Releases: chenwilliam77/RiskAdjustedLinearizations.jl
Releases · chenwilliam77/RiskAdjustedLinearizations.jl
v1.0.1
v1.0.0
New Features
- Caching for sparse arrays
- Sparse differentiation for Jacobians of key objects comprising a risk-adjusted linearization
- Sparse differentiation for Jacobians computed during calls to
nlsolve
- New example scripts showcasing the speed gains from exploiting sparsity
v0.5.3
Updates
- Fix bug in Taylor rule for NK with Capital example
v0.5.2
Updates
- Bug fix for multivariate Gauss-Hermite quadrature (incorrect use of
CartesianIndices
) - Bug fix when computing deterministic steady state (failure to store
Psi
matrix) - Mild updates to documentation
- Example script of New Keynesian model with capital accumulation and adjustment costs
v0.5.1
- Fix bug in solving deterministic steady state (Psi was not calculated)
- Add an example solving the textbook New Keynesian model
- Timing tests comparing package to MATLAB
v0.5.0
- Extend Lambda and Sigma to permit dependence on jump variables
- Implement Euler equation error diagnostics
- Provide functions for calculating expectations with Gauss-Hermite quadrature
- Add a new example based on Coeurdacier, Rey, and Winant (2011)
- Update RBC with Campbell-Cochrane Habits model to show how to add strips for asset pricing
v0.4.0
Updates
- Implement functions for simulating risk-adjusted linearizations and calculating impulse responses
v0.3.1
Updates
- Fix bugs in handling of keywords for
solve!
- Fix compat restrictions for tests
v0.3.0
Updates
- Implement example with endogenous risk (an RBC model with Campbell-Cochrane Habits)
- Bug fixes to dimension mismatch errors when creating caches and checking Blanchard-Kahn conditions
- Anderson acceleration can be used with the relaxation algorithm
- Document helper and wrapper types
v0.2.0
New Features
- Automatic differentiation can be used in the relaxation algorithm
- Caching for all functions has been updated by using
DiffCache
fromDiffEqBase
and extendingDiffCache
with the new cache typesTwoDiffCache
andThreeDiffCache
, e.g. to handle nested Jacobians. - Use of new helper types
RALF1
andRALF2
to automate conversion of in-place and out-of-place functions to be non-allocating.
Ongoing Issues
- Current approach for setting chunk sizes is not guaranteed to work when using automatic differentiation in the relaxation algorithm and is definitely not working in the homotopy algorithm.