Skip to content

Latest commit

 

History

History
178 lines (133 loc) · 6.64 KB

News.md

File metadata and controls

178 lines (133 loc) · 6.64 KB

Changes for QuantLib 1.32:

QuantLib 1.32 includes 34 pull requests from several contributors.

Some of the most notable changes are included below. A detailed list of changes is available in ChangeLog.txt and at https://github.com/lballabio/QuantLib/milestone/29?closed=1.

Portability

  • Possibly breaking change: the protected evaluationDate_ data member of the SwaptionVolatilityDiscrete class was renamed to cachedReferenceDate_.

  • Future end of support: we're targeting the future release 1.35 as the last to support Visual C++ 2015, g++ up to version 6.x, and clang up to version 4; support for those compilers will be dropped in release 1.36, about one year from now. From that point onwards, this will allows us to enable the use of C++17 in the code base.

  • Future end of support: at the same time as the above, we'll also remove the configure switch that allows to use boost::tuple, boost::function and boost::bind instead of their std counterparts; starting from this release, the std classes are already the default.

  • Reorganized the CMake presets; thanks to the XAD team (@auto-differentiation-dev).

Cash flows

  • All cash flows are now lazy; thanks to Peter Caspers (@pcaspers).

Instruments

  • Overnight-indexed swaps can now have different schedules and nominals on the two legs; thanks to Tom Anderson (@tomwhoiscontrary).

  • Margrabe options, compound options and chooser options were moved from experimental to core (@lballabio).

  • Introduced common base class FixedVsFloatingSwap for vanilla swap and overnight-indexed swaps; this will be used in the future to help a few existing swap engines support OIS (@lballabio).

  • Added optional redemptions argument to amortizing bond constructors. This allows them to be used for pools of loans where a certain proportion of the underlying loans are subject to defaults and losses. Thanks to Gyan Sinha (@gyansinha).

  • It is now possible to manually prune the notification tree for swaps and bonds if one knows that the cashflows won't change pricer; thanks to Peter Caspers (@pcaspers).

Models

  • Fixed the algorithm to add instruments to the calibration set of the Markov model; thanks to Peter Caspers (@pcaspers) for the fix and Giuseppe Trapani (@lePidduN7) for the heads-up.

Term structures

  • Time-to-date conversion in some swaption volatility classes could return the wrong date before the first exercise date; this is now fixed, thanks to Peter Caspers (@pcaspers).

  • It's now possible to specify the maximum number of iteration for the solver inside a bootstrapped term structure; thanks to Jonathan Sweemer (@sweemer) for the change and Daniel Ángeles Ortiz (@Danie8) for the heads-up.

  • Reduced the number of notifications for bootstrap helpers; thanks to Peter Caspers (@pcaspers).

Random numbers

  • Added the xoshiro265** random-number generator; thanks to Ralf Konrad (@ralfkonrad). It is faster than the Mersenne Twister and might be used as default in the future.

Examples

  • The code of the examples has been modernized a bit; thanks to Jonathan Sweemer (@sweemer).

Patterns

  • Avoided a possible crash when using observables in a multi-threaded setting; thanks to Peter Caspers (@pcaspers).

Deprecated features

  • Removed features deprecated in version 1.27:

    • The QL_NULL_INTEGER, QL_NULL_REAL, QL_NOEXCEPT, QL_CONSTEXPR and QL_USE_STD_UNIQUE_PTR macros.
    • The MultiCurveSensitivities class.
    • The constant, identity, square, cube, fourth_power, add, subtract, subtract_from, multiply_by, divide, divide_by, less_than, greater_than, greater_or_equal_to, not_zero, not_null, everywhere, nowhere, equal_within, clipped_function, clip, composed_function, compose, binary_compose3_function and compose3 functors.
    • The PdeShortRate, ShoutCondition, FDShoutCondition, FDStepConditionEngine and FDEngineAdapter classes from the old finite-differences framework.
    • The dsd::inner_product function.
    • The FDDividendEngineBase, FDDividendEngineMerton73, FDDividendEngineShiftScale and FDDividendEngine pricing engines.
    • The empty headers ql/auto_ptr.hpp, ql/math/initializers.hpp, ql/methods/finitedifferences/americancondition.hpp, ql/methods/finitedifferences/onefactoroperator.hpp, ql/pricingengines/vanilla/fddividendshoutengine.hpp, ql/pricingengines/vanilla/fdshoutengine.hpp and ql/utilities/disposable.hpp.

  • Deprecated the overload of the withReplication method in the DigitalIborLeg, DigitalCmsLeg and DigitalCmsSpreadLeg classes that takes no arguments; use the other overload instead.

  • Deprecated the StandardFiniteDifferenceModel, StandardSystemFiniteDifferenceModel and StandardStepCondition typedefs; define your own typedefs if needed.

  • Deprecated the FDVanillaEngine, FDMultiPeriodEngine, StepConditionSet, ParallelEvolverTraits, ParallelEvolver and SampledCurveclasses and the BSMTermOperator and SampledCurveSet typedefs; use the new finite-differences framework instead.

  • Deprecated the QL_NULL_FUNCTION macro; to check if a function is empty, use it in a bool context instead.

  • Deprecated the now empty headers ql/experimental/exoticoptions/margrabeoption.hpp, ql/experimental/exoticoptions/analyticcomplexchooserengine.hpp, ql/experimental/exoticoptions/analyticeuropeanmargrabeengine.hpp, ql/experimental/exoticoptions/analyticcompoundoptionengine.hpp, ql/experimental/exoticoptions/simplechooseroption.hpp, ql/experimental/exoticoptions/compoundoption.hpp, ql/experimental/exoticoptions/analyticamericanmargrabeengine.hpp, ql/experimental/exoticoptions/analyticsimplechooserengine.hpp, ql/experimental/exoticoptions/complexchooseroption.hpp, ql/experimental/termstructures/multicurvesensitivities.hpp, ql/methods/finitedifferences/shoutcondition.hpp, ql/methods/finitedifferences/pdeshortrate.hpp, ql/pricingengines/vanilla/fddividendengine.hpp, ql/pricingengines/vanilla/fdstepconditionengine.hpp, ql/pricingengines/vanilla/fdconditions.hpp and ql/models/marketmodels/duffsdeviceinnerproduct.hpp.

Thanks go also to Jonathan Sweemer (@sweemer), Ralf Konrad (@ralfkonrad), Klaus Spanderen (@klausspanderen), Peter Caspers (@pcaspers), Tom Anderson (@tomwhoiscontrary), Fredrik Gerdin Börjesson (@gbfredrik), Guillaume Horel (@thrasibule) and the XAD team (@auto-differentiation-dev) for a number of smaller fixes and improvements.