$ and $beta$. An analytical formalism for the formation redshift of halos is also derived.
التعليقات
جاري جلب التعليقات جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
mircosoft-partner

هل ترغب بارسال اشعارات عن اخر التحديثات في شمرا-اكاديميا