A frontal solver is an approach to solving sparse linear systems which is used extensively in finite element analysis.
[1] Algorithms of this kind are variants of Gauss elimination that automatically avoids a large number of operations involving zero terms due to the fact that the matrix is only sparse.
[2] The development of frontal solvers is usually considered as dating back to work by Bruce Irons.
[3] A frontal solver builds a LU or Cholesky decomposition of a sparse matrix.
A variation of frontal solvers is the multifrontal method that originates in work of Duff and Reid.