# Isocurvature initial conditions for second order Boltzmann solvers

## TL;DR

This paper develops initial conditions for second order cosmological perturbations involving adiabatic and isocurvature modes, highlighting the importance of mixed modes and their non-trivial evolution, to improve non-linear cosmological calculations.

## Contribution

It provides approximate initial solutions for second order transfer functions with mixed modes, including isocurvature and adiabatic combinations, in the context of the b1b4CDM model.

## Key findings

- Mixed modes are sourced by two different initial conditions and are crucial at second order.
- Non-trivial initial evolution occurs for the compensated isocurvature mode when mixed with adiabatic modes.
- Neutrino velocity isocurvature mode generates non-regular decaying modes at second order.

## Abstract

We study how to set the initial evolution of general cosmological fluctuations at second order, after neutrino decoupling. We compute approximate initial solutions for the transfer functions of all the relevant cosmological variables sourced by quadratic combinations of adiabatic and isocurvature modes. We perform these calculations in synchronous gauge, assuming a Universe described by the $\Lambda$CDM model and composed of neutrinos, photons, baryons and dark matter. We highlight the importance of mixed modes, which are sourced by two different isocurvature or adiabatic modes and do not exist at the linear level. In particular, we investigate the so-called compensated isocurvature mode and find non-trivial initial evolution when it is mixed with the adiabatic mode, in contrast to the result at linear order and even at second order for the unmixed mode. Non-trivial evolution also arises when this compensated isocurvature is mixed with the neutrino density isocurvature mode. Regarding the neutrino velocity isocurvature mode, we show it unavoidably generates non-regular (decaying) modes at second order. Our results can be applied to second order Boltzmann solvers to calculate the effects of isocurvatures on non-linear observables.