Speculative computing in the condensed graphs machine

The condensed graphs model encompasses both lazy and eager evaluation strategies. By altering the topology of a condensed graph, any mixture of these strategies can be obtained for a given algorithm. Choosing between these strategies, exploiting a conditional operator which allows for the simultaneous reduction of both branches and using a dedicated speculation operator together provide the ingredients used to exploit speculation and to determine when it has become irrelevant. The hierarchical nature of condensed graphs also ensures that all speculative computations have a single coherent representation, facilitating straight-forward termination of irrelevant tasks.