My particular GA problem creates a large percentage of dead offspring. In fact, each crossover or mutation has less than a 1% chance of producing viable offspring.
What is the best way to replace the DEAD????
Are there any known papers or texts that deal with this in detail?
It's a genetic algorithm. It will start out slow, but will eventually start to flourish. Hopefully your fitness function takes into account the characteristics of this 1%. Give it time, it will evolve.
Think of it like the cambrian explosion in the evolutionary history of Earth. There were only single-celled organisms for a very long time until out of the blue multicellular organisms started popping out everywhere.
Give it time.