baram2584/FESADev
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

perf(hex8): reuse identical paired Hessian subexpressions per trial

Browse Source
This commit is contained in:
KOKO\Mimi
2026-10-14 17:14:05 +09:00
parent 1d77e6be83
commit 6d709fd572
1 changed files with 22 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/fesa/elements/hex8_high_low.cpp
+22 -5
View File
@@ -204,6 +204,18 @@ Result<Hex8HighLowResponse> Hex8Element::EvaluateHighLow(
for (std::size_t a = 0; a < 8; ++a) for (std::size_t a = 0; a < 8; ++a)
for (std::size_t i = 0; i < 3; ++i) for (std::size_t i = 0; i < 3; ++i)
integrated_fg[a][i] += fg[a][i] * HighLow{point.weight}; integrated_fg[a][i] += fg[a][i] * HighLow{point.weight};
// Reuse identical first products within this trial point only. The second
// multiplication and the original left association remain unchanged.
std::array<Triple, 8> deviatoric_fg{}, shear_fg{};
Tensor shear_fft{};
for (std::size_t a = 0; a < 8; ++a)
for (std::size_t i = 0; i < 3; ++i) {
deviatoric_fg[a][i] = deviatoric_coefficient * fg[a][i];
shear_fg[a][i] = HighLow{mu_} * fg[a][i];
}
for (std::size_t i = 0; i < 3; ++i)
for (std::size_t k = 0; k < 3; ++k)
shear_fft[i][k] = HighLow{mu_} * fft[i][k];
for (std::size_t a = 0; a < 8; ++a) for (std::size_t a = 0; a < 8; ++a)
for (std::size_t b = 0; b < 8; ++b) { for (std::size_t b = 0; b < 8; ++b) {
const auto dot = Dot(gradients[a], gradients[b]); const auto dot = Dot(gradients[a], gradients[b]);
@@ -216,22 +228,27 @@ Result<Hex8HighLowResponse> Hex8Element::EvaluateHighLow(
for (std::size_t k = 0; k < 3; ++k) { for (std::size_t k = 0; k < 3; ++k) {
const auto row = 3U * a + i; const auto row = 3U * a + i;
const auto column = 3U * b + k; const auto column = 3U * b + k;
const auto mat = deviatoric_coefficient * fg[a][i] * fg[b][k] + const auto mat = deviatoric_fg[a][i] * fg[b][k] +
HighLow{mu_} * fg[b][i] * fg[a][k] + shear_fg[b][i] * fg[a][k] +
HighLow{mu_} * fft[i][k] * dot; shear_fft[i][k] * dot;
material[row][column] += mat * HighLow{point.weight}; material[row][column] += mat * HighLow{point.weight};
if (i == k) if (i == k)
geometric[row][column] += geo * HighLow{point.weight}; geometric[row][column] += geo * HighLow{point.weight};
} }
} }
} }
if (nonlinear || !paired_linear_tangent_ready_) if (nonlinear || !paired_linear_tangent_ready_) {
std::array<Triple, 8> bulk_fg{};
for (std::size_t a = 0; a < 8; ++a)
for (std::size_t i = 0; i < 3; ++i)
bulk_fg[a][i] = bulk_coefficient * integrated_fg[a][i];
for (std::size_t a = 0; a < 8; ++a) for (std::size_t a = 0; a < 8; ++a)
for (std::size_t b = 0; b < 8; ++b) for (std::size_t b = 0; b < 8; ++b)
for (std::size_t i = 0; i < 3; ++i) for (std::size_t i = 0; i < 3; ++i)
for (std::size_t k = 0; k < 3; ++k) for (std::size_t k = 0; k < 3; ++k)
material[3 * a + i][3 * b + k] += material[3 * a + i][3 * b + k] +=
bulk_coefficient * integrated_fg[a][i] * integrated_fg[b][k]; bulk_fg[a][i] * integrated_fg[b][k];
}
if (!IsNormalized(energy)) if (!IsNormalized(energy))
return Result<Hex8HighLowResponse>::Failure( return Result<Hex8HighLowResponse>::Failure(
Failure("Hex8 high+low energy is unsupported or nonfinite.")); Failure("Hex8 high+low energy is unsupported or nonfinite."));