In this paper, for a Borel probability measure P on a Euclidean space Rk, we extend the definitions of nth unconstrained quantization error, unconstrained quantization dimension, and unconstrained quantization coefficient, which traditionally in the literature known as nth quantization error, quantization dimension, and quantization coefficient, to the definitions of nth constrained quantization error, constrained quantization dimension, and constrained quantization coefficient. The work in this paper extends the theory of quantization and opens a new area of research. In unconstrained quantization, the elements in an optimal set are the conditional expectations in their own Voronoi regions, and it is not true in constrained quantization. In unconstrained quantization, if the support of P contains infinitely many elements, then an optimal set of n-means always contains exactly n elements, and it is not true in constrained quantization. It is known that the unconstrained quantization dimension for an absolutely continuous probability measure equals the Euclidean dimension of the underlying space. In this paper, we show that this fact is not true as well for the constrained quantization dimension. It is known that the unconstrained quantization coefficient for an absolutely continuous probability measure exists as a unique finite positive number. From work in this paper, it can be seen that the constrained quantization coefficient for an absolutely continuous probability measure can be any nonnegative number depending on the constraint that occurs in the definition of nth constrained quantization error.
Pandey, Megha, and Mrinal K. Roychowdhury. "Constrained quantization for probability distributions." arXiv preprint arXiv:2305.11110 (2023).
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 International License.