gtest coverage for the 53-bit double-backed M2::ARingRR real-number aring. More...

#include <cstdio>
#include <string>
#include <iostream>
#include <sstream>
#include <memory>
#include <gtest/gtest.h>
#include <mpfr.h>
#include "aring-RR.hpp"
#include "ARingTest.hpp"

Functions
bool	almostEqual (const M2::ARingRR &R, unsigned long nbits, const M2::ARingRR::ElementType &a, const M2::ARingRR::ElementType &b)
template<>
void	getElement< M2::ARingRR > (const M2::ARingRR &R, int index, M2::ARingRR::ElementType &result)
	TEST (ARingRR, create)
void	testRingNegateRR (const M2::ARingRR &R, int ntrials)
	TEST (ARingRR, negate)
	TEST (ARingRR, add)
	TEST (ARingRR, subtract)
	TEST (ARingRR, multDivide)
	TEST (ARingRR, axioms)
	TEST (ARingRR, power_and_invert)

Detailed Description

gtest coverage for the 53-bit double-backed M2::ARingRR real-number aring.

Note: AI-generated documentation. Verify against the source before relying on it.

Plugs M2::ARingRR into the ARingTest.hpp harness, defines the per-backend almostEqual(R, nbits, a, b) epsilon checker the floating-point tests use in place of exact equality, and provides the getElement<M2::ARingRR> specialisation that feeds the shared testSomeMore driver with deterministic edge-case prefix elements (zero, plus/minus one, near-overflow exponents) followed by random doubles. The almostEqual helper scales 2^-nbits to match each test's expected propagation of round-off; arithmetic tests use ~50 bits, iterated sqrt / pow round-trips loosen the tolerance as needed.

Sibling files ARingRRRTest.cpp, ARingRRiTest.cpp, ARingCCTest.cpp, and ARingCCCTest.cpp (covered by the same file-aring-real-complex-tests markdown) follow the same template against MPFR / MPFI / complex<double> / MPC backends, including the cross-backend agreement check that ARingRRR at 53 bits stays within 2^-50 of ARingRR.

See also: ARingTest.hpp; aring-RR.hpp; ARingRRRTest.cpp; ARingRRiTest.cpp; ARingCCTest.cpp

Definition in file ARingRRTest.cpp.

Functions

Detailed Description