Effect of non-randomness on random variates and discrete event simulation
Latchireddi, Seethapathi Rao
MetadataShow full item record
Random numbers and random variables are the most important aspect of simulation studies. Non-randomness is unavoidable with the use of digital computers to generate random numbers. In such a scenario it is important to understand how non-randomness would affect simulation modeling. A detailed understanding is expected to minimize inconsistencies of the results of use of pseudo random numbers. In this research the impact of typical non-randomness artificially duplicated into a random number generator on random variates and resulting use in discrete event simulation is discussed. Normal, exponential and triangular distributions are studied to understand the effect of non-randomness observed in RANDU type generator on random variate performance. A detailed analysis based on statistical tests is used for hypothesis testing and the findings are expected to help analyze how faulty random number generators used in early Monte Carlo simulations, dozens of research publications which are cited as reference to this day, along with its possible consequences in current research.
Showing items related by title, author, creator and subject.
Stamp, Mark Steven (Texas Tech University, 1988-12)In this paper we present an algorithm for constructing a polynomial system corresponding to a given directed labeled graph. Three specific classes of polynomial systems are considered in detail. Then we briefly consider ...
Central processing unit built-in self-test and enhanced statistical defect analysis compression for static random access memory Francis, David A (Texas Tech University, 2003-08)The percentage of digital signal processors (DSPs) occupied by embedded static random access memory (SRAM) has increased drastically over the years. As a result, static memory has become a prime contributor to the overall ...
Brackett, Benjamin (Texas Tech University, 2004-05)As the size of on-chip caches in continues to increase, wire delays become a more dominant factor in limiting the speed of the cache. The fact that the wire length for each bit in the array is different causes a significant ...