[Mlir-commits] [mlir] [MLIR][Presburger] Implement function to evaluate the number of terms in a generating function. (PR #78078)
llvmlistbot at llvm.org
llvmlistbot at llvm.org
Sun Jan 14 12:06:49 PST 2024
https://github.com/Abhinav271828 updated https://github.com/llvm/llvm-project/pull/78078
>From 1ec4bd325ab80f6d9f1dfbe1db1100ce9fb22219 Mon Sep 17 00:00:00 2001
From: Abhinav271828 <abhinav.m at research.iiit.ac.in>
Date: Sun, 14 Jan 2024 09:39:25 +0530
Subject: [PATCH 01/24] Initial commit
---
.../mlir/Analysis/Presburger/Barvinok.h | 4 +
mlir/lib/Analysis/Presburger/Barvinok.cpp | 197 ++++++++++++++++++
2 files changed, 201 insertions(+)
diff --git a/mlir/include/mlir/Analysis/Presburger/Barvinok.h b/mlir/include/mlir/Analysis/Presburger/Barvinok.h
index edee19f0e1a535..2e2273dab4bc9d 100644
--- a/mlir/include/mlir/Analysis/Presburger/Barvinok.h
+++ b/mlir/include/mlir/Analysis/Presburger/Barvinok.h
@@ -99,6 +99,10 @@ QuasiPolynomial getCoefficientInRationalFunction(unsigned power,
ArrayRef<QuasiPolynomial> num,
ArrayRef<Fraction> den);
+/// Substitute the generating function with the unit vector
+/// to find the number of terms.
+QuasiPolynomial substituteWithUnitVector(GeneratingFunction);
+
} // namespace detail
} // namespace presburger
} // namespace mlir
diff --git a/mlir/lib/Analysis/Presburger/Barvinok.cpp b/mlir/lib/Analysis/Presburger/Barvinok.cpp
index 4ba4462af0317f..0e9d3be7a3289f 100644
--- a/mlir/lib/Analysis/Presburger/Barvinok.cpp
+++ b/mlir/lib/Analysis/Presburger/Barvinok.cpp
@@ -245,3 +245,200 @@ QuasiPolynomial mlir::presburger::detail::getCoefficientInRationalFunction(
}
return coefficients[power].simplify();
}
+
+// We have a generating function of the form
+// f_p(x) = \sum_i sign_i * (x^n_i(p)) / (\prod_j (1 - x^d_{ij})
+//
+// where sign_i is ±1,
+// n_i \in Q^p -> Q^d is a d-vector of affine functions on p parameters, and
+// d_{ij} \in Q^d are vectors.
+//
+// We need to find the number of terms of the form x^t in the expansion of
+// this function, for which we substitute x = (1, ..., 1).
+// However, direct substitution leads to an undefined answer due to the
+// form of the denominator.
+//
+// We therefore use the following procedure instead:
+// Find a vector μ that is not orthogonal to any of the d_{ij}.
+// Substitute x_i = (s+1)^μ_i. As μ_i is not orthogonal to d_{ij},
+// we never have (1 - (s+1)^0) = 0 in any of the terms in denominator.
+// We then find the constant term in this function, i.e., we evaluate it
+// at s = 0, which is equivalent to x = (1, ..., 1).
+//
+// Now, we have a function of the form
+// f_p(s) = \sum_i sign_i * (s+1)^n_i / (\prod_j (1 - (s+1)^d'_{ij}))
+// in which we need to find the constant term.
+// For the i'th term, we first convert all the d'_{ij} to their
+// absolute values by multiplying and dividing by (s+1)^(-d'_{ij}) if it is
+// negative. We change the sign accordingly.
+// Then, we replace each (1 - (s+1)^(d'_{ij})) with
+// (-s)(\sum_{0 ≤ k < d'_{ij}} (s+1)^k).
+// Thus the term overall has now the form
+// sign'_i * (s+1)^n'_i / (s^r * \prod_j (\sum_k (s+1)^k)).
+// This means that
+// the numerator is a polynomial in s, with coefficients as quasipolynomials,
+// and the denominator is polynomial in s, with fractional coefficients.
+// We need to find the constant term in the expansion of this term,
+// which is the same as finding the coefficient of s^r in
+// sign'_i * (s+1)^n'_i / (\prod_j (\sum_k (s+1)^k)),
+// for which we use the `getCoefficientInRationalFunction()` function.
+//
+// Verdoolaege, Sven, et al. "Counting integer points in parametric polytopes
+// using Barvinok's rational functions." Algorithmica 48 (2007): 37-66.
+QuasiPolynomial
+mlir::presburger::detail::substituteWithUnitVector(GeneratingFunction gf) {
+ std::vector<Point> allDenominators;
+ for (std::vector<Point> den : gf.getDenominators())
+ allDenominators.insert(allDenominators.end(), den.begin(), den.end());
+ Point mu = getNonOrthogonalVector(allDenominators);
+
+ unsigned num_params = gf.getNumParams();
+ unsigned num_dims = mu.size();
+ unsigned num_terms = gf.getDenominators().size();
+
+ std::vector<Fraction> dens;
+
+ std::vector<QuasiPolynomial> numeratorCoefficients;
+ std::vector<Fraction> singleTermDenCoefficients, denominatorCoefficients;
+ std::vector<std::vector<Fraction>> eachTermDenCoefficients;
+ std::vector<Fraction> convolution;
+
+ QuasiPolynomial totalTerm(num_params, 0);
+ for (unsigned i = 0; i < num_terms; i++) {
+ int sign = gf.getSigns()[i];
+ ParamPoint v = gf.getNumerators()[i];
+ std::vector<Point> ds = gf.getDenominators()[i];
+
+ // Substitute x_i = (s+1)^μ_i
+ // Then the exponent in the numerator becomes
+ // - (μ • u_1) * (floor(first col of v))
+ // - (μ • u_2) * (floor(second col of v)) - ...
+ // - (μ • u_d) * (floor(d'th col of v))
+ // So we store the negation of the dot produts.
+
+ // We have d terms, each of whose coefficient is the negative dot product,
+ SmallVector<Fraction> coefficients;
+ coefficients.reserve(num_dims);
+ for (Point d : ds)
+ coefficients.push_back(-dotProduct(mu, d));
+
+ // and whose affine fn is a single floor expression, given by the
+ // corresponding column of v.
+ std::vector<std::vector<SmallVector<Fraction>>> affine(num_dims);
+ for (unsigned j = 0; j < num_dims; j++)
+ SmallVector<Fraction> jthCol(v.transpose().getRow(j));
+
+ QuasiPolynomial num(num_params, coefficients, affine);
+ num = num.simplify();
+
+ // Now the numerator is (s+1)^num.
+
+ dens.clear();
+ // Similarly, each term in the denominator has exponent
+ // given by the dot product of μ with u_i.
+ for (Point d : ds)
+ dens.push_back(dotProduct(d, mu));
+ // This term in the denominator is
+ // (1 - (s+1)^dens.back())
+
+ // We track the number of exponents that are negative in the
+ // denominator, and convert them to their absolute values
+ // (see lines 361-71).
+ unsigned numNegExps = 0;
+ Fraction sumNegExps(0, 1);
+ for (unsigned j = 0; j < dens.size(); j++) {
+ if (dens[j] < Fraction(0, 1)) {
+ numNegExps += 1;
+ sumNegExps = sumNegExps + dens[j];
+ }
+ // All exponents will be made positive; then reduce
+ // (1 - (s+1)^x)
+ // to
+ // (-s)*(Σ_{x-1} (s+1)^j) because x > 0
+ dens[j] = abs(dens[j]) - 1;
+ }
+
+ // If we have (1 - (s+1)^-c) in the denominator,
+ // multiply and divide by (s+1)^c.
+ // We convert all negative-exponent terms at once; therefore
+ // we multiply and divide by (s+1)^sumNegExps.
+ // Then we get
+ // -(1 - (s+1)^c) in the denominator,
+ // increase the numerator by c, and
+ // flip the sign.
+ if (numNegExps % 2 == 1)
+ sign = -sign;
+ num = num - QuasiPolynomial(num_params, sumNegExps);
+
+ // Take all the (-s) out, from line 328.
+ unsigned r = dens.size();
+ if (r % 2 == 1)
+ sign = -sign;
+
+ // Now the expression is
+ // (s+1)^num /
+ // (s^r * Π_(0 ≤ i < r) (Σ_{0 ≤ j ≤ dens[i]} (s+1)^j))
+
+ // Letting P(s) = (s+1)^num and Q(s) = Π_r (...),
+ // we need to find the coefficient of s^r in
+ // P(s)/Q(s).
+
+ // First, the coefficients of P(s), which are binomial coefficients.
+ // We need r+1 of these.
+ numeratorCoefficients.clear();
+ numeratorCoefficients.push_back(
+ QuasiPolynomial(num_params, 1)); // Coeff of s^0
+ for (unsigned j = 1; j <= r; j++)
+ numeratorCoefficients.push_back(
+ (numeratorCoefficients[j - 1] *
+ (num - QuasiPolynomial(num_params, j - 1)) / Fraction(j, 1))
+ .simplify());
+ // Coeff of s^j
+
+ // Then the coefficients of each individual term in Q(s),
+ // which are (di+1) C (k+1) for 0 ≤ k ≤ di
+ eachTermDenCoefficients.clear();
+ for (Fraction den : dens) {
+ singleTermDenCoefficients.clear();
+ singleTermDenCoefficients.push_back(den + 1);
+ for (unsigned j = 1; j <= den; j++)
+ singleTermDenCoefficients.push_back(singleTermDenCoefficients[j - 1] *
+ (den - (j - 1)) / (j + 1));
+
+ eachTermDenCoefficients.push_back(singleTermDenCoefficients);
+ }
+
+ // Now we find the coefficients in Q(s) itself
+ // by taking the convolution of the coefficients
+ // of all the terms.
+ denominatorCoefficients.clear();
+ denominatorCoefficients = eachTermDenCoefficients[0];
+ for (unsigned j = 1; j < eachTermDenCoefficients.size(); j++) {
+ // The length of the convolution is the maximum of the lengths
+ // of the two sequences. We pad the shorter one with zeroes.
+ unsigned convlen = std::max(denominatorCoefficients.size(),
+ eachTermDenCoefficients[j].size());
+ for (unsigned k = denominatorCoefficients.size(); k < convlen; k++)
+ denominatorCoefficients.push_back(0);
+ for (unsigned k = eachTermDenCoefficients[j].size(); k < convlen; k++)
+ eachTermDenCoefficients[j].push_back(0);
+
+ convolution.clear();
+ for (unsigned k = 0; k < convlen; k++) {
+ Fraction sum(0, 1);
+ for (unsigned l = 0; l <= k; l++)
+ sum = sum +
+ denominatorCoefficients[l] * eachTermDenCoefficients[j][k - l];
+ convolution.push_back(sum);
+ }
+ denominatorCoefficients = convolution;
+ }
+
+ totalTerm =
+ totalTerm + getCoefficientInRationalFunction(r, numeratorCoefficients,
+ denominatorCoefficients) *
+ QuasiPolynomial(num_params, sign);
+ }
+
+ return totalTerm.simplify();
+}
\ No newline at end of file
>From f286b7b216f98a139b78d7339086dd8c650d336f Mon Sep 17 00:00:00 2001
From: Abhinav271828 <abhinav.m at research.iiit.ac.in>
Date: Sun, 14 Jan 2024 18:17:25 +0530
Subject: [PATCH 02/24] Bug fix
---
mlir/lib/Analysis/Presburger/Barvinok.cpp | 9 +--
.../Analysis/Presburger/BarvinokTest.cpp | 56 +++++++++++++++++++
2 files changed, 61 insertions(+), 4 deletions(-)
diff --git a/mlir/lib/Analysis/Presburger/Barvinok.cpp b/mlir/lib/Analysis/Presburger/Barvinok.cpp
index 0e9d3be7a3289f..785f61b578242f 100644
--- a/mlir/lib/Analysis/Presburger/Barvinok.cpp
+++ b/mlir/lib/Analysis/Presburger/Barvinok.cpp
@@ -324,9 +324,10 @@ mlir::presburger::detail::substituteWithUnitVector(GeneratingFunction gf) {
// and whose affine fn is a single floor expression, given by the
// corresponding column of v.
- std::vector<std::vector<SmallVector<Fraction>>> affine(num_dims);
+ std::vector<std::vector<SmallVector<Fraction>>> affine;
+ affine.reserve(num_dims);
for (unsigned j = 0; j < num_dims; j++)
- SmallVector<Fraction> jthCol(v.transpose().getRow(j));
+ affine.push_back({SmallVector<Fraction>(v.transpose().getRow(j))});
QuasiPolynomial num(num_params, coefficients, affine);
num = num.simplify();
@@ -343,7 +344,7 @@ mlir::presburger::detail::substituteWithUnitVector(GeneratingFunction gf) {
// We track the number of exponents that are negative in the
// denominator, and convert them to their absolute values
- // (see lines 361-71).
+ // (see lines 362-72).
unsigned numNegExps = 0;
Fraction sumNegExps(0, 1);
for (unsigned j = 0; j < dens.size(); j++) {
@@ -370,7 +371,7 @@ mlir::presburger::detail::substituteWithUnitVector(GeneratingFunction gf) {
sign = -sign;
num = num - QuasiPolynomial(num_params, sumNegExps);
- // Take all the (-s) out, from line 328.
+ // Take all the (-s) out, from line 359.
unsigned r = dens.size();
if (r % 2 == 1)
sign = -sign;
diff --git a/mlir/unittests/Analysis/Presburger/BarvinokTest.cpp b/mlir/unittests/Analysis/Presburger/BarvinokTest.cpp
index e304f81de21f0b..8badf8983a2a39 100644
--- a/mlir/unittests/Analysis/Presburger/BarvinokTest.cpp
+++ b/mlir/unittests/Analysis/Presburger/BarvinokTest.cpp
@@ -124,3 +124,59 @@ TEST(BarvinokTest, getCoefficientInRationalFunction) {
coeff = getCoefficientInRationalFunction(3, numerator, denominator);
EXPECT_EQ(coeff.getConstantTerm(), Fraction(55, 64));
}
+
+// The following test is taken from
+//
+TEST(BarvinokTest, substituteWithUnitVector) {
+ GeneratingFunction gf(
+ 1, {1, 1, 1},
+ {makeFracMatrix(2, 2, {{0, Fraction(1, 2)}, {0, 0}}),
+ makeFracMatrix(2, 2, {{0, Fraction(1, 2)}, {0, 0}}),
+ makeFracMatrix(2, 2, {{0, 0}, {0, 0}})},
+ {{{-1, 1}, {-1, 0}},
+ {{1, -1}, {0, -1}},
+ {{1, 0}, {0, 1}}});
+
+ QuasiPolynomial numPoints = substituteWithUnitVector(gf);
+ EXPECT_EQ(
+ numPoints.getCoefficients(),
+ SmallVector<Fraction>(
+ {Fraction(-1, 2), Fraction(-1, 2), Fraction(1, 2), Fraction(-1, 2),
+ Fraction(-1, 2), Fraction(1, 2), Fraction(1, 1), Fraction(3, 2),
+ Fraction(-1, 2), Fraction(3, 2), Fraction(9, 4), Fraction(-3, 4),
+ Fraction(-1, 2), Fraction(-3, 4), Fraction(1, 8), Fraction(1, 8)}));
+ EXPECT_EQ(
+ numPoints.getAffine(),
+ std::vector<std::vector<Point>>(
+ {{{Fraction(1, 2), Fraction(0, 1)}, {Fraction(1, 2), Fraction(0, 1)}},
+ {{Fraction(1, 2), Fraction(0, 1)}},
+ {{Fraction(1, 2), Fraction(0, 1)}},
+ {{Fraction(1, 2), Fraction(0, 1)}},
+ {},
+ {},
+ {{Fraction(1, 2), Fraction(0, 1)}, {Fraction(1, 2), Fraction(0, 1)}},
+ {{Fraction(1, 2), Fraction(0, 1)}},
+ {{Fraction(1, 2), Fraction(0, 1)}},
+ {{Fraction(1, 2), Fraction(0, 1)}},
+ {},
+ {},
+ {{Fraction(1, 2), Fraction(0, 1)}},
+ {},
+ {},
+ {}}));
+
+ // We can gather the like terms because we know there's only
+ // either ⌊p/2⌋^2, ⌊p/2⌋, or constants.
+ Fraction pSquaredCoeff = 0, pCoeff = 0, constantTerm = 0;
+ for (unsigned i = 0; i < numPoints.getCoefficients().size(); i++)
+ if (numPoints.getAffine()[i].size() == 2)
+ pSquaredCoeff = pSquaredCoeff + numPoints.getCoefficients()[i];
+ else if (numPoints.getAffine()[i].size() == 1)
+ pCoeff = pCoeff + numPoints.getCoefficients()[i];
+ else
+ constantTerm = constantTerm + numPoints.getCoefficients()[i];
+
+ EXPECT_EQ(pSquaredCoeff, Fraction(1, 2));
+ EXPECT_EQ(pCoeff, Fraction(3, 2));
+ EXPECT_EQ(constantTerm, Fraction(1, 1));
+}
\ No newline at end of file
>From 19ade35598076669bccebfbc184b7956a1cd9fec Mon Sep 17 00:00:00 2001
From: Abhinav271828 <abhinav.m at research.iiit.ac.in>
Date: Sun, 14 Jan 2024 18:24:09 +0530
Subject: [PATCH 03/24] Add test
---
.../Analysis/Presburger/BarvinokTest.cpp | 54 +++++++------------
1 file changed, 19 insertions(+), 35 deletions(-)
diff --git a/mlir/unittests/Analysis/Presburger/BarvinokTest.cpp b/mlir/unittests/Analysis/Presburger/BarvinokTest.cpp
index 8badf8983a2a39..2842fc14eb8298 100644
--- a/mlir/unittests/Analysis/Presburger/BarvinokTest.cpp
+++ b/mlir/unittests/Analysis/Presburger/BarvinokTest.cpp
@@ -126,57 +126,41 @@ TEST(BarvinokTest, getCoefficientInRationalFunction) {
}
// The following test is taken from
-//
+//
TEST(BarvinokTest, substituteWithUnitVector) {
GeneratingFunction gf(
1, {1, 1, 1},
{makeFracMatrix(2, 2, {{0, Fraction(1, 2)}, {0, 0}}),
makeFracMatrix(2, 2, {{0, Fraction(1, 2)}, {0, 0}}),
makeFracMatrix(2, 2, {{0, 0}, {0, 0}})},
- {{{-1, 1}, {-1, 0}},
- {{1, -1}, {0, -1}},
- {{1, 0}, {0, 1}}});
+ {{{-1, 1}, {-1, 0}}, {{1, -1}, {0, -1}}, {{1, 0}, {0, 1}}});
QuasiPolynomial numPoints = substituteWithUnitVector(gf);
- EXPECT_EQ(
- numPoints.getCoefficients(),
- SmallVector<Fraction>(
- {Fraction(-1, 2), Fraction(-1, 2), Fraction(1, 2), Fraction(-1, 2),
- Fraction(-1, 2), Fraction(1, 2), Fraction(1, 1), Fraction(3, 2),
- Fraction(-1, 2), Fraction(3, 2), Fraction(9, 4), Fraction(-3, 4),
- Fraction(-1, 2), Fraction(-3, 4), Fraction(1, 8), Fraction(1, 8)}));
- EXPECT_EQ(
- numPoints.getAffine(),
- std::vector<std::vector<Point>>(
- {{{Fraction(1, 2), Fraction(0, 1)}, {Fraction(1, 2), Fraction(0, 1)}},
- {{Fraction(1, 2), Fraction(0, 1)}},
- {{Fraction(1, 2), Fraction(0, 1)}},
- {{Fraction(1, 2), Fraction(0, 1)}},
- {},
- {},
- {{Fraction(1, 2), Fraction(0, 1)}, {Fraction(1, 2), Fraction(0, 1)}},
- {{Fraction(1, 2), Fraction(0, 1)}},
- {{Fraction(1, 2), Fraction(0, 1)}},
- {{Fraction(1, 2), Fraction(0, 1)}},
- {},
- {},
- {{Fraction(1, 2), Fraction(0, 1)}},
- {},
- {},
- {}}));
-
- // We can gather the like terms because we know there's only
+
+ // First, we make sure that all the affine functions are of the form ⌊p/2⌋.
+ for (const std::vector<SmallVector<Fraction>> &term : numPoints.getAffine()) {
+ for (const SmallVector<Fraction> &aff : term) {
+ EXPECT_EQ(aff.size(), 2u);
+ EXPECT_EQ(aff[0], Fraction(1, 2));
+ EXPECT_EQ(aff[1], Fraction(0, 1));
+ }
+ }
+
+ // Now, we can gather the like terms because we know there's only
// either ⌊p/2⌋^2, ⌊p/2⌋, or constants.
+ // The total coefficient of ⌊p/2⌋^2 is the sum of coefficients of all
+ // terms with 2 affine functions, and
+ // the coefficient of total ⌊p/2⌋ is the sum of coefficients of all
+ // terms with 1 affine function,
Fraction pSquaredCoeff = 0, pCoeff = 0, constantTerm = 0;
for (unsigned i = 0; i < numPoints.getCoefficients().size(); i++)
if (numPoints.getAffine()[i].size() == 2)
pSquaredCoeff = pSquaredCoeff + numPoints.getCoefficients()[i];
else if (numPoints.getAffine()[i].size() == 1)
pCoeff = pCoeff + numPoints.getCoefficients()[i];
- else
- constantTerm = constantTerm + numPoints.getCoefficients()[i];
+ // We expect the answer to be (1/2)⌊p/2⌋^2 + (3/2)⌊p/2⌋ + 1.
EXPECT_EQ(pSquaredCoeff, Fraction(1, 2));
EXPECT_EQ(pCoeff, Fraction(3, 2));
- EXPECT_EQ(constantTerm, Fraction(1, 1));
+ EXPECT_EQ(numPoints.getConstantTerm(), Fraction(1, 1));
}
\ No newline at end of file
>From 39dd6a8a1c87ac33d6b6e31a92374c7a2409892c Mon Sep 17 00:00:00 2001
From: Abhinav271828 <abhinav.m at research.iiit.ac.in>
Date: Sun, 14 Jan 2024 18:27:59 +0530
Subject: [PATCH 04/24] Minor fixes
---
mlir/lib/Analysis/Presburger/Barvinok.cpp | 26 +++++++++++------------
1 file changed, 13 insertions(+), 13 deletions(-)
diff --git a/mlir/lib/Analysis/Presburger/Barvinok.cpp b/mlir/lib/Analysis/Presburger/Barvinok.cpp
index 785f61b578242f..e580e40f6034cc 100644
--- a/mlir/lib/Analysis/Presburger/Barvinok.cpp
+++ b/mlir/lib/Analysis/Presburger/Barvinok.cpp
@@ -288,7 +288,7 @@ QuasiPolynomial mlir::presburger::detail::getCoefficientInRationalFunction(
QuasiPolynomial
mlir::presburger::detail::substituteWithUnitVector(GeneratingFunction gf) {
std::vector<Point> allDenominators;
- for (std::vector<Point> den : gf.getDenominators())
+ for (ArrayRef<Point> den : gf.getDenominators())
allDenominators.insert(allDenominators.end(), den.begin(), den.end());
Point mu = getNonOrthogonalVector(allDenominators);
@@ -304,7 +304,7 @@ mlir::presburger::detail::substituteWithUnitVector(GeneratingFunction gf) {
std::vector<Fraction> convolution;
QuasiPolynomial totalTerm(num_params, 0);
- for (unsigned i = 0; i < num_terms; i++) {
+ for (unsigned i = 0; i < num_terms; ++i) {
int sign = gf.getSigns()[i];
ParamPoint v = gf.getNumerators()[i];
std::vector<Point> ds = gf.getDenominators()[i];
@@ -326,7 +326,7 @@ mlir::presburger::detail::substituteWithUnitVector(GeneratingFunction gf) {
// corresponding column of v.
std::vector<std::vector<SmallVector<Fraction>>> affine;
affine.reserve(num_dims);
- for (unsigned j = 0; j < num_dims; j++)
+ for (unsigned j = 0; j < num_dims; ++j)
affine.push_back({SmallVector<Fraction>(v.transpose().getRow(j))});
QuasiPolynomial num(num_params, coefficients, affine);
@@ -347,10 +347,10 @@ mlir::presburger::detail::substituteWithUnitVector(GeneratingFunction gf) {
// (see lines 362-72).
unsigned numNegExps = 0;
Fraction sumNegExps(0, 1);
- for (unsigned j = 0; j < dens.size(); j++) {
- if (dens[j] < Fraction(0, 1)) {
+ for (unsigned j = 0, e = dens.size(); j < e; ++j) {
+ if (dens[j] < 0) {
numNegExps += 1;
- sumNegExps = sumNegExps + dens[j];
+ sumNegExps += dens[j];
}
// All exponents will be made positive; then reduce
// (1 - (s+1)^x)
@@ -389,7 +389,7 @@ mlir::presburger::detail::substituteWithUnitVector(GeneratingFunction gf) {
numeratorCoefficients.clear();
numeratorCoefficients.push_back(
QuasiPolynomial(num_params, 1)); // Coeff of s^0
- for (unsigned j = 1; j <= r; j++)
+ for (unsigned j = 1; j <= r; ++j)
numeratorCoefficients.push_back(
(numeratorCoefficients[j - 1] *
(num - QuasiPolynomial(num_params, j - 1)) / Fraction(j, 1))
@@ -402,7 +402,7 @@ mlir::presburger::detail::substituteWithUnitVector(GeneratingFunction gf) {
for (Fraction den : dens) {
singleTermDenCoefficients.clear();
singleTermDenCoefficients.push_back(den + 1);
- for (unsigned j = 1; j <= den; j++)
+ for (unsigned j = 1; j <= den; ++j)
singleTermDenCoefficients.push_back(singleTermDenCoefficients[j - 1] *
(den - (j - 1)) / (j + 1));
@@ -414,20 +414,20 @@ mlir::presburger::detail::substituteWithUnitVector(GeneratingFunction gf) {
// of all the terms.
denominatorCoefficients.clear();
denominatorCoefficients = eachTermDenCoefficients[0];
- for (unsigned j = 1; j < eachTermDenCoefficients.size(); j++) {
+ for (unsigned j = 1, e = eachTermDenCoefficients.size(); j < e; ++j) {
// The length of the convolution is the maximum of the lengths
// of the two sequences. We pad the shorter one with zeroes.
unsigned convlen = std::max(denominatorCoefficients.size(),
eachTermDenCoefficients[j].size());
- for (unsigned k = denominatorCoefficients.size(); k < convlen; k++)
+ for (unsigned k = denominatorCoefficients.size(); k < convlen; ++k)
denominatorCoefficients.push_back(0);
- for (unsigned k = eachTermDenCoefficients[j].size(); k < convlen; k++)
+ for (unsigned k = eachTermDenCoefficients[j].size(); k < convlen; ++k)
eachTermDenCoefficients[j].push_back(0);
convolution.clear();
- for (unsigned k = 0; k < convlen; k++) {
+ for (unsigned k = 0; k < convlen; ++k) {
Fraction sum(0, 1);
- for (unsigned l = 0; l <= k; l++)
+ for (unsigned l = 0; l <= k; ++l)
sum = sum +
denominatorCoefficients[l] * eachTermDenCoefficients[j][k - l];
convolution.push_back(sum);
>From 64a55bc0a8991270f3a7e7ff56c33de6fba3de8b Mon Sep 17 00:00:00 2001
From: Abhinav271828 <abhinav.m at research.iiit.ac.in>
Date: Sun, 14 Jan 2024 18:33:17 +0530
Subject: [PATCH 05/24] Use const
---
mlir/lib/Analysis/Presburger/Barvinok.cpp | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/mlir/lib/Analysis/Presburger/Barvinok.cpp b/mlir/lib/Analysis/Presburger/Barvinok.cpp
index e580e40f6034cc..180c3f047c6010 100644
--- a/mlir/lib/Analysis/Presburger/Barvinok.cpp
+++ b/mlir/lib/Analysis/Presburger/Barvinok.cpp
@@ -319,7 +319,7 @@ mlir::presburger::detail::substituteWithUnitVector(GeneratingFunction gf) {
// We have d terms, each of whose coefficient is the negative dot product,
SmallVector<Fraction> coefficients;
coefficients.reserve(num_dims);
- for (Point d : ds)
+ for (const Point &d : ds)
coefficients.push_back(-dotProduct(mu, d));
// and whose affine fn is a single floor expression, given by the
@@ -337,7 +337,7 @@ mlir::presburger::detail::substituteWithUnitVector(GeneratingFunction gf) {
dens.clear();
// Similarly, each term in the denominator has exponent
// given by the dot product of μ with u_i.
- for (Point d : ds)
+ for (const Point &d : ds)
dens.push_back(dotProduct(d, mu));
// This term in the denominator is
// (1 - (s+1)^dens.back())
@@ -399,7 +399,7 @@ mlir::presburger::detail::substituteWithUnitVector(GeneratingFunction gf) {
// Then the coefficients of each individual term in Q(s),
// which are (di+1) C (k+1) for 0 ≤ k ≤ di
eachTermDenCoefficients.clear();
- for (Fraction den : dens) {
+ for (const Fraction &den : dens) {
singleTermDenCoefficients.clear();
singleTermDenCoefficients.push_back(den + 1);
for (unsigned j = 1; j <= den; ++j)
>From b3331b2c736d1a99fe360c90e17fd98ce2ac08b0 Mon Sep 17 00:00:00 2001
From: Abhinav271828 <abhinav.m at research.iiit.ac.in>
Date: Sun, 14 Jan 2024 18:41:09 +0530
Subject: [PATCH 06/24] Push down declarations
---
mlir/lib/Analysis/Presburger/Barvinok.cpp | 51 +++++++++++------------
1 file changed, 25 insertions(+), 26 deletions(-)
diff --git a/mlir/lib/Analysis/Presburger/Barvinok.cpp b/mlir/lib/Analysis/Presburger/Barvinok.cpp
index 180c3f047c6010..8b3a91acdc83a9 100644
--- a/mlir/lib/Analysis/Presburger/Barvinok.cpp
+++ b/mlir/lib/Analysis/Presburger/Barvinok.cpp
@@ -292,19 +292,12 @@ mlir::presburger::detail::substituteWithUnitVector(GeneratingFunction gf) {
allDenominators.insert(allDenominators.end(), den.begin(), den.end());
Point mu = getNonOrthogonalVector(allDenominators);
- unsigned num_params = gf.getNumParams();
- unsigned num_dims = mu.size();
- unsigned num_terms = gf.getDenominators().size();
+ unsigned numParams = gf.getNumParams();
+ unsigned numDims = mu.size();
+ unsigned numTerms = gf.getDenominators().size();
- std::vector<Fraction> dens;
-
- std::vector<QuasiPolynomial> numeratorCoefficients;
- std::vector<Fraction> singleTermDenCoefficients, denominatorCoefficients;
- std::vector<std::vector<Fraction>> eachTermDenCoefficients;
- std::vector<Fraction> convolution;
-
- QuasiPolynomial totalTerm(num_params, 0);
- for (unsigned i = 0; i < num_terms; ++i) {
+ QuasiPolynomial totalTerm(numParams, 0);
+ for (unsigned i = 0; i < numTerms; ++i) {
int sign = gf.getSigns()[i];
ParamPoint v = gf.getNumerators()[i];
std::vector<Point> ds = gf.getDenominators()[i];
@@ -318,23 +311,24 @@ mlir::presburger::detail::substituteWithUnitVector(GeneratingFunction gf) {
// We have d terms, each of whose coefficient is the negative dot product,
SmallVector<Fraction> coefficients;
- coefficients.reserve(num_dims);
+ coefficients.reserve(numDims);
for (const Point &d : ds)
coefficients.push_back(-dotProduct(mu, d));
// and whose affine fn is a single floor expression, given by the
// corresponding column of v.
std::vector<std::vector<SmallVector<Fraction>>> affine;
- affine.reserve(num_dims);
- for (unsigned j = 0; j < num_dims; ++j)
+ affine.reserve(numDims);
+ for (unsigned j = 0; j < numDims; ++j)
affine.push_back({SmallVector<Fraction>(v.transpose().getRow(j))});
- QuasiPolynomial num(num_params, coefficients, affine);
+ QuasiPolynomial num(numParams, coefficients, affine);
num = num.simplify();
// Now the numerator is (s+1)^num.
- dens.clear();
+ std::vector<Fraction> dens;
+ dens.reserve(ds.size());
// Similarly, each term in the denominator has exponent
// given by the dot product of μ with u_i.
for (const Point &d : ds)
@@ -344,7 +338,7 @@ mlir::presburger::detail::substituteWithUnitVector(GeneratingFunction gf) {
// We track the number of exponents that are negative in the
// denominator, and convert them to their absolute values
- // (see lines 362-72).
+ // (see lines 356-66).
unsigned numNegExps = 0;
Fraction sumNegExps(0, 1);
for (unsigned j = 0, e = dens.size(); j < e; ++j) {
@@ -369,9 +363,9 @@ mlir::presburger::detail::substituteWithUnitVector(GeneratingFunction gf) {
// flip the sign.
if (numNegExps % 2 == 1)
sign = -sign;
- num = num - QuasiPolynomial(num_params, sumNegExps);
+ num = num - QuasiPolynomial(numParams, sumNegExps);
- // Take all the (-s) out, from line 359.
+ // Take all the (-s) out, from line 353.
unsigned r = dens.size();
if (r % 2 == 1)
sign = -sign;
@@ -386,19 +380,22 @@ mlir::presburger::detail::substituteWithUnitVector(GeneratingFunction gf) {
// First, the coefficients of P(s), which are binomial coefficients.
// We need r+1 of these.
- numeratorCoefficients.clear();
+ std::vector<QuasiPolynomial> numeratorCoefficients;
+ numeratorCoefficients.reserve(r + 1);
numeratorCoefficients.push_back(
- QuasiPolynomial(num_params, 1)); // Coeff of s^0
+ QuasiPolynomial(numParams, 1)); // Coeff of s^0
for (unsigned j = 1; j <= r; ++j)
numeratorCoefficients.push_back(
(numeratorCoefficients[j - 1] *
- (num - QuasiPolynomial(num_params, j - 1)) / Fraction(j, 1))
+ (num - QuasiPolynomial(numParams, j - 1)) / Fraction(j, 1))
.simplify());
// Coeff of s^j
// Then the coefficients of each individual term in Q(s),
// which are (di+1) C (k+1) for 0 ≤ k ≤ di
- eachTermDenCoefficients.clear();
+ std::vector<std::vector<Fraction>> eachTermDenCoefficients;
+ std::vector<Fraction> singleTermDenCoefficients;
+ eachTermDenCoefficients.reserve(r);
for (const Fraction &den : dens) {
singleTermDenCoefficients.clear();
singleTermDenCoefficients.push_back(den + 1);
@@ -412,7 +409,7 @@ mlir::presburger::detail::substituteWithUnitVector(GeneratingFunction gf) {
// Now we find the coefficients in Q(s) itself
// by taking the convolution of the coefficients
// of all the terms.
- denominatorCoefficients.clear();
+ std::vector<Fraction> denominatorCoefficients;
denominatorCoefficients = eachTermDenCoefficients[0];
for (unsigned j = 1, e = eachTermDenCoefficients.size(); j < e; ++j) {
// The length of the convolution is the maximum of the lengths
@@ -424,6 +421,8 @@ mlir::presburger::detail::substituteWithUnitVector(GeneratingFunction gf) {
for (unsigned k = eachTermDenCoefficients[j].size(); k < convlen; ++k)
eachTermDenCoefficients[j].push_back(0);
+ std::vector<Fraction> convolution;
+ convolution.reserve(convlen);
convolution.clear();
for (unsigned k = 0; k < convlen; ++k) {
Fraction sum(0, 1);
@@ -438,7 +437,7 @@ mlir::presburger::detail::substituteWithUnitVector(GeneratingFunction gf) {
totalTerm =
totalTerm + getCoefficientInRationalFunction(r, numeratorCoefficients,
denominatorCoefficients) *
- QuasiPolynomial(num_params, sign);
+ QuasiPolynomial(numParams, sign);
}
return totalTerm.simplify();
>From f78ef764817c10bbbf2375556dd1fb502d2686e6 Mon Sep 17 00:00:00 2001
From: Abhinav271828 <abhinav.m at research.iiit.ac.in>
Date: Sun, 14 Jan 2024 23:02:15 +0530
Subject: [PATCH 07/24] Mark getters as const
---
.../mlir/Analysis/Presburger/GeneratingFunction.h | 10 ++++++----
1 file changed, 6 insertions(+), 4 deletions(-)
diff --git a/mlir/include/mlir/Analysis/Presburger/GeneratingFunction.h b/mlir/include/mlir/Analysis/Presburger/GeneratingFunction.h
index eaf0449fe8a118..c38eab6efd0fc1 100644
--- a/mlir/include/mlir/Analysis/Presburger/GeneratingFunction.h
+++ b/mlir/include/mlir/Analysis/Presburger/GeneratingFunction.h
@@ -62,13 +62,15 @@ class GeneratingFunction {
#endif // NDEBUG
}
- unsigned getNumParams() { return numParam; }
+ unsigned getNumParams() const { return numParam; }
- SmallVector<int> getSigns() { return signs; }
+ SmallVector<int> getSigns() const { return signs; }
- std::vector<ParamPoint> getNumerators() { return numerators; }
+ std::vector<ParamPoint> getNumerators() const { return numerators; }
- std::vector<std::vector<Point>> getDenominators() { return denominators; }
+ std::vector<std::vector<Point>> getDenominators() const {
+ return denominators;
+ }
GeneratingFunction operator+(GeneratingFunction &gf) const {
assert(numParam == gf.getNumParams() &&
>From 06ca87a6de82a4d36ba54508bc979cc0eb5ef5de Mon Sep 17 00:00:00 2001
From: Abhinav271828 <abhinav.m at research.iiit.ac.in>
Date: Sun, 14 Jan 2024 23:02:41 +0530
Subject: [PATCH 08/24] Fix doc
---
.../mlir/Analysis/Presburger/Barvinok.h | 6 +-
mlir/lib/Analysis/Presburger/Barvinok.cpp | 95 +++++++++++--------
2 files changed, 58 insertions(+), 43 deletions(-)
diff --git a/mlir/include/mlir/Analysis/Presburger/Barvinok.h b/mlir/include/mlir/Analysis/Presburger/Barvinok.h
index 2e2273dab4bc9d..d72332d220d2b5 100644
--- a/mlir/include/mlir/Analysis/Presburger/Barvinok.h
+++ b/mlir/include/mlir/Analysis/Presburger/Barvinok.h
@@ -99,9 +99,9 @@ QuasiPolynomial getCoefficientInRationalFunction(unsigned power,
ArrayRef<QuasiPolynomial> num,
ArrayRef<Fraction> den);
-/// Substitute the generating function with the unit vector
-/// to find the number of terms.
-QuasiPolynomial substituteWithUnitVector(GeneratingFunction);
+/// Find the number of terms in the generating function corresponding to
+/// a polytope.
+QuasiPolynomial computeNumTerms(const GeneratingFunction& gf);
} // namespace detail
} // namespace presburger
diff --git a/mlir/lib/Analysis/Presburger/Barvinok.cpp b/mlir/lib/Analysis/Presburger/Barvinok.cpp
index 8b3a91acdc83a9..9eac53a96e9e0b 100644
--- a/mlir/lib/Analysis/Presburger/Barvinok.cpp
+++ b/mlir/lib/Analysis/Presburger/Barvinok.cpp
@@ -246,47 +246,62 @@ QuasiPolynomial mlir::presburger::detail::getCoefficientInRationalFunction(
return coefficients[power].simplify();
}
-// We have a generating function of the form
-// f_p(x) = \sum_i sign_i * (x^n_i(p)) / (\prod_j (1 - x^d_{ij})
-//
-// where sign_i is ±1,
-// n_i \in Q^p -> Q^d is a d-vector of affine functions on p parameters, and
-// d_{ij} \in Q^d are vectors.
-//
-// We need to find the number of terms of the form x^t in the expansion of
-// this function, for which we substitute x = (1, ..., 1).
-// However, direct substitution leads to an undefined answer due to the
-// form of the denominator.
-//
-// We therefore use the following procedure instead:
-// Find a vector μ that is not orthogonal to any of the d_{ij}.
-// Substitute x_i = (s+1)^μ_i. As μ_i is not orthogonal to d_{ij},
-// we never have (1 - (s+1)^0) = 0 in any of the terms in denominator.
-// We then find the constant term in this function, i.e., we evaluate it
-// at s = 0, which is equivalent to x = (1, ..., 1).
-//
-// Now, we have a function of the form
-// f_p(s) = \sum_i sign_i * (s+1)^n_i / (\prod_j (1 - (s+1)^d'_{ij}))
-// in which we need to find the constant term.
-// For the i'th term, we first convert all the d'_{ij} to their
-// absolute values by multiplying and dividing by (s+1)^(-d'_{ij}) if it is
-// negative. We change the sign accordingly.
-// Then, we replace each (1 - (s+1)^(d'_{ij})) with
-// (-s)(\sum_{0 ≤ k < d'_{ij}} (s+1)^k).
-// Thus the term overall has now the form
-// sign'_i * (s+1)^n'_i / (s^r * \prod_j (\sum_k (s+1)^k)).
-// This means that
-// the numerator is a polynomial in s, with coefficients as quasipolynomials,
-// and the denominator is polynomial in s, with fractional coefficients.
-// We need to find the constant term in the expansion of this term,
-// which is the same as finding the coefficient of s^r in
-// sign'_i * (s+1)^n'_i / (\prod_j (\sum_k (s+1)^k)),
-// for which we use the `getCoefficientInRationalFunction()` function.
-//
-// Verdoolaege, Sven, et al. "Counting integer points in parametric polytopes
-// using Barvinok's rational functions." Algorithmica 48 (2007): 37-66.
+/// We have a generating function of the form
+/// f_p(x) = \sum_i sign_i * (x^n_i(p)) / (\prod_j (1 - x^d_{ij})
+///
+/// where sign_i is ±1,
+/// n_i \in Q^p -> Q^d is a d-vector of affine functions on p parameters, and
+/// d_{ij} \in Q^d are vectors.
+///
+/// We need to find the number of terms of the form x^t in the expansion of
+/// this function, for which we substitute x = (1, ..., 1).
+/// However, direct substitution causes the denominator to become zero.
+///
+/// We therefore use the following procedure instead:
+/// 1. Substitute x_i = (s+1)^μ_i for some vector μ. This makes the generating
+/// a function of a scalar s.
+/// 2. Write each term in this function as P(s)/Q(s), where P and Q are
+/// polynomials. P has coefficients as quasipolynomials in d parameters, while
+/// Q has coefficients as scalars.
+/// 3. Find the constant term in the expansion of each term P(s)/Q(s). This is
+/// equivalent to substituting s = 0, which by step 1's substitution is
+/// equivalent to letting x = (1, ..., 1).
+/// In this step, we cancel the factor with root zero from the numerator and
+/// denominator, thus preventing the denominator from becoming zero.
+/// Step (1) We need to find a μ_i such that we can substitute x_i =
+/// (s+1)^μ_i. After this substitution, the exponent of (s+1) in the
+/// denominator is (μ_i • d_{ij}) in each term. Clearly, this cannot become
+/// zero. Thus we find a vector μ that is not orthogonal to any of the
+/// d_{ij} and substitute x accordingly.
+///
+/// Step (2) We need to express the terms in the function as quotients of
+/// polynomials. Each term is now of the form
+/// sign_i * (s+1)^n_i / (\prod_j (1 - (s+1)^d'_{ij}))
+/// For the i'th term, we first convert all the d'_{ij} to their
+/// absolute values by multiplying and dividing by (s+1)^(-d'_{ij}) if it is
+/// negative. We change the sign accordingly.
+/// Then, we replace each (1 - (s+1)^(d'_{ij})) with
+/// (-s)(\sum_{0 ≤ k < d'_{ij}} (s+1)^k).
+/// Thus the term overall has now the form
+/// sign'_i * (s+1)^n'_i / (s^r * \prod_j (\sum_k (s+1)^k)).
+/// This means that
+/// the numerator is a polynomial in s, with coefficients as
+/// quasipolynomials (given by binomial coefficients), and the denominator
+/// is polynomial in s, with fractional coefficients (given by taking the
+/// convolution over all j).
+///
+/// Step (3) We need to find the constant term in the expansion of each
+/// term. Since each term has s^r as a factor in the denominator, we avoid
+/// substituting s = 0 directly; instead, we find the coefficient of s^r in
+/// sign'_i * (s+1)^n'_i / (\prod_j (\sum_k (s+1)^k)),
+/// for which we use the `getCoefficientInRationalFunction()` function.
+///
+/// Verdoolaege, Sven, et al. "Counting integer points in parametric
+/// polytopes using Barvinok's rational functions." Algorithmica 48 (2007):
+/// 37-66.
+
QuasiPolynomial
-mlir::presburger::detail::substituteWithUnitVector(GeneratingFunction gf) {
+mlir::presburger::detail::computeNumTerms(const GeneratingFunction &gf) {
std::vector<Point> allDenominators;
for (ArrayRef<Point> den : gf.getDenominators())
allDenominators.insert(allDenominators.end(), den.begin(), den.end());
>From 92a73fceb9b547917c4a2f0f6dcb68c1aab2794c Mon Sep 17 00:00:00 2001
From: Abhinav271828 <abhinav.m at research.iiit.ac.in>
Date: Sun, 14 Jan 2024 23:32:06 +0530
Subject: [PATCH 09/24] Abstract convolution
---
mlir/lib/Analysis/Presburger/Barvinok.cpp | 63 ++++++++++---------
.../Analysis/Presburger/BarvinokTest.cpp | 4 +-
2 files changed, 34 insertions(+), 33 deletions(-)
diff --git a/mlir/lib/Analysis/Presburger/Barvinok.cpp b/mlir/lib/Analysis/Presburger/Barvinok.cpp
index 9eac53a96e9e0b..ca3edd68c82483 100644
--- a/mlir/lib/Analysis/Presburger/Barvinok.cpp
+++ b/mlir/lib/Analysis/Presburger/Barvinok.cpp
@@ -246,6 +246,28 @@ QuasiPolynomial mlir::presburger::detail::getCoefficientInRationalFunction(
return coefficients[power].simplify();
}
+static std::vector<Fraction> convolution(std::vector<Fraction> a,
+ std::vector<Fraction> b) {
+ // The length of the convolution is the maximum of the lengths
+ // of the two sequences. We pad the shorter one with zeroes.
+ unsigned convlen = std::max(a.size(), b.size());
+ for (unsigned k = a.size(); k < convlen; ++k)
+ a.push_back(0);
+ for (unsigned k = b.size(); k < convlen; ++k)
+ b.push_back(0);
+
+ std::vector<Fraction> convolution;
+ convolution.reserve(convlen);
+ convolution.clear();
+ for (unsigned k = 0; k < convlen; ++k) {
+ Fraction sum(0, 1);
+ for (unsigned l = 0; l <= k; ++l)
+ sum = sum + a[l] * b[k - l];
+ convolution.push_back(sum);
+ }
+ return convolution;
+}
+
/// We have a generating function of the form
/// f_p(x) = \sum_i sign_i * (x^n_i(p)) / (\prod_j (1 - x^d_{ij})
///
@@ -254,7 +276,7 @@ QuasiPolynomial mlir::presburger::detail::getCoefficientInRationalFunction(
/// d_{ij} \in Q^d are vectors.
///
/// We need to find the number of terms of the form x^t in the expansion of
-/// this function, for which we substitute x = (1, ..., 1).
+/// this function.
/// However, direct substitution causes the denominator to become zero.
///
/// We therefore use the following procedure instead:
@@ -264,10 +286,8 @@ QuasiPolynomial mlir::presburger::detail::getCoefficientInRationalFunction(
/// polynomials. P has coefficients as quasipolynomials in d parameters, while
/// Q has coefficients as scalars.
/// 3. Find the constant term in the expansion of each term P(s)/Q(s). This is
-/// equivalent to substituting s = 0, which by step 1's substitution is
-/// equivalent to letting x = (1, ..., 1).
-/// In this step, we cancel the factor with root zero from the numerator and
-/// denominator, thus preventing the denominator from becoming zero.
+/// equivalent to substituting s = 0.
+///
/// Step (1) We need to find a μ_i such that we can substitute x_i =
/// (s+1)^μ_i. After this substitution, the exponent of (s+1) in the
/// denominator is (μ_i • d_{ij}) in each term. Clearly, this cannot become
@@ -279,7 +299,8 @@ QuasiPolynomial mlir::presburger::detail::getCoefficientInRationalFunction(
/// sign_i * (s+1)^n_i / (\prod_j (1 - (s+1)^d'_{ij}))
/// For the i'th term, we first convert all the d'_{ij} to their
/// absolute values by multiplying and dividing by (s+1)^(-d'_{ij}) if it is
-/// negative. We change the sign accordingly.
+/// negative. We change the sign accordingly to keep the denominator in the
+/// same form.
/// Then, we replace each (1 - (s+1)^(d'_{ij})) with
/// (-s)(\sum_{0 ≤ k < d'_{ij}} (s+1)^k).
/// Thus the term overall has now the form
@@ -299,7 +320,6 @@ QuasiPolynomial mlir::presburger::detail::getCoefficientInRationalFunction(
/// Verdoolaege, Sven, et al. "Counting integer points in parametric
/// polytopes using Barvinok's rational functions." Algorithmica 48 (2007):
/// 37-66.
-
QuasiPolynomial
mlir::presburger::detail::computeNumTerms(const GeneratingFunction &gf) {
std::vector<Point> allDenominators;
@@ -314,7 +334,6 @@ mlir::presburger::detail::computeNumTerms(const GeneratingFunction &gf) {
QuasiPolynomial totalTerm(numParams, 0);
for (unsigned i = 0; i < numTerms; ++i) {
int sign = gf.getSigns()[i];
- ParamPoint v = gf.getNumerators()[i];
std::vector<Point> ds = gf.getDenominators()[i];
// Substitute x_i = (s+1)^μ_i
@@ -332,10 +351,11 @@ mlir::presburger::detail::computeNumTerms(const GeneratingFunction &gf) {
// and whose affine fn is a single floor expression, given by the
// corresponding column of v.
+ ParamPoint vTranspose = gf.getNumerators()[i].transpose();
std::vector<std::vector<SmallVector<Fraction>>> affine;
affine.reserve(numDims);
for (unsigned j = 0; j < numDims; ++j)
- affine.push_back({SmallVector<Fraction>(v.transpose().getRow(j))});
+ affine.push_back({SmallVector<Fraction>(vTranspose.getRow(j))});
QuasiPolynomial num(numParams, coefficients, affine);
num = num.simplify();
@@ -426,28 +446,9 @@ mlir::presburger::detail::computeNumTerms(const GeneratingFunction &gf) {
// of all the terms.
std::vector<Fraction> denominatorCoefficients;
denominatorCoefficients = eachTermDenCoefficients[0];
- for (unsigned j = 1, e = eachTermDenCoefficients.size(); j < e; ++j) {
- // The length of the convolution is the maximum of the lengths
- // of the two sequences. We pad the shorter one with zeroes.
- unsigned convlen = std::max(denominatorCoefficients.size(),
- eachTermDenCoefficients[j].size());
- for (unsigned k = denominatorCoefficients.size(); k < convlen; ++k)
- denominatorCoefficients.push_back(0);
- for (unsigned k = eachTermDenCoefficients[j].size(); k < convlen; ++k)
- eachTermDenCoefficients[j].push_back(0);
-
- std::vector<Fraction> convolution;
- convolution.reserve(convlen);
- convolution.clear();
- for (unsigned k = 0; k < convlen; ++k) {
- Fraction sum(0, 1);
- for (unsigned l = 0; l <= k; ++l)
- sum = sum +
- denominatorCoefficients[l] * eachTermDenCoefficients[j][k - l];
- convolution.push_back(sum);
- }
- denominatorCoefficients = convolution;
- }
+ for (unsigned j = 1, e = eachTermDenCoefficients.size(); j < e; ++j)
+ denominatorCoefficients =
+ convolution(denominatorCoefficients, eachTermDenCoefficients[j]);
totalTerm =
totalTerm + getCoefficientInRationalFunction(r, numeratorCoefficients,
diff --git a/mlir/unittests/Analysis/Presburger/BarvinokTest.cpp b/mlir/unittests/Analysis/Presburger/BarvinokTest.cpp
index 2842fc14eb8298..82af731b54f71a 100644
--- a/mlir/unittests/Analysis/Presburger/BarvinokTest.cpp
+++ b/mlir/unittests/Analysis/Presburger/BarvinokTest.cpp
@@ -127,7 +127,7 @@ TEST(BarvinokTest, getCoefficientInRationalFunction) {
// The following test is taken from
//
-TEST(BarvinokTest, substituteWithUnitVector) {
+TEST(BarvinokTest, computeNumTerms) {
GeneratingFunction gf(
1, {1, 1, 1},
{makeFracMatrix(2, 2, {{0, Fraction(1, 2)}, {0, 0}}),
@@ -135,7 +135,7 @@ TEST(BarvinokTest, substituteWithUnitVector) {
makeFracMatrix(2, 2, {{0, 0}, {0, 0}})},
{{{-1, 1}, {-1, 0}}, {{1, -1}, {0, -1}}, {{1, 0}, {0, 1}}});
- QuasiPolynomial numPoints = substituteWithUnitVector(gf);
+ QuasiPolynomial numPoints = computeNumTerms(gf);
// First, we make sure that all the affine functions are of the form ⌊p/2⌋.
for (const std::vector<SmallVector<Fraction>> &term : numPoints.getAffine()) {
>From aad6ca25ac2ac56091dce748a0e980841397b7a4 Mon Sep 17 00:00:00 2001
From: Abhinav271828 <abhinav.m at research.iiit.ac.in>
Date: Sun, 14 Jan 2024 23:38:25 +0530
Subject: [PATCH 10/24] Abstract out mu substitution
---
mlir/lib/Analysis/Presburger/Barvinok.cpp | 59 +++++++++++++----------
1 file changed, 34 insertions(+), 25 deletions(-)
diff --git a/mlir/lib/Analysis/Presburger/Barvinok.cpp b/mlir/lib/Analysis/Presburger/Barvinok.cpp
index ca3edd68c82483..92e942bc42430c 100644
--- a/mlir/lib/Analysis/Presburger/Barvinok.cpp
+++ b/mlir/lib/Analysis/Presburger/Barvinok.cpp
@@ -268,6 +268,38 @@ static std::vector<Fraction> convolution(std::vector<Fraction> a,
return convolution;
}
+/// Substitute x_i = (s+1)^μ_i in one term of a generating function,
+/// returning a quasipolynomial which represents the exponent of the
+/// numerator of the result.
+QuasiPolynomial substituteMuInTerm(unsigned numParams, ParamPoint v,
+ std::vector<Point> ds, Point mu) {
+ unsigned numDims = mu.size();
+ // First, the exponent in the numerator becomes
+ // - (μ • u_1) * (floor(first col of v))
+ // - (μ • u_2) * (floor(second col of v)) - ...
+ // - (μ • u_d) * (floor(d'th col of v))
+ // So we store the negation of the dot produts.
+
+ // We have d terms, each of whose coefficient is the negative dot product,
+ SmallVector<Fraction> coefficients;
+ coefficients.reserve(numDims);
+ for (const Point &d : ds)
+ coefficients.push_back(-dotProduct(mu, d));
+
+ // Then, the affine fn is a single floor expression, given by the
+ // corresponding column of v.
+ ParamPoint vTranspose = v.transpose();
+ std::vector<std::vector<SmallVector<Fraction>>> affine;
+ affine.reserve(numDims);
+ for (unsigned j = 0; j < numDims; ++j)
+ affine.push_back({SmallVector<Fraction>(vTranspose.getRow(j))});
+
+ QuasiPolynomial num(numParams, coefficients, affine);
+ num = num.simplify();
+
+ return num;
+}
+
/// We have a generating function of the form
/// f_p(x) = \sum_i sign_i * (x^n_i(p)) / (\prod_j (1 - x^d_{ij})
///
@@ -328,7 +360,6 @@ mlir::presburger::detail::computeNumTerms(const GeneratingFunction &gf) {
Point mu = getNonOrthogonalVector(allDenominators);
unsigned numParams = gf.getNumParams();
- unsigned numDims = mu.size();
unsigned numTerms = gf.getDenominators().size();
QuasiPolynomial totalTerm(numParams, 0);
@@ -336,30 +367,8 @@ mlir::presburger::detail::computeNumTerms(const GeneratingFunction &gf) {
int sign = gf.getSigns()[i];
std::vector<Point> ds = gf.getDenominators()[i];
- // Substitute x_i = (s+1)^μ_i
- // Then the exponent in the numerator becomes
- // - (μ • u_1) * (floor(first col of v))
- // - (μ • u_2) * (floor(second col of v)) - ...
- // - (μ • u_d) * (floor(d'th col of v))
- // So we store the negation of the dot produts.
-
- // We have d terms, each of whose coefficient is the negative dot product,
- SmallVector<Fraction> coefficients;
- coefficients.reserve(numDims);
- for (const Point &d : ds)
- coefficients.push_back(-dotProduct(mu, d));
-
- // and whose affine fn is a single floor expression, given by the
- // corresponding column of v.
- ParamPoint vTranspose = gf.getNumerators()[i].transpose();
- std::vector<std::vector<SmallVector<Fraction>>> affine;
- affine.reserve(numDims);
- for (unsigned j = 0; j < numDims; ++j)
- affine.push_back({SmallVector<Fraction>(vTranspose.getRow(j))});
-
- QuasiPolynomial num(numParams, coefficients, affine);
- num = num.simplify();
-
+ QuasiPolynomial num =
+ substituteMuInTerm(numParams, gf.getNumerators()[i], ds, mu);
// Now the numerator is (s+1)^num.
std::vector<Fraction> dens;
>From 520576574b29719a6104ed2f1c6039dd32e05727 Mon Sep 17 00:00:00 2001
From: Abhinav271828 <abhinav.m at research.iiit.ac.in>
Date: Sun, 14 Jan 2024 23:54:20 +0530
Subject: [PATCH 11/24] Fix doc
---
mlir/lib/Analysis/Presburger/Barvinok.cpp | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/mlir/lib/Analysis/Presburger/Barvinok.cpp b/mlir/lib/Analysis/Presburger/Barvinok.cpp
index 92e942bc42430c..de78ab8e2493be 100644
--- a/mlir/lib/Analysis/Presburger/Barvinok.cpp
+++ b/mlir/lib/Analysis/Presburger/Barvinok.cpp
@@ -304,7 +304,8 @@ QuasiPolynomial substituteMuInTerm(unsigned numParams, ParamPoint v,
/// f_p(x) = \sum_i sign_i * (x^n_i(p)) / (\prod_j (1 - x^d_{ij})
///
/// where sign_i is ±1,
-/// n_i \in Q^p -> Q^d is a d-vector of affine functions on p parameters, and
+/// n_i \in Q^p -> Q^d is the sum of the vectors d_{ij}, weighted by the
+/// floors of d affine functions.
/// d_{ij} \in Q^d are vectors.
///
/// We need to find the number of terms of the form x^t in the expansion of
>From fc0dba7ffe164b8c03ac279b4affb32604c7e849 Mon Sep 17 00:00:00 2001
From: Abhinav271828 <abhinav.m at research.iiit.ac.in>
Date: Sun, 14 Jan 2024 23:58:22 +0530
Subject: [PATCH 12/24] Fix doc
---
mlir/include/mlir/Analysis/Presburger/Barvinok.h | 2 +-
mlir/lib/Analysis/Presburger/Barvinok.cpp | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/mlir/include/mlir/Analysis/Presburger/Barvinok.h b/mlir/include/mlir/Analysis/Presburger/Barvinok.h
index d72332d220d2b5..50d6742019161c 100644
--- a/mlir/include/mlir/Analysis/Presburger/Barvinok.h
+++ b/mlir/include/mlir/Analysis/Presburger/Barvinok.h
@@ -101,7 +101,7 @@ QuasiPolynomial getCoefficientInRationalFunction(unsigned power,
/// Find the number of terms in the generating function corresponding to
/// a polytope.
-QuasiPolynomial computeNumTerms(const GeneratingFunction& gf);
+QuasiPolynomial computeNumTerms(const GeneratingFunction &gf);
} // namespace detail
} // namespace presburger
diff --git a/mlir/lib/Analysis/Presburger/Barvinok.cpp b/mlir/lib/Analysis/Presburger/Barvinok.cpp
index de78ab8e2493be..809dfdca6376d9 100644
--- a/mlir/lib/Analysis/Presburger/Barvinok.cpp
+++ b/mlir/lib/Analysis/Presburger/Barvinok.cpp
@@ -314,7 +314,7 @@ QuasiPolynomial substituteMuInTerm(unsigned numParams, ParamPoint v,
///
/// We therefore use the following procedure instead:
/// 1. Substitute x_i = (s+1)^μ_i for some vector μ. This makes the generating
-/// a function of a scalar s.
+/// function a function of a scalar s.
/// 2. Write each term in this function as P(s)/Q(s), where P and Q are
/// polynomials. P has coefficients as quasipolynomials in d parameters, while
/// Q has coefficients as scalars.
>From d3e758e8a4fc261db7e5bf8b9007ed6f44430fd5 Mon Sep 17 00:00:00 2001
From: Abhinav271828 <abhinav.m at research.iiit.ac.in>
Date: Mon, 15 Jan 2024 00:01:31 +0530
Subject: [PATCH 13/24] Fix doc
---
mlir/lib/Analysis/Presburger/Barvinok.cpp | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/mlir/lib/Analysis/Presburger/Barvinok.cpp b/mlir/lib/Analysis/Presburger/Barvinok.cpp
index 809dfdca6376d9..497987676d4d3b 100644
--- a/mlir/lib/Analysis/Presburger/Barvinok.cpp
+++ b/mlir/lib/Analysis/Presburger/Barvinok.cpp
@@ -324,12 +324,12 @@ QuasiPolynomial substituteMuInTerm(unsigned numParams, ParamPoint v,
/// Step (1) We need to find a μ_i such that we can substitute x_i =
/// (s+1)^μ_i. After this substitution, the exponent of (s+1) in the
/// denominator is (μ_i • d_{ij}) in each term. Clearly, this cannot become
-/// zero. Thus we find a vector μ that is not orthogonal to any of the
+/// zero. Hence we find a vector μ that is not orthogonal to any of the
/// d_{ij} and substitute x accordingly.
///
/// Step (2) We need to express the terms in the function as quotients of
/// polynomials. Each term is now of the form
-/// sign_i * (s+1)^n_i / (\prod_j (1 - (s+1)^d'_{ij}))
+/// sign_i * (s+1)^n'_i / (\prod_j (1 - (s+1)^d'_{ij}))
/// For the i'th term, we first convert all the d'_{ij} to their
/// absolute values by multiplying and dividing by (s+1)^(-d'_{ij}) if it is
/// negative. We change the sign accordingly to keep the denominator in the
>From ac6b0388aa8d43c7b2426fc023a05f1efb64dcf3 Mon Sep 17 00:00:00 2001
From: Abhinav271828 <abhinav.m at research.iiit.ac.in>
Date: Mon, 15 Jan 2024 00:08:00 +0530
Subject: [PATCH 14/24] Fix doc
---
mlir/lib/Analysis/Presburger/Barvinok.cpp | 10 +++++-----
1 file changed, 5 insertions(+), 5 deletions(-)
diff --git a/mlir/lib/Analysis/Presburger/Barvinok.cpp b/mlir/lib/Analysis/Presburger/Barvinok.cpp
index 497987676d4d3b..cde2ce3dcd91c2 100644
--- a/mlir/lib/Analysis/Presburger/Barvinok.cpp
+++ b/mlir/lib/Analysis/Presburger/Barvinok.cpp
@@ -330,7 +330,8 @@ QuasiPolynomial substituteMuInTerm(unsigned numParams, ParamPoint v,
/// Step (2) We need to express the terms in the function as quotients of
/// polynomials. Each term is now of the form
/// sign_i * (s+1)^n'_i / (\prod_j (1 - (s+1)^d'_{ij}))
-/// For the i'th term, we first convert all the d'_{ij} to their
+/// For the i'th term, we first normalize the denominator to have only
+/// positive exponents. We convert all the d'_{ij} to their
/// absolute values by multiplying and dividing by (s+1)^(-d'_{ij}) if it is
/// negative. We change the sign accordingly to keep the denominator in the
/// same form.
@@ -366,7 +367,7 @@ mlir::presburger::detail::computeNumTerms(const GeneratingFunction &gf) {
QuasiPolynomial totalTerm(numParams, 0);
for (unsigned i = 0; i < numTerms; ++i) {
int sign = gf.getSigns()[i];
- std::vector<Point> ds = gf.getDenominators()[i];
+ const std::vector<Point> &ds = gf.getDenominators()[i];
QuasiPolynomial num =
substituteMuInTerm(numParams, gf.getNumerators()[i], ds, mu);
@@ -382,8 +383,7 @@ mlir::presburger::detail::computeNumTerms(const GeneratingFunction &gf) {
// (1 - (s+1)^dens.back())
// We track the number of exponents that are negative in the
- // denominator, and convert them to their absolute values
- // (see lines 356-66).
+ // denominator, and convert them to their absolute values.
unsigned numNegExps = 0;
Fraction sumNegExps(0, 1);
for (unsigned j = 0, e = dens.size(); j < e; ++j) {
@@ -410,7 +410,7 @@ mlir::presburger::detail::computeNumTerms(const GeneratingFunction &gf) {
sign = -sign;
num = num - QuasiPolynomial(numParams, sumNegExps);
- // Take all the (-s) out, from line 353.
+ // Take all the (-s) out.
unsigned r = dens.size();
if (r % 2 == 1)
sign = -sign;
>From 94fe37002b0685a60b9bcf93b0f91fb5927f3e6e Mon Sep 17 00:00:00 2001
From: Abhinav271828 <abhinav.m at research.iiit.ac.in>
Date: Mon, 15 Jan 2024 00:15:14 +0530
Subject: [PATCH 15/24] Fix build
---
mlir/lib/Analysis/Presburger/Barvinok.cpp | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/mlir/lib/Analysis/Presburger/Barvinok.cpp b/mlir/lib/Analysis/Presburger/Barvinok.cpp
index cde2ce3dcd91c2..0419e590d3481f 100644
--- a/mlir/lib/Analysis/Presburger/Barvinok.cpp
+++ b/mlir/lib/Analysis/Presburger/Barvinok.cpp
@@ -367,7 +367,7 @@ mlir::presburger::detail::computeNumTerms(const GeneratingFunction &gf) {
QuasiPolynomial totalTerm(numParams, 0);
for (unsigned i = 0; i < numTerms; ++i) {
int sign = gf.getSigns()[i];
- const std::vector<Point> &ds = gf.getDenominators()[i];
+ std::vector<Point> ds = gf.getDenominators()[i];
QuasiPolynomial num =
substituteMuInTerm(numParams, gf.getNumerators()[i], ds, mu);
>From fd7879a583f6fc1f4749ac30cf999fa27dca796e Mon Sep 17 00:00:00 2001
From: Abhinav271828 <abhinav.m at research.iiit.ac.in>
Date: Mon, 15 Jan 2024 00:42:32 +0530
Subject: [PATCH 16/24] Reorg
---
mlir/lib/Analysis/Presburger/Barvinok.cpp | 120 ++++++++++++----------
1 file changed, 65 insertions(+), 55 deletions(-)
diff --git a/mlir/lib/Analysis/Presburger/Barvinok.cpp b/mlir/lib/Analysis/Presburger/Barvinok.cpp
index 0419e590d3481f..7fdb5d44c0c6b7 100644
--- a/mlir/lib/Analysis/Presburger/Barvinok.cpp
+++ b/mlir/lib/Analysis/Presburger/Barvinok.cpp
@@ -269,16 +269,20 @@ static std::vector<Fraction> convolution(std::vector<Fraction> a,
}
/// Substitute x_i = (s+1)^μ_i in one term of a generating function,
-/// returning a quasipolynomial which represents the exponent of the
-/// numerator of the result.
-QuasiPolynomial substituteMuInTerm(unsigned numParams, ParamPoint v,
- std::vector<Point> ds, Point mu) {
+/// returning
+/// a quasipolynomial which represents the exponent of the numerator
+/// of the result, and
+/// a vector which represents the exponents of the denominator of the
+/// result.
+std::pair<QuasiPolynomial, std::vector<Fraction>>
+substituteMuInTerm(unsigned numParams, ParamPoint v, std::vector<Point> ds,
+ Point mu) {
unsigned numDims = mu.size();
// First, the exponent in the numerator becomes
// - (μ • u_1) * (floor(first col of v))
// - (μ • u_2) * (floor(second col of v)) - ...
// - (μ • u_d) * (floor(d'th col of v))
- // So we store the negation of the dot produts.
+ // So we store the negation of the dot products.
// We have d terms, each of whose coefficient is the negative dot product,
SmallVector<Fraction> coefficients;
@@ -297,7 +301,52 @@ QuasiPolynomial substituteMuInTerm(unsigned numParams, ParamPoint v,
QuasiPolynomial num(numParams, coefficients, affine);
num = num.simplify();
- return num;
+ std::vector<Fraction> dens;
+ dens.reserve(ds.size());
+ // Similarly, each term in the denominator has exponent
+ // given by the dot product of μ with u_i.
+ for (const Point &d : ds)
+ dens.push_back(dotProduct(d, mu));
+ // This term in the denominator is
+ // (1 - (s+1)^dens.back())
+
+ return std::make_pair(num, dens);
+}
+
+void normalizeDenominatorExponents(int &sign, QuasiPolynomial &num,
+ std::vector<Fraction> &dens) {
+ // We track the number of exponents that are negative in the
+ // denominator, and convert them to their absolute values.
+ unsigned numNegExps = 0;
+ Fraction sumNegExps(0, 1);
+ for (unsigned j = 0, e = dens.size(); j < e; ++j) {
+ if (dens[j] < 0) {
+ numNegExps += 1;
+ sumNegExps += dens[j];
+ }
+ // All exponents will be made positive; then reduce
+ // (1 - (s+1)^x)
+ // to
+ // (-s)*(Σ_{x-1} (s+1)^j) because x > 0
+ dens[j] = abs(dens[j]) - 1;
+ }
+
+ // If we have (1 - (s+1)^-c) in the denominator,
+ // multiply and divide by (s+1)^c.
+ // We convert all negative-exponent terms at once; therefore
+ // we multiply and divide by (s+1)^sumNegExps.
+ // Then we get
+ // -(1 - (s+1)^c) in the denominator,
+ // increase the numerator by c, and
+ // flip the sign.
+ if (numNegExps % 2 == 1)
+ sign = -sign;
+ num = num - QuasiPolynomial(num.getNumInputs(), sumNegExps);
+
+ // Take all the (-s) out.
+ unsigned r = dens.size();
+ if (r % 2 == 1)
+ sign = -sign;
}
/// We have a generating function of the form
@@ -362,58 +411,18 @@ mlir::presburger::detail::computeNumTerms(const GeneratingFunction &gf) {
Point mu = getNonOrthogonalVector(allDenominators);
unsigned numParams = gf.getNumParams();
- unsigned numTerms = gf.getDenominators().size();
+ const std::vector<std::vector<Point>> &ds = gf.getDenominators();
QuasiPolynomial totalTerm(numParams, 0);
- for (unsigned i = 0; i < numTerms; ++i) {
+ for (unsigned i = 0, e = ds.size(); i < e; ++i) {
int sign = gf.getSigns()[i];
- std::vector<Point> ds = gf.getDenominators()[i];
-
- QuasiPolynomial num =
- substituteMuInTerm(numParams, gf.getNumerators()[i], ds, mu);
- // Now the numerator is (s+1)^num.
-
- std::vector<Fraction> dens;
- dens.reserve(ds.size());
- // Similarly, each term in the denominator has exponent
- // given by the dot product of μ with u_i.
- for (const Point &d : ds)
- dens.push_back(dotProduct(d, mu));
- // This term in the denominator is
- // (1 - (s+1)^dens.back())
-
- // We track the number of exponents that are negative in the
- // denominator, and convert them to their absolute values.
- unsigned numNegExps = 0;
- Fraction sumNegExps(0, 1);
- for (unsigned j = 0, e = dens.size(); j < e; ++j) {
- if (dens[j] < 0) {
- numNegExps += 1;
- sumNegExps += dens[j];
- }
- // All exponents will be made positive; then reduce
- // (1 - (s+1)^x)
- // to
- // (-s)*(Σ_{x-1} (s+1)^j) because x > 0
- dens[j] = abs(dens[j]) - 1;
- }
- // If we have (1 - (s+1)^-c) in the denominator,
- // multiply and divide by (s+1)^c.
- // We convert all negative-exponent terms at once; therefore
- // we multiply and divide by (s+1)^sumNegExps.
- // Then we get
- // -(1 - (s+1)^c) in the denominator,
- // increase the numerator by c, and
- // flip the sign.
- if (numNegExps % 2 == 1)
- sign = -sign;
- num = num - QuasiPolynomial(numParams, sumNegExps);
-
- // Take all the (-s) out.
- unsigned r = dens.size();
- if (r % 2 == 1)
- sign = -sign;
+ auto [numExp, dens] =
+ substituteMuInTerm(numParams, gf.getNumerators()[i], ds[i], mu);
+ // Now the numerator is (s+1)^numExp
+ // and the denominator is \prod_j (1 - (s+1)^dens[j]).
+
+ normalizeDenominatorExponents(sign, numExp, dens);
// Now the expression is
// (s+1)^num /
@@ -423,6 +432,7 @@ mlir::presburger::detail::computeNumTerms(const GeneratingFunction &gf) {
// we need to find the coefficient of s^r in
// P(s)/Q(s).
+ unsigned r = dens.size();
// First, the coefficients of P(s), which are binomial coefficients.
// We need r+1 of these.
std::vector<QuasiPolynomial> numeratorCoefficients;
@@ -432,7 +442,7 @@ mlir::presburger::detail::computeNumTerms(const GeneratingFunction &gf) {
for (unsigned j = 1; j <= r; ++j)
numeratorCoefficients.push_back(
(numeratorCoefficients[j - 1] *
- (num - QuasiPolynomial(numParams, j - 1)) / Fraction(j, 1))
+ (numExp - QuasiPolynomial(numParams, j - 1)) / Fraction(j, 1))
.simplify());
// Coeff of s^j
>From 819ec57a60e99ced09624eb5d652c63763cb5d3f Mon Sep 17 00:00:00 2001
From: Abhinav271828 <abhinav.m at research.iiit.ac.in>
Date: Mon, 15 Jan 2024 00:43:09 +0530
Subject: [PATCH 17/24] Typo
---
mlir/lib/Analysis/Presburger/Barvinok.cpp | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/mlir/lib/Analysis/Presburger/Barvinok.cpp b/mlir/lib/Analysis/Presburger/Barvinok.cpp
index 7fdb5d44c0c6b7..85bdfda4a79abd 100644
--- a/mlir/lib/Analysis/Presburger/Barvinok.cpp
+++ b/mlir/lib/Analysis/Presburger/Barvinok.cpp
@@ -290,7 +290,7 @@ substituteMuInTerm(unsigned numParams, ParamPoint v, std::vector<Point> ds,
for (const Point &d : ds)
coefficients.push_back(-dotProduct(mu, d));
- // Then, the affine fn is a single floor expression, given by the
+ // Then, the affine function is a single floor expression, given by the
// corresponding column of v.
ParamPoint vTranspose = v.transpose();
std::vector<std::vector<SmallVector<Fraction>>> affine;
>From eeb020c739eb5aff0f77bbc39985bbd6003ef54c Mon Sep 17 00:00:00 2001
From: Abhinav271828 <abhinav.m at research.iiit.ac.in>
Date: Mon, 15 Jan 2024 00:54:33 +0530
Subject: [PATCH 18/24] Reorg doc
---
mlir/lib/Analysis/Presburger/Barvinok.cpp | 74 +++++++++++------------
1 file changed, 36 insertions(+), 38 deletions(-)
diff --git a/mlir/lib/Analysis/Presburger/Barvinok.cpp b/mlir/lib/Analysis/Presburger/Barvinok.cpp
index 85bdfda4a79abd..a1cfdd8114c13a 100644
--- a/mlir/lib/Analysis/Presburger/Barvinok.cpp
+++ b/mlir/lib/Analysis/Presburger/Barvinok.cpp
@@ -313,6 +313,9 @@ substituteMuInTerm(unsigned numParams, ParamPoint v, std::vector<Point> ds,
return std::make_pair(num, dens);
}
+/// Normalize all denominator exponents `dens` to their absolute values
+/// by multiplying and dividing by the inverses.
+/// Also, take the factors (-s) out of each term of the product.
void normalizeDenominatorExponents(int &sign, QuasiPolynomial &num,
std::vector<Fraction> &dens) {
// We track the number of exponents that are negative in the
@@ -370,41 +373,16 @@ void normalizeDenominatorExponents(int &sign, QuasiPolynomial &num,
/// 3. Find the constant term in the expansion of each term P(s)/Q(s). This is
/// equivalent to substituting s = 0.
///
-/// Step (1) We need to find a μ_i such that we can substitute x_i =
-/// (s+1)^μ_i. After this substitution, the exponent of (s+1) in the
-/// denominator is (μ_i • d_{ij}) in each term. Clearly, this cannot become
-/// zero. Hence we find a vector μ that is not orthogonal to any of the
-/// d_{ij} and substitute x accordingly.
-///
-/// Step (2) We need to express the terms in the function as quotients of
-/// polynomials. Each term is now of the form
-/// sign_i * (s+1)^n'_i / (\prod_j (1 - (s+1)^d'_{ij}))
-/// For the i'th term, we first normalize the denominator to have only
-/// positive exponents. We convert all the d'_{ij} to their
-/// absolute values by multiplying and dividing by (s+1)^(-d'_{ij}) if it is
-/// negative. We change the sign accordingly to keep the denominator in the
-/// same form.
-/// Then, we replace each (1 - (s+1)^(d'_{ij})) with
-/// (-s)(\sum_{0 ≤ k < d'_{ij}} (s+1)^k).
-/// Thus the term overall has now the form
-/// sign'_i * (s+1)^n'_i / (s^r * \prod_j (\sum_k (s+1)^k)).
-/// This means that
-/// the numerator is a polynomial in s, with coefficients as
-/// quasipolynomials (given by binomial coefficients), and the denominator
-/// is polynomial in s, with fractional coefficients (given by taking the
-/// convolution over all j).
-///
-/// Step (3) We need to find the constant term in the expansion of each
-/// term. Since each term has s^r as a factor in the denominator, we avoid
-/// substituting s = 0 directly; instead, we find the coefficient of s^r in
-/// sign'_i * (s+1)^n'_i / (\prod_j (\sum_k (s+1)^k)),
-/// for which we use the `getCoefficientInRationalFunction()` function.
-///
/// Verdoolaege, Sven, et al. "Counting integer points in parametric
/// polytopes using Barvinok's rational functions." Algorithmica 48 (2007):
/// 37-66.
QuasiPolynomial
mlir::presburger::detail::computeNumTerms(const GeneratingFunction &gf) {
+ // Step (1) We need to find a μ such that we can substitute x_i =
+ // (s+1)^μ_i. After this substitution, the exponent of (s+1) in the
+ // denominator is (μ_i • d_{ij}) in each term. Clearly, this cannot become
+ // zero. Hence we find a vector μ that is not orthogonal to any of the
+ // d_{ij} and substitute x accordingly.
std::vector<Point> allDenominators;
for (ArrayRef<Point> den : gf.getDenominators())
allDenominators.insert(allDenominators.end(), den.begin(), den.end());
@@ -417,22 +395,42 @@ mlir::presburger::detail::computeNumTerms(const GeneratingFunction &gf) {
for (unsigned i = 0, e = ds.size(); i < e; ++i) {
int sign = gf.getSigns()[i];
+ // Compute the new numerator and denominator after substituting μ.
auto [numExp, dens] =
substituteMuInTerm(numParams, gf.getNumerators()[i], ds[i], mu);
// Now the numerator is (s+1)^numExp
// and the denominator is \prod_j (1 - (s+1)^dens[j]).
+ // Step (2) We need to express the terms in the function as quotients of
+ // polynomials. Each term is now of the form
+ // sign_i * (s+1)^numExp / (\prod_j (1 - (s+1)^dens[j]))
+ // For the i'th term, we first normalize the denominator to have only
+ // positive exponents. We convert all the dens[j] to their
+ // absolute values by multiplying and dividing by (s+1)^(-dens[j]) if it is
+ // negative. We change the sign accordingly to keep the denominator in the
+ // same form.
+ // Then, we replace each (1 - (s+1)^(dens[j])) with
+ // (-s)(\sum_{0 ≤ k < dens[j]} (s+1)^k).
normalizeDenominatorExponents(sign, numExp, dens);
- // Now the expression is
- // (s+1)^num /
- // (s^r * Π_(0 ≤ i < r) (Σ_{0 ≤ j ≤ dens[i]} (s+1)^j))
-
- // Letting P(s) = (s+1)^num and Q(s) = Π_r (...),
- // we need to find the coefficient of s^r in
- // P(s)/Q(s).
-
+ // Thus the term overall has now the form
+ // sign'_i * (s+1)^numExp /
+ // (s^r * \prod_j (\sum_{0 ≤ k < dens[j]} (s+1)^k)).
+ // This means that
+ // the numerator is a polynomial in s, with coefficients as
+ // quasipolynomials (given by binomial coefficients), and the denominator
+ // is polynomial in s, with fractional coefficients (given by taking the
+ // convolution over all j).
+
+ // Step (3) We need to find the constant term in the expansion of each
+ // term. Since each term has s^r as a factor in the denominator, we avoid
+ // substituting s = 0 directly; instead, we find the coefficient of s^r in
+ // sign'_i * (s+1)^numExp / (\prod_j (\sum_k (s+1)^k)),
+ // Letting P(s) = (s+1)^numExp and Q(s) = \prod_j (...),
+ // we need to find the coefficient of s^r in P(s)/Q(s),
+ // for which we use the `getCoefficientInRationalFunction()` function.
unsigned r = dens.size();
+
// First, the coefficients of P(s), which are binomial coefficients.
// We need r+1 of these.
std::vector<QuasiPolynomial> numeratorCoefficients;
>From 0bda7729d154a2f4232f9926d4905b5007bdbd96 Mon Sep 17 00:00:00 2001
From: Abhinav271828 <abhinav.m at research.iiit.ac.in>
Date: Mon, 15 Jan 2024 01:24:05 +0530
Subject: [PATCH 19/24] Abstract out binomial coefficient computation
---
mlir/lib/Analysis/Presburger/Barvinok.cpp | 61 +++++++++++++++--------
1 file changed, 40 insertions(+), 21 deletions(-)
diff --git a/mlir/lib/Analysis/Presburger/Barvinok.cpp b/mlir/lib/Analysis/Presburger/Barvinok.cpp
index a1cfdd8114c13a..cf42b9494c8bc1 100644
--- a/mlir/lib/Analysis/Presburger/Barvinok.cpp
+++ b/mlir/lib/Analysis/Presburger/Barvinok.cpp
@@ -352,6 +352,34 @@ void normalizeDenominatorExponents(int &sign, QuasiPolynomial &num,
sign = -sign;
}
+/// Compute the binomial coefficients nCi for 0 ≤ i ≤ r,
+/// where n is a QuasiPolynomial.
+std::vector<QuasiPolynomial> getBinomialCoefficients(QuasiPolynomial n,
+ unsigned r) {
+ unsigned numParams = n.getNumInputs();
+ std::vector<QuasiPolynomial> coefficients;
+ coefficients.reserve(r + 1);
+ coefficients.push_back(QuasiPolynomial(numParams, 1));
+ for (unsigned j = 1; j <= r; ++j)
+ // We use the recursive formula for binomial coefficients here and below.
+ coefficients.push_back(
+ (coefficients[j - 1] * (n - QuasiPolynomial(numParams, j - 1)) /
+ Fraction(j, 1))
+ .simplify());
+ return coefficients;
+}
+
+/// Compute the binomial coefficients nCi for 0 ≤ i ≤ r,
+/// where n is a QuasiPolynomial.
+std::vector<Fraction> getBinomialCoefficients(Fraction n, Fraction r) {
+ std::vector<Fraction> coefficients;
+ coefficients.reserve((int64_t)floor(r));
+ coefficients.push_back(1);
+ for (unsigned j = 1; j <= r; ++j)
+ coefficients.push_back(coefficients[j - 1] * (n - (j - 1)) / (j));
+ return coefficients;
+}
+
/// We have a generating function of the form
/// f_p(x) = \sum_i sign_i * (x^n_i(p)) / (\prod_j (1 - x^d_{ij})
///
@@ -409,7 +437,8 @@ mlir::presburger::detail::computeNumTerms(const GeneratingFunction &gf) {
// absolute values by multiplying and dividing by (s+1)^(-dens[j]) if it is
// negative. We change the sign accordingly to keep the denominator in the
// same form.
- // Then, we replace each (1 - (s+1)^(dens[j])) with
+ // Then, using the formula for geometric series, we replace each (1 -
+ // (s+1)^(dens[j])) with
// (-s)(\sum_{0 ≤ k < dens[j]} (s+1)^k).
normalizeDenominatorExponents(sign, numExp, dens);
@@ -431,32 +460,22 @@ mlir::presburger::detail::computeNumTerms(const GeneratingFunction &gf) {
// for which we use the `getCoefficientInRationalFunction()` function.
unsigned r = dens.size();
- // First, the coefficients of P(s), which are binomial coefficients.
- // We need r+1 of these.
- std::vector<QuasiPolynomial> numeratorCoefficients;
- numeratorCoefficients.reserve(r + 1);
- numeratorCoefficients.push_back(
- QuasiPolynomial(numParams, 1)); // Coeff of s^0
- for (unsigned j = 1; j <= r; ++j)
- numeratorCoefficients.push_back(
- (numeratorCoefficients[j - 1] *
- (numExp - QuasiPolynomial(numParams, j - 1)) / Fraction(j, 1))
- .simplify());
- // Coeff of s^j
+ // First, we compute the coefficients of P(s),
+ // which are binomial coefficients.
+ // We only need the first r+1 of these, as higher-order terms do not
+ // contribute to the coefficient of s^r.
+ std::vector<QuasiPolynomial> numeratorCoefficients =
+ getBinomialCoefficients(numExp, r);
// Then the coefficients of each individual term in Q(s),
- // which are (di+1) C (k+1) for 0 ≤ k ≤ di
+ // which are (di+1) C (k+1) for 0 ≤ k ≤ di.
std::vector<std::vector<Fraction>> eachTermDenCoefficients;
std::vector<Fraction> singleTermDenCoefficients;
eachTermDenCoefficients.reserve(r);
for (const Fraction &den : dens) {
- singleTermDenCoefficients.clear();
- singleTermDenCoefficients.push_back(den + 1);
- for (unsigned j = 1; j <= den; ++j)
- singleTermDenCoefficients.push_back(singleTermDenCoefficients[j - 1] *
- (den - (j - 1)) / (j + 1));
-
- eachTermDenCoefficients.push_back(singleTermDenCoefficients);
+ singleTermDenCoefficients = getBinomialCoefficients(den + 1, den + 1);
+ eachTermDenCoefficients.push_back(
+ ArrayRef<Fraction>(singleTermDenCoefficients).slice(1));
}
// Now we find the coefficients in Q(s) itself
>From e811dfda57ef3e389d9e869a5425c240cb2a8600 Mon Sep 17 00:00:00 2001
From: Abhinav271828 <abhinav.m at research.iiit.ac.in>
Date: Mon, 15 Jan 2024 01:29:23 +0530
Subject: [PATCH 20/24] Fix conv
---
mlir/lib/Analysis/Presburger/Barvinok.cpp | 15 ++++++---------
1 file changed, 6 insertions(+), 9 deletions(-)
diff --git a/mlir/lib/Analysis/Presburger/Barvinok.cpp b/mlir/lib/Analysis/Presburger/Barvinok.cpp
index cf42b9494c8bc1..f8b3bd6226ed19 100644
--- a/mlir/lib/Analysis/Presburger/Barvinok.cpp
+++ b/mlir/lib/Analysis/Presburger/Barvinok.cpp
@@ -250,19 +250,16 @@ static std::vector<Fraction> convolution(std::vector<Fraction> a,
std::vector<Fraction> b) {
// The length of the convolution is the maximum of the lengths
// of the two sequences. We pad the shorter one with zeroes.
- unsigned convlen = std::max(a.size(), b.size());
- for (unsigned k = a.size(); k < convlen; ++k)
- a.push_back(0);
- for (unsigned k = b.size(); k < convlen; ++k)
- b.push_back(0);
+ unsigned len = std::max(a.size(), b.size());
+ a.resize(len);
+ b.resize(len);
std::vector<Fraction> convolution;
- convolution.reserve(convlen);
- convolution.clear();
- for (unsigned k = 0; k < convlen; ++k) {
+ convolution.reserve(len);
+ for (unsigned k = 0; k < len; ++k) {
Fraction sum(0, 1);
for (unsigned l = 0; l <= k; ++l)
- sum = sum + a[l] * b[k - l];
+ sum += a[l] * b[k - l];
convolution.push_back(sum);
}
return convolution;
>From 3b40fb5243694f162b9ca20e7c715a1bf55b8af7 Mon Sep 17 00:00:00 2001
From: Abhinav271828 <abhinav.m at research.iiit.ac.in>
Date: Mon, 15 Jan 2024 01:31:47 +0530
Subject: [PATCH 21/24] Fix conv
---
mlir/lib/Analysis/Presburger/Barvinok.cpp | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/mlir/lib/Analysis/Presburger/Barvinok.cpp b/mlir/lib/Analysis/Presburger/Barvinok.cpp
index f8b3bd6226ed19..cbc5d30a29dab1 100644
--- a/mlir/lib/Analysis/Presburger/Barvinok.cpp
+++ b/mlir/lib/Analysis/Presburger/Barvinok.cpp
@@ -250,7 +250,7 @@ static std::vector<Fraction> convolution(std::vector<Fraction> a,
std::vector<Fraction> b) {
// The length of the convolution is the maximum of the lengths
// of the two sequences. We pad the shorter one with zeroes.
- unsigned len = std::max(a.size(), b.size());
+ unsigned len = a.size() + b.size();
a.resize(len);
b.resize(len);
>From 0ef21593f729cbce801d3682ae1ce710bd47501f Mon Sep 17 00:00:00 2001
From: Abhinav271828 <abhinav.m at research.iiit.ac.in>
Date: Mon, 15 Jan 2024 01:33:50 +0530
Subject: [PATCH 22/24] Fix doc
---
mlir/lib/Analysis/Presburger/Barvinok.cpp | 12 ++++++++----
1 file changed, 8 insertions(+), 4 deletions(-)
diff --git a/mlir/lib/Analysis/Presburger/Barvinok.cpp b/mlir/lib/Analysis/Presburger/Barvinok.cpp
index cbc5d30a29dab1..f9cda3f34c9291 100644
--- a/mlir/lib/Analysis/Presburger/Barvinok.cpp
+++ b/mlir/lib/Analysis/Presburger/Barvinok.cpp
@@ -271,6 +271,9 @@ static std::vector<Fraction> convolution(std::vector<Fraction> a,
/// of the result, and
/// a vector which represents the exponents of the denominator of the
/// result.
+/// v represents the affine functions whose floors are multiplied by
+/// by the generators, and
+/// ds represents the list of generators.
std::pair<QuasiPolynomial, std::vector<Fraction>>
substituteMuInTerm(unsigned numParams, ParamPoint v, std::vector<Point> ds,
Point mu) {
@@ -302,12 +305,13 @@ substituteMuInTerm(unsigned numParams, ParamPoint v, std::vector<Point> ds,
dens.reserve(ds.size());
// Similarly, each term in the denominator has exponent
// given by the dot product of μ with u_i.
- for (const Point &d : ds)
+ for (const Point &d : ds) {
+ // This term in the denominator is
+ // (1 - (s+1)^dens.back())
dens.push_back(dotProduct(d, mu));
- // This term in the denominator is
- // (1 - (s+1)^dens.back())
+ }
- return std::make_pair(num, dens);
+ return {num, dens};
}
/// Normalize all denominator exponents `dens` to their absolute values
>From f20bb63f988dc12b8b1243aa636f44c47adcaa67 Mon Sep 17 00:00:00 2001
From: Abhinav271828 <abhinav.m at research.iiit.ac.in>
Date: Mon, 15 Jan 2024 01:35:03 +0530
Subject: [PATCH 23/24] Fix doc
---
mlir/unittests/Analysis/Presburger/BarvinokTest.cpp | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/mlir/unittests/Analysis/Presburger/BarvinokTest.cpp b/mlir/unittests/Analysis/Presburger/BarvinokTest.cpp
index 82af731b54f71a..b0948334740fe6 100644
--- a/mlir/unittests/Analysis/Presburger/BarvinokTest.cpp
+++ b/mlir/unittests/Analysis/Presburger/BarvinokTest.cpp
@@ -126,7 +126,9 @@ TEST(BarvinokTest, getCoefficientInRationalFunction) {
}
// The following test is taken from
-//
+/// Verdoolaege, Sven, et al. "Counting integer points in parametric
+/// polytopes using Barvinok's rational functions." Algorithmica 48 (2007):
+/// 37-66.
TEST(BarvinokTest, computeNumTerms) {
GeneratingFunction gf(
1, {1, 1, 1},
>From 98ae71e2649a31ba619f773b7b81a2e148c65a5e Mon Sep 17 00:00:00 2001
From: Abhinav271828 <abhinav.m at research.iiit.ac.in>
Date: Mon, 15 Jan 2024 01:36:28 +0530
Subject: [PATCH 24/24] Fix doc
---
mlir/lib/Analysis/Presburger/Barvinok.cpp | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/mlir/lib/Analysis/Presburger/Barvinok.cpp b/mlir/lib/Analysis/Presburger/Barvinok.cpp
index f9cda3f34c9291..969c327ab60fa8 100644
--- a/mlir/lib/Analysis/Presburger/Barvinok.cpp
+++ b/mlir/lib/Analysis/Presburger/Barvinok.cpp
@@ -469,7 +469,7 @@ mlir::presburger::detail::computeNumTerms(const GeneratingFunction &gf) {
getBinomialCoefficients(numExp, r);
// Then the coefficients of each individual term in Q(s),
- // which are (di+1) C (k+1) for 0 ≤ k ≤ di.
+ // which are (dens[i]+1) C (k+1) for 0 ≤ k ≤ dens[i].
std::vector<std::vector<Fraction>> eachTermDenCoefficients;
std::vector<Fraction> singleTermDenCoefficients;
eachTermDenCoefficients.reserve(r);
More information about the Mlir-commits
mailing list