Tomographic Forecasts of f_NL with SPHEREx: Optimizing Binning and Probing Systematic Bias

The primordial non-Gaussianity parameter f_NL provides a key signature of inflation, enabling discrimination between models of the early Universe. Through its impact on the scale-dependent galaxy bias, we use Fisher formalism to forecast constraints on f_NL​ using simulated data from the recently launched SPHEREx mission. Unlike previous forecasts, we implement a realistic tomographic approach, identifying how constraints depend on binning and redshift precision of the samples. We find that constraints are driven by high redshift, low-precision galaxies, and only a few redshift bins saturate the available information in the galaxy clustering.

We also investigate one of the dominant systematics for such analyses: spurious fluctuations from errors in galactic dust extinction maps. A new DESI dust map covering ~1/3 of the sky suggests typical errors of ~0.01 mag in the widely used SFD98 map. We forecast how these errors bias f_NL constraints and explore mitigation approaches, including removing angular scales as well as reducing sky area. These results guide analysis choices for upcoming SPHEREx data and underscore the importance of accurate dust maps for precise f_NL​ estimates.