Custom Emulator
We may also wish to emulate a completely new quantity, \(f(\theta)\). In this spirit, we propose how to do this with our library here. Throughout, we will assume that we have a tabular output (target), \(y\in \mathbb{R}^{N_{\theta}\times N}\) and a tabular input (cosmological parameters), \(\theta \in \mathbb{R}^{N_{\theta}\times p}\), where \(N_{\theta}\) is the number of training points, \(N\) is the number of nodes, for example, \(\ell\) or \(k\) and \(p\) is the dimensionality of the problem (number of cosmological parameters).