E-graphs are a prominent data structure that has been increasing in popularity in recent years due to their expanding range of applications in various formal reasoning tasks. Often, they are used for equality saturation, a process of deriving consequences through repeatedly applying universally quantified equality formulas via term rewriting. They handle equality reasoning over a large spaces of terms, but are severely limited in their handling of case splitting and other types of logical cuts, especially when compared to other reasoning techniques such as sequent calculi and resolution. The main difficulty is when equality reasoning requires multiple inconsistent assumptions to reach a single conclusion. Ad-hoc solutions, such as duplicating the e-graph for each assumption, are available, but they are notably resource-intensive. We introduce a key observation is that each duplicate e-graph (with an added assumption) corresponds to coarsened congruence relation. Based on that, we present an extension to e-graphs, called Colored E-Graphs, as a way to represent all of the coarsened congruence relations in a single structure. A colored e-graph is a memory-efficient equivalent of multiple copies of an e-graph, with a much lower overhead. This is attained by sharing as much as possible between different cases, while carefully tracking which conclusion is true under which assumption. Support for multiple relations can be thought of as adding multiple "color-coded" layers on top of the original e-graph structure, leading to a large degree of sharing. In our implementation, we introduce optimizations to rebuilding and e-matching. We run experiments and demonstrate that our colored e-graphs can support hundreds of assumptions and millions of terms with space requirements that are an order of magnitude lower, and with similar time requirements.
翻译:暂无翻译