Oracle lower bounds for stochastic gradient sampling algorithms (Q2137007)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Oracle lower bounds for stochastic gradient sampling algorithms
scientific article

    Statements

    Oracle lower bounds for stochastic gradient sampling algorithms (English)
    0 references
    0 references
    0 references
    0 references
    0 references
    16 May 2022
    0 references
    The paper considers a problem of sampling from a strongly log-concave density in \(\mathbb R^d,\) and proves that an information theoretic lower bound on the number of stochastic gradient queries of the log density is needed. It is achieved by applying several popular sampling algorithms (including many Markov chain Monte Carlo methods) by using stochastic gradients of the log density to generate a sample. The results of the paper establish an information theoretic limit for all these algorithms.
    0 references
    information theoretic lower bounds
    0 references
    Markov chain Monte Carlo
    0 references
    sampling lower bounds
    0 references
    stochastic gradient Monte Carlo
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references

    Identifiers

    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references