Segfault in closed_shell_CC_spintrace #426
Description
I am trying to run the following
#include<SeQuant/core/context.hpp>
#include<SeQuant/core/op.hpp>
#include<SeQuant/domain/mbpt/convention.hpp>
#include<SeQuant/domain/mbpt/models/cc.hpp>
#include<SeQuant/domain/mbpt/spin.hpp>
using namespace sequant;
using namespace sequant::mbpt;
int main() {
set_default_context({.index_space_registry_shared_ptr = make_min_sr_spaces(),
.vacuum = Vacuum::SingleProduct});
auto t_eqs = CC{2}.t();
auto t2_eq = t_eqs[2];
auto t2_rhf = closed_shell_CC_spintrace(t2_eq);
std::wcout << "T2 (RHF): " << to_latex(t2_rhf) << "\n";
return 0;
}
but it is segfaulting seemingly in this call. If I change the build to Debug, it fails an assertion with
Assertion failed: (this->m_holder.m_size > n), function operator[], file vector.hpp, line 1742.
Compilation:
-- [range-v3]: C++ std=2a
-- std::variant uses shared storage: TRUE
-- PV_EXPLOIT_SHARED_STORAGE: ON
-- Fetching and building dependencies...
-- Done fetching dependencies
-- Skipping configuration of libPerm-specific LTO setup when built as a static library -> Global CMAKE_INTERPROCEDURAL_OPTIMIZATION takes effect.
-- Build spdlog: 1.15.3
-- Build type: Release
-- Using the multi-header code from /Users/oliver.backhouse/git/sequant/build/_deps/nlohmann_json-src/include/
-- Google Benchmark version: v1.9.4, normalized to 1.9.4
-- Enabling additional flags: -DINCLUDE_DIRECTORIES=/Users/oliver.backhouse/git/sequant/build/_deps/googlebenchmark-src/include
-- Performing Test HAVE_THREAD_SAFETY_ATTRIBUTES -- success
-- Performing Test HAVE_STD_REGEX -- success
-- Performing Test HAVE_GNU_POSIX_REGEX -- failed to compile
-- Performing Test HAVE_POSIX_REGEX -- success
-- Performing Test HAVE_STEADY_CLOCK -- success
-- Performing Test HAVE_PTHREAD_AFFINITY -- failed to compile
-- pybind11 v3.0.1
-- === SeQuant Package/Feature Info ===
-- The following features have been enabled:
* PYTHON, PySeQuant: Python bindings to SeQuant
* SEQUANT_USE_SYSTEM_BOOST_HASH, SeQuant uses system Boost for hashing (thus results depend on the Boost version)
* SEQUANT_CONTEXT_MANIPULATION_THREADSAFE, The default context can be changed from multiple threads
* TESTS, Enable building of test cases
-- The following OPTIONAL packages have been found:
* boost_charconv (required version == 1.89.0)
* boost_atomic (required version == 1.89.0)
* boost_chrono (required version == 1.89.0)
* boost_container (required version == 1.89.0)
* boost_date_time (required version == 1.89.0)
* boost_exception (required version == 1.89.0)
* boost_thread (required version == 1.89.0)
* boost_locale (required version == 1.89.0)
* boost_headers (required version == 1.89.0)
* boost_regex (required version == 1.89.0)
* Boost (required version >= 1.81)
* Eigen3
-- The following REQUIRED packages have been found:
* Threads
* Python (required version >= 3)
* Git
-- The following features have been disabled:
* TBB, Intel Thread-Building Blocks is an optional prerequisite for C++'s parallel STL
* EVAL_TESTS, Build evaluation tests (if true, will look for and/or build TiledArray)
* BUILD_DOCS, Build SeQuant documentation. Doxygen and Sphinx are required.