Abstract
We present MULTIMODECODE, a Fortran 95/2000 package for the numerical exploration of multifield inflation models. This program facilitates efficient Monte Carlo sampling of prior probabilities for inflationary model parameters and initial conditions and is the first publicly available code that can efficiently generate large sample-sets for inflation models with $O(100)$ fields. The code numerically solves the equations of motion for the background and first-order perturbations of multi-field inflation models with canonical kinetic terms and arbitrary potentials, providing the adiabatic, isocurvature, and tensor power spectra at the end of inflation. For models with sum-separable potentials, MULTIMODECODE also computes the slow-roll prediction via the $\delta N$ formalism for easy model exploration and validation. We pay particular attention to the isocurvature perturbations as the system approaches the adiabatic limit, showing how to avoid numerical instabilities that affect some other approaches to this problem.
Introduction: Scaling Up the Search for Cosmic Origins
In the quest to understand the first fractions of a second after the Big Bang, physicists rely on the theory of cosmic inflation—a period of exponential expansion that smoothed out the universe and planted the seeds for large-scale structures. While simple models involving a single scalar field (the "inflaton") match current data well, high-energy theories like string theory often predict a much more complex landscape involving hundreds of interacting fields.
Navigating this "multifield" complexity requires high-performance numerical tools. MultiModeCode, introduced by Price et al., is a robust Fortran-based package designed to simulate universes with over 100 fields, providing a fast and stable way to calculate primordial power spectra.
The Challenge of Multifield Inflation
Most current understanding of the early universe stems from single-field inflation. While mathematically elegant, these models are often considered "unnatural" from the perspective of fundamental physics, which suggests the early universe was likely populated by a vast "landscape" of fields.
The primary challenge in simulating multifield models is computational expense. These models involve systems of coupled, nonlinear ordinary differential equations (ODEs). Furthermore, as these models evolve, they can approach an "adiabatic limit" where different types of perturbations—specifically adiabatic and isocurvature modes—become nearly indistinguishable. Standard numerical methods often fail in this regime because they attempt to calculate the difference between two very large, nearly identical numbers, leading to catastrophic rounding errors.
Technical Innovations in MultiModeCode
MultiModeCode addresses these challenges through several key technical features:
- Mode Function Evolution: The code solves the Klein-Gordon equations for the background universe and the Mukhanov-Sasaki equations for first-order perturbations.
- Two-Index Mode Matrix: To account for the mixing between different fields, the code tracks a matrix of perturbations rather than a single value.
- Kinematic Basis for Stability: To solve the "adiabatic limit" instability, the code uses a modified definition of isocurvature perturbations based on a kinematic basis. This ensures the calculation remains proportional to the decaying modes themselves rather than a difference between large values, maintaining numerical precision for 27 orders of magnitude longer than previous methods.
- $\delta N$ Formalism: For models with "sum-separable" potentials, the code includes a fast slow-roll approximation that bypasses the need to solve the full perturbation equations.
The table above summarizes the various power spectra and observables that MultiModeCode can compute, ranging from standard scalar and tensor spectra to more complex entropic and cross-spectra.
Validation and Performance
As a software-focused project, MultiModeCode was validated against several inflationary models, including $N_f$-quadratic models and complex 10-field models with potential "steps." The code demonstrated significant scaling efficiency: the computational cost for full perturbations scales as $O(N_f^2)$, while background-only or $\delta N$ calculations scale as $O(N_f)$. This allows the code to handle thousands of fields if only the background dynamics are required.
Furthermore, in a 10-field simulation with a potential step, the code revealed oscillatory features in both the adiabatic and isocurvature power spectra. Interestingly, these oscillations were nearly identical in shape but scaled differently, suggesting a deep connection between how different perturbations are sourced in multifield environments.
Conclusion and Future Directions
MultiModeCode is the first publicly available tool capable of efficiently generating large sample sets for models with $O(100)$ fields. While currently focused on canonical kinetic terms and linear perturbations, future versions are expected to implement non-canonical terms and direct calculations of higher-order statistics like non-Gaussianity. The authors also plan to integrate the tool with standard cosmological sampling packages like CosmoMC and MultiNest to further facilitate parameter estimation in complex inflationary landscapes.
Reference: Price, L. C., Frazer, J., Xu, J., Peiris, H. V., & Easther, R. (2014). MultiModeCode: An efficient numerical solver for multifield inflation. Journal of Cosmology and Astroparticle Physics. arXiv:1410.0685
Summary updated