Spintracing Related Issues

[ajay-mk]

To reproduce, try the following (term from perturbed T amplitudes in LR-CCSD):

  auto expr1 = sequant::parse_expr(L"-1 ω * A{i1,i2;a1,a2} * t{a1,a2;i1,i2}");
  auto st_expr1 = mbpt::closed_shell_CC_spintrace(expr1);
  std::wcout << "ST Expr1: " << st_expr1.to_latex() << "\n";

1. closed_shell_CC_spintrace methods expect Sum only
   

   SeQuant/SeQuant/domain/mbpt/spin.cpp

   Lines 1133 to 1136 in af78423

   	ExprPtr closed_shell_CC_spintrace_v2(ExprPtr const& expr,
   	ClosedShellCCSpintraceOptions options) {
   	SEQUANT_ASSERT(options.method == BiorthogonalizationMethod::V2);
   	SEQUANT_ASSERT(expr->is<Sum>());

2. symmetrize_expr expects the first factor to be the A tensor. So it doesn't work when ω is present.
   

   SeQuant/SeQuant/domain/mbpt/spin.cpp

   Lines 602 to 606 in af78423

   	ExprPtr symmetrize_expr(const ProductPtr& product) {
   	auto result = std::make_shared<Sum>();
   	// Assumes canonical sequence of tensors in the product
   	if (product->factor(0)->as<Tensor>().label() != L"A") return product;

[ABesharat]

The issue was addressed in PR #441