This article is planned.

Topics to cover:

• representing indexed tensor expressions
• canonicalization and simplification
• lowering symbolic expressions to kernels
• static checks for tensor ranks and indices
• bridges to MLIR and C++ backends