• 한국산림과학회지
• /
• 제67권1호
• /
• pp.52-59
• /
• 1984

A Gentan probability q(j) is the probability that a newly planted forest will be felled at age-class j. A future change in growing stock and yield of the forests can be predicted by means of this probability. On the other hand a state of the forests is described in terms of an n-vector whose components are the areas of each age-class. This vector, called age-class vector, flows in a n-1 dimensional simplex by means of $n{\times}n$ matrices, whose components are the age-class transition probabilities derived from the Gentan probabilities. In the simplex there exists a fixed point, into which an arbitrary forest age vector sinks. Theoretically this point means a normal state of the forest. To each age-class-transition matrix there corresponds a single normal state; this means that there are infinitely many normal states of the forests.

• 한국지능시스템학회논문지
• /
• 제17권7호
• /
• pp.855-861
• /
• 2007

실제 시스템 적용에 있어서, IMM에 기초한 위치 추적 알고리즘은 불확실한 표적 기동에 대해서 강인한 성능, 적은 연산량, 간편한 설계 절차를 필요로 한다. 이 문제들을 해결하기 위해서 잘 정의된 기저 부모델 및 잘 조정된 모델 천이 확률에 기초한 적응 퍼지 IMM 알고리즘을 제안하였다. 시뮬레이션 결과는 제안된 알고리즘이 IMM에 기초한 알고리즘의 실제 적용에서 존재하는 문제점들을 효과적으로 해결할 수 있음을 보여준다.

• International Journal of Control, Automation, and Systems
• /
• 제6권1호
• /
• pp.109-118
• /
• 2008

We present two estimators for discrete non-Gaussian and nonstationary probability density estimation based on a dynamic Bayesian network (DBN). The first estimator is for off line computation and consists of a DBN whose transition distribution is represented in terms of kernel functions. The estimator parameters are the weights and shifts of the kernel functions. The parameters are determined through a recursive learning algorithm using maximum likelihood (ML) estimation. The second estimator is a DBN whose parameters form the transition probabilities. We use an asymptotically convergent, recursive, on-line algorithm to update the parameters using observation data. The DBN calculates the state probabilities using the estimated parameters. We provide examples that demonstrate the usefulness and simplicity of the two proposed estimators.

• Communications for Statistical Applications and Methods
• /
• 제25권1호
• /
• pp.29-42
• /
• 2018

We combine the integer-valued GARCH(1, 1) model with a generalized regime-switching model to propose a dynamic count time series model. Our model adopts Markov-chains with time-varying dependent transition probabilities to model dynamic count time series called the generalized regime-switching integer-valued GARCH(1, 1) (GRS-INGARCH(1, 1)) models. We derive a recursive formula of the conditional probability of the regime in the Markov-chain given the past information, in terms of transition probabilities of the Markov-chain and the Poisson parameters of the INGARCH(1, 1) process. In addition, we also study the forecasting of the Poisson parameter as well as the cumulative impulse response function of the model, which is a measure for the persistence of volatility. A Monte-Carlo simulation is conducted to see the performances of volatility forecasting and behaviors of cumulative impulse response coefficients as well as conditional maximum likelihood estimation; consequently, a real data application is given.

• 한국정보통신학회논문지
• /
• 제26권2호
• /
• pp.317-320
• /
• 2022

This paper studies real-time error in the context of monitoring a symmetric binary information source over a delay system. To obtain the average real-time error, the delay system is modeled and analyzed as a discrete time Markov chain with a finite state space. Numerical analysis is performed on various system parameters such as state transition probabilities of information source, transmission times, and transmission frequencies. Given state transition probabilities and transmission times, we investigate the relationship between the transmission frequency and the average real-time error. The results can be used to investigate the relationship between real-time errors and age of information.

• 대한화학회지
• /
• 제17권4호
• /
• pp.314-318
• /
• 1973

• 한국환경과학회지
• /
• 제6권2호
• /
• pp.95-106
• /
• 1997

The characteristics on the transtion probabilities and periodicity for the daily precipitation occurrence in Korean peninsula are investigated by applying the Markov chain properties to daily precipitation occurrence. In order to examine the responses of Markov Chain properties to the applied period and their magnitudes, three cases (Case A: 1956~ 1985 at 14 stations, Case B: 1965~ 1994 at 14 stations, and Case C: 1985~ 1994 at 63 stations) are considered In this study. The transition probabilities from wet day to wet day for all cases are about 0.50 and in summer, especially July, are higher. In addition, considering them in each station we can find that they are the highest at Ullung-do and lowest at Inchon for all cases. The annual equilibrium probabilities of a wet day appear 0.31 In Case A, 0.30 Case B, and 0. 29 Case C, respectively. This may explain that as the data-period used becomes shorter, the higher the equilibrium probability is. The seasonal distributions of equilibrium probabilities are appeared the lowest(0.23~0.28) in winter and the highest(more than 0.39) in spring and monthly in .truly and in October, repectively. The annual mean wet duration for all cases is 2.04 days in Case A, 1.99 Case B, and 1.89 Case C, repectively. The weather cycle obtained from the annual mean wet and dry duration is 6.54~6.59 days, which are closely associated with the movement of synoptic systems. And the statistical tests show that the transitions of daily precipitation occurrence for all cases may have two-state first Markov chain property, being the stationarity in time and heterogeneity in space.

• The Journal of the Acoustical Society of Korea
• /
• 제18권1E호
• /
• pp.3-6
• /
• 1999

In digital mobile communication system, the transmission errors affect the quality of output speech seriously. There are many error concealment techniques using a posteriori probability which provides information about any transmitted parameter. They need knowledge about channel transition probability as well as the 1st order Markov transition probability of codec parameters for estimation of transmitted parameters. However, in applications of mobile communication systems, the channel transition probability varies depending on nonstationary channel characteristics. The mismatch of designed channel transition probability of the estimator to actual channel transition probability degrades the performance of the estimator. In this paper, we proposed a new parameter estimator which adapts to the channel characteristics using short time average of maximum a posteriori probabilities(MAPs). The proposed scheme, when applied to the LSP parameter estimation, performed better than the conventional estimator which do not adapt to the channel characteristics.

• 경영과학
• /
• 제8권1호
• /
• pp.3-11
• /
• 1991

Non-Stationary transition probabilities models which is incorporated into a Markov framework with exogenous variables to account for some of variability are discussed, and extended for alternative procedure. Also as an application of the methodology, the size change of aggregate time-series data on the planted system of agricultural crops is estimated, and evaluated for the precision of time-varying evolution statistically.

• Bulletin of the Korean Chemical Society
• /
• 제22권7호
• /
• pp.721-726
• /
• 2001

The vibrational transition probability expressions for the forced Morse oscillator have been derived using the commutation relations of the anharmonic Boson operators. The formulation is based on the collinear collision model with the exponential repulsive potential in the framework of semiclassical collision dynamics. The sample calculation results for H2+ He collision system, where the anharmonicity is large, are in excellent agreement with those from an exact, numerical quantum mechanical study by Clark and Dickinson, using the reactance matrix. Our results, however, are markedly different from those of Ree, Kim and Shin's in which they approximate the commutation operator I。 as unity, the harmonic oscillator limit. We have concluded that the quantum number dependence in I。 must be retained to get accurate vibrational transition probabilities for the Morse oscillator.