The CSL model predicts a progressive breakdown of the quantum superposition principle, with a noise randomly driving the state of the system towards a localized one, thus accounting for the emergence of a classical world within a quantum framework. In the original model the noise is supposed to be white, but since white noises do not exist in nature, it becomes relevant to identify some of its spectral properties. Experimental data set an upper bound on its frequencies, while in this paper we bound it from below. We do so in two ways: by considering a 'minimal' measurement setup, requiring that the collapse is completed within the measurement time; and in a measurement modelling-independent way, by requiring that the fluctuations average to zero before the measurement time. We find that the lower bound is comparable to the upper bound coming from bulk heating experiments.
I. INTRODUCTIONCollapse models are a phenomenological solution to the quantum measurement problem, where the Schrödinger dynamics is modified by adding non-linear stochastic terms, which trigger the collapse of the wave function [1,2]. In the widely popular mass-proportional CSL model [3], two parameters control the collapse: a collapse rate λ and the noise space correlator r C . Their numerical values are chosen in such a way that the collapse is negligible for microscopic objects, thus recovering the standard quantum predictions, and become increasingly stronger for larger objects. While there is a generic consensus on the potential value of r C (∼ 10 −7 m), the value of λ is rather open [4,5]. If one requires the collapse to be effective at the level of latent image formation [5], then λ ∼ 10 −8 s −1 , with an uncertainty of about 2 orders of magnitude. This is the *