In a nutshell, it would be helpful to have the ability to add chromatic Gaussian events to PINT timing models, in much the same way that we can add chromatic exponential dips.
The IPTA’s third data set includes several pulsars which other PTAs have found needed Gaussian chromatic terms in their timing models. These are predominantly Southern Hemisphere pulsars, in MPTA and/or PPTA data sets. The poster child is J1603-7202, which appeared to exhibit an extreme scattering event (ESE) one decade ago (Coles+2015). Subsequent PTA data releases featuring J1603-7202 have modeled this using a Gaussian delay in time, centered on the ESE, with a chromatic index of 2 (e.g. Reardon+2023; PPTA DR3). We will be including this feature in our vanilla noise modeling for DR3.
Other Gaussian events have been used in PTA timing models. Reardon+2023 also used a Gaussian – an achromatic one – to model a feature in J1600-3053. Additionally, the MPTA’s 4.5-year data release (Miles+2025) found that 15 pulsars in its data set were favored to include Gaussian events with widely-varying chromaticities, interpreted as evidence of complex discrete structures within the interstellar medium. A variable chromatic index is definitely needed here.
Unlike the exponential dips, a Gaussian has no discontinuities, so computing the derivatives with respect to each timing parameter shouldn’t require any workarounds akin to the logistic function used to approximate the exponential dips in PINT.
This isn’t something that needs to be added to PINT for DR3; our current strategy includes the capacity for Gaussian chromatic events only for J1603-7202, and does so in the noise modelling stage through some custom code (a Gaussian function is defined, and then implemented using enterprise.signals. However, it may be useful to someone down the line – and I might open a companion issue for Enterprise, asking it to be built in there, too.
In a nutshell, it would be helpful to have the ability to add chromatic Gaussian events to PINT timing models, in much the same way that we can add chromatic exponential dips.
The IPTA’s third data set includes several pulsars which other PTAs have found needed Gaussian chromatic terms in their timing models. These are predominantly Southern Hemisphere pulsars, in MPTA and/or PPTA data sets. The poster child is J1603-7202, which appeared to exhibit an extreme scattering event (ESE) one decade ago (Coles+2015). Subsequent PTA data releases featuring J1603-7202 have modeled this using a Gaussian delay in time, centered on the ESE, with a chromatic index of 2 (e.g. Reardon+2023; PPTA DR3). We will be including this feature in our vanilla noise modeling for DR3.
Other Gaussian events have been used in PTA timing models. Reardon+2023 also used a Gaussian – an achromatic one – to model a feature in J1600-3053. Additionally, the MPTA’s 4.5-year data release (Miles+2025) found that 15 pulsars in its data set were favored to include Gaussian events with widely-varying chromaticities, interpreted as evidence of complex discrete structures within the interstellar medium. A variable chromatic index is definitely needed here.
Unlike the exponential dips, a Gaussian has no discontinuities, so computing the derivatives with respect to each timing parameter shouldn’t require any workarounds akin to the logistic function used to approximate the exponential dips in PINT.
This isn’t something that needs to be added to PINT for DR3; our current strategy includes the capacity for Gaussian chromatic events only for J1603-7202, and does so in the noise modelling stage through some custom code (a Gaussian function is defined, and then implemented using
enterprise.signals. However, it may be useful to someone down the line – and I might open a companion issue for Enterprise, asking it to be built in there, too.