CALT: A Library for Computer Algebra with Transformer

2025-06-10

AI Summary

Non-deep-learning experts face significant barriers in applying Transformer models to symbolic computation tasks. Method: This paper introduces SymTran—a lightweight, Transformer-based Python library that systematically encapsulates sequence-to-sequence modeling capabilities into programmable, computer algebra system (CAS)-oriented APIs for the first time. It employs abstract syntax tree (AST)-driven tokenization of symbolic expressions, structured preprocessing, and an end-to-end PyTorch training framework, enabling unified modeling across diverse symbolic tasks—including algebraic simplification, differentiation, and integration. Contribution/Results: Experiments demonstrate that SymTran achieves high accuracy on multiple symbolic computation benchmarks. By providing a user-friendly, CAS-integrated interface with minimal deep learning prerequisites, SymTran substantially lowers the technical barrier to AI-enhanced symbolic computation and effectively bridges the gap between deep learning and symbolic mathematics communities.

Technology Category

• Machine Learning: Neuro-Symbolic Learning
• Cognitive Modeling & Cognitive Systems: Symbolic Representations
• Computer Vision: Visual Reasoning & Symbolic Representations

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Recent advances in artificial intelligence have demonstrated the learnability of symbolic computation through end-to-end deep learning. Given a sufficient number of examples of symbolic expressions before and after the target computation, Transformer models - highly effective learners of sequence-to-sequence functions - can be trained to emulate the computation. This development opens up several intriguing challenges and new research directions, which require active contributions from the symbolic computation community. In this work, we introduce Computer Algebra with Transformer (CALT), a user-friendly Python library designed to help non-experts in deep learning train models for symbolic computation tasks.

Problem

• Train Transformer models for symbolic computation tasks
• Simplify symbolic computation for non-deep learning experts
• Provide a user-friendly Python library for Computer Algebra

Innovation

• Transformer models for symbolic computation
• Python library for non-experts
• End-to-end deep learning approach