81 research outputs found
Decoders with Initial State Invariance for Multivalued Encodings
Multivaluedencodings constitute an interesting generalization of ordinary encodings in that they allow each source symbol to be encoded by more than one codeword. In this paper we characterize the class of multivaluedencodings that admit invariant decoders and provide an algorithm for constructing such decoders. Invariant decoders have the useful property that their behavior does not depend on the state in which they are, thus exhibiting optimal tolerance to accidental state transitions and/or errors in the input sequence
Efficient q-ary Immutable Codes
A fixed length code is called immutable if no codeword can be transformed into another codeword by using only a restricted set of symbol changes. Immutablecodes are used to prevent undetectable updates of information stored over write-once memories [14]. In this paper we consider immutablecodes on the alphabet Q={0,..., q−1}. We prove that a maximum size immutablecode of block length n can be obtained by taking the set of all vectors in Q^n of weight ⌈n(q−1)⧸2⌉. Furthermore, we propose an encoding rule to map information sequences of length k into codewords of an immutablecode of length k+p. The number k of information digits and the number p of parity digits must satisfy the inequality k≤2(q^p−1)⧸(q−1)−p. The proposed encoding algorithm has computational complexity O(k)
A continuous Markovian model for neuronal activity
A diffusion model for the description of neurons' membrane potential fluctuations is proposed. Though retaining the well known feature consisting of the spontaneous exponential decay of the membrane potential to its resting value, the model discussed differs substantially from the ones in the current literature. Moreover, the Fokker-Planck equation now describing the membrane potential fluctuations is singular. The neuron's firing times probability density function is calculated in closed form as in a first passage time problem, and its expectation value and variance are evaluated. A detailed study of the mode of the firing times probability density function as related to the noise's intensity is performed. Some other auxiliary results are also obtained
A diffusion model for population growth in random environment
The growth of a population in a randomly varying environment is modeled by replacing the Malthusian growth rate with a delta-correlated normal process. The population size is then shown to be a random process, lognormally distributed, obeying a diffusion equation of the Fokker-Planck type. The first passage time p.d.f. through any arbitrarily assigned value and the probability of absorption are derived. The asymptotic behavior of the population size is investigated
A note on growth processes in random environment
The purpose of this note is to discuss certain features of population growth models carlier proposed and to construct an alternative diffusion model for regulated growth in random environment. This model is shown to be the analogue of the Malthusian one, although it is a generalization of the latter due to the presence of regulation
- …
