A note on the Laplace-Varadhan integral lemma (Q1812192): Difference between revisions

When trying to establish a large deviation principle, one is often led to use an ad hoc modification of the Laplace method, since some of the hypotheses in Varadhan's integral lemma might fail to be satisfied (e.g. continuity of the exponent entering in the change of probabilities). The present paper could be useful to anybody facing this kind of situation, since it provides a carefully crafted complement to Varadhan's lemma, where the exponent defining the change of probabilities is replaced by successive approximations depending on the scale parameter and on some auxiliary parameters. The authors illustrate the usefulness of their approach by giving a unified proof of several large deviation principles on path spaces in the context of interacting particle systems. The reader should nevertheless beware of several inaccuracies or misprints.