Pozměňování či falzifikace psaných textů jsou bezesporu staré jako sám vynález písma. Důvody pro takové jednání byly různé, nicméně skutečné stáří, respektive autenticita, daného rukopisu mají velký dopad na jeho význam, ať již pro historii, nebo - v případě současných rukopisů - z hlediska právního. U některých historických rukopisů je otázka jejich datování řešena s použitím spektroskopických technik, jejichž hlavní výhodou je nedestruktivnost nebo minimální invazivnost, neohrožující samu existenci zkoumaného dokumentu. V článku jsou zmíněny nejčastější spektroskopické metody používané k těmto účelům, včetně příkladů konkrétních studovaných rukopisů., The alteration or falsification of written texts is undoubtedly old as invention of scripture itself. The reasons for such behaviour are different, but the actual age or authenticity of the manuscript had a great impact on its signification, whether for history or, in the case of contemporary manuscripts, from a legal point of view. For historical manuscripts, the question of their dating is solved using spectroscopic techniques whose main advantage is non-destructiveness or minimal invasiveness, not endangering the very existence of the document under study. In the article the most frequent spectroscopic methods used for these purposes are mentioned and examples of particular studied manuscripts are given., Karel Nesměrák., and Obsahuje bibliografické odkazy
Obstructive sleep apnoea syndrome (OSAS) is a common disorder associated with upper airway muscle dysfunction. Agents that improve respiratory muscle performance may have considerable therapeutic value. We examined the effects of acute exposure to sustained and intermittent hypoxia on rat pharyngeal dilator muscle function. Additionally, we sought to test the efficacy of antioxidant treatment in ameliorating or preventing hypoxia-related muscle dysfunction. Isometric contractile and endurance properties of isolated rat sternohyoid muscle bundles were examined at 35 °C in vitro. Muscle bundles were exposed to one of four gas treatments: hyperoxia (control), sustained hypoxia (SH), intermittent hypoxia (IH) or hypoxia/reoxygenation (HR), in the absence or presence of the superoxide scavenger – Tempol (10 mM). Stress-frequency relationship was determined in response to electrical stimulation (10-100 Hz in increments of 10-20 Hz, train duration: 300 ms). Muscle performance was also assessed during repetitive muscle stimulation (40 Hz, 300 ms every 2 s for 2.5 min). Compared to control, IH and HR treatments significantly decreased sternohyoid muscle force. The negative inotropic effect of the two gas protocols was similar, but both were of lesser magnitude than the effects of SH. SH, but not IH and HR, increased muscle fatigue. Tempol significantly increased sensitivity to stimulation in all muscle preparations and caused a leftward shift in the stressfrequency relationship of IH and SH treated muscles. Tempol did not ameliorate sternohyoid muscle fatigue during SH. We conclude that Tempol increases upper airway muscle sensitivity to stimulation but only modestly ameliorates respiratory muscle weakness during intermittent and sustained hypoxic conditions in vitro. Respiratory muscle fatigue during sustained hypoxia appears unrelated to oxidative stress., J. R. Skelly, ... [et al.]., and Obsahuje seznam literatury
It is one of the fundamental and challenging problems to determine the node numbers of hidden layers in neural networks. Various efforts have been made to study the relations between the approximation ability and the number of hidden nodes of some specific neural networks, such as single-hidden-layer and two-hiddenlayer feedforward neural networks with specific or conditional activation functions. However, for arbitrary feedforward neural networks, there are few theoretical results on such issues. This paper gives an upper bound on the node number of each hidden layer for the most general feedforward neural networks called multilayer perceptrons (MLP), from an algebraic point of view. First, we put forward the method of expansion linear spaces to investigate the algebraic structure and properties of the outputs of MLPs. Then it is proved that given k distinct training samples, for any MLP with k nodes in each hidden layer, if a certain optimization problem has solutions, the approximation error keeps invariant with adding nodes to hidden layers. Furthermore, it is shown that for any MLP whose activation function for the output layer is bounded on R, at most k hidden nodes in each hidden layer are needed to learn k training samples.