I will talk about a grid-based parameter estimation algorithm, called Snake, which can be used to sample the likelihood distribution of multidimensional parameter space in order to find the set of parameters that best fit the data. The sampling technique and accuracy of Snake has been tested using 2- and 10-dimensional Gaussian distributions, and its ability to explore cosmological parameter space was investigated using the 7-year data release from the WMAP satellite and compared to results from the Marcov Chain Monte-Carlo (MCMC) approach of CosmoMC, the most commonly used parameter estimation algorithm in cosmology.
Furthermore, the grid regularity of Snake allows for very clean statistics and it is trivial to marginalize over parameters, and to calculate the Bayesian evidence. This evidence may be used to compare how well different models fit the data, and I will present the results of evidence evaluations of the standard 6 parameter cosmological model and a 5 parameter model where a parameter has been fixed. I will also discuss the planned improvements of the algorithm to make it better, faster and more user friendly, as well as where to go next.