Consider an alternate way of constructing an SSA program, that do not necessarily adhere to the minimal-SSA (with liveness-based pruning) through dominance-frontiers, and yet produces an SSA program. Can such an alternate way result in an SSA program that has fewer phi nodes than the minimal-SSA (with liveness-based pruning) algorithm (for the same original input program)? If not, provide a brief and clear justification of why not. If yes, show an example program and an example alternate way of constructing an SSA program produces an output with fewer phi nodes than minimal-SSA (with liveness-based pruning).