[Mlir-commits] [mlir] [MLIR][Presburger] Definitions for basic functions related to cones (PR #76650)

[llvmlistbot at llvm.org]

================
@@ -0,0 +1,69 @@
+//===- Barvinok.h - Barvinok's Algorithm ------------------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+//
+// Functions relating to Barvinok's algorithm.
+// These include functions to manipulate cones (define a cone object, get its
+// dual, and find its index).
+//
+//===----------------------------------------------------------------------===//
+
+#ifndef MLIR_ANALYSIS_PRESBURGER_BARVINOK_H
+#define MLIR_ANALYSIS_PRESBURGER_BARVINOK_H
+
+#include "mlir/Analysis/Presburger/IntegerRelation.h"
+#include "mlir/Analysis/Presburger/Matrix.h"
+#include <optional>
+
+namespace mlir {
+namespace presburger {
+
+// A polyhedron in H-representation is a set of inequalities
+// in d variables with integer coefficients.
+using PolyhedronH = IntegerRelation;
+
+// A polyhedron in V-representation is a set of rays and points, i.e.,
+// vectors, stored as rows of a matrix.
+using PolyhedronV = IntMatrix;
+
+// A cone in either representation is a special case of
+// a polyhedron in that representation.
+using ConeH = PolyhedronH;
+using ConeV = PolyhedronV;
+
+inline ConeH defineHRep(int num_vars) {
+  // We don't distinguish between domain and range variables, so
+  // we set the number of domain variables as 0 and the number of
+  // range variables as the number of actual variables.
+  // There are no symbols (we don't work with parametric cones) and no local
+  // (existentially quantified) variables.
+  // Once the cone is defined, we use `addInequality()` to set inequalities.
+  return ConeH(PresburgerSpace::getSetSpace(/*numDims=*/num_vars,
+                                            /*numSymbols=*/0,
+                                            /*numLocals=*/0));
+}
----------------
Abhinav271828 wrote:

I found that it really helps me with writing the code, so I would rather keep them as long as possible tbh
Is it okay if we do a find-replace as part of the last patch?

https://github.com/llvm/llvm-project/pull/76650