See also our versioning policy.
Breaking changes
-
Changed the default steady-state method to
integrationOnly
(by @dweindl in #2574)The default mode for computing steady states and sensitivities at steady state was changed to
integrationOnly
(from previouslyintegrateIfNewtonFails
).This was done for a more robust default behavior. For example, the evaluation in https://doi.org/10.1371/journal.pone.0312148 shows that - at least for some models - Newton's method may easily lead to physically impossible solutions.
To keep the previous behavior, use:
amici_model.setSteadyStateComputationMode(amici.SteadyStateComputationMode.integrateIfNewtonFails) amici_model.setSteadyStateSensitivityMode(amici.SteadyStateSensitivityMode.integrateIfNewtonFails)
Fixes
-
PEtab import: Fixed potentially incorrect sensitivities with observable/state-dependent sigmas. This was fixed for all cases amici can handle, others cases will now result in
ValueError
s (#2563).by @dweindl in #2562
-
Fixed potentially incorrect disabling of Newton's method
by @dweindl in #2576
-
Fixed
ModelStateDerived
copy ctor, where previously dangling pointers could lead to crashes in some situationsby @dweindl in #2583
-
Added missing simulation status codes
by @dweindl in #2560
-
Check for unsupported observable IDs in sigma expressions
by @dweindl in #2563
Features
-
Optional warning in
fill_in_parameters
by @dweindl in #2578
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.27.0...v0.28.0
This release comes with an updated version of the SUNDIALS package (7.1.1) (#2513). For C++ users of some of AMICI's internal RAII classes, this may include some breaking changes. The Python API is not affected.
Note regarding editable installations (pip install -e ...
):
Due to the SUNDIALS update, it will be necessary to clean out some temporary
build directories (at least ThirdParty/sundials/build/
,
python/sdist/build/
) before rebuilding the package.
Fixes
-
Fixed a bug that led to program termination if a root-after-reinitialization error (potentially also others) occurred at an output timepoint
by @dweindl in #2555
-
CMake: Fixes compilation errors for models named
model
by @dweindl in #2547
-
Updated CMake export config, making it easier to use AMICI in CMake projects and fixing some potential issues with interferring packages
by @dweindl in #2540
-
CMake: Set policies for CMake 3.31
by @dweindl in #2539
-
Documentation fixes by @FFroehlich, @ChocolateCharlie, @dweindl
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.26.3...v0.27.0
Fixes
-
Skip building SuiteSparse shared libraries and build all subprojects together for slightly faster package installation
-
Got rid of petab
DeprecationWarnings
when using theamici_import_petab
CLIby @dweindl in #2517
-
Now also sundials and suitesparse are built in debug mode when installing with
ENABLE_AMICI_DEBUGGING=TRUE
by @dweindl in #2515
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.26.2...v0.26.3
Fixes
-
Fixed a sympy float comparison issue in spline code that would cause an
AssertionError
by @dweindl in #2499
-
Fixed some warnings from recent CMake versions
by @dweindl in #2492
-
Fixed a potential issue when including AMICI in a CMake project
by @dweindl in #2493
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.26.1...v0.26.2
Fixes
-
Fixed some C++ exception handling that previously could crash Python under certain conditions
by @dweindl in #2484
-
Disabled turning warnings into errors when building amici on GitHub Actions in downstream projects
by @dweindl in #2481
-
Fixed CMP0167 warning with CMake 3.30
by @dweindl in #2480
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.26.0...v0.26.1
AMICI v0.26.0 requires sympy>=1.12.1 and petab>=0.4.0.
Policy changes
-
Updated AMICI's versioning / deprecation policy
We will start removing deprecated features that had a deprecation warning for longer than six months in the next minor release.
Deprecations
-
Passing individual tables to
amici_import_petab
is now deprecated. Use apetab.Problem
instance instead.by @dweindl in #2464
Fixes
-
Fixed a bug where during installation of AMICI, an incorrect sundials CMake would be used resulting in installation errors.
by @dweindl in #2468
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.25.2...v0.26.0
Fixes
- Fixed a bug in PEtab import which led to incorrect gradients
w.r.t. estimated initial values specified via the condition table
BREAKING CHANGE:
amici.petab.sbml_import.{import_model_sbml,import_model}
no longer supports passing individual PEtab tables, but only the PEtab problem object. This functionality was deprecated since v0.12.0 (2022-08-26). - Fixes for numpy 2.0 compatibility
NOTE: As long as some amici dependencies don't support numpy 2.0 yet,
you may need to pin numpy to <2.0 in your requirements
(
pip install amici "numpy<2.0"
).
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.25.1...v0.25.2
Fixes
- Avoid clashes with sympy-entities in
plot_expressions
by @dweindl in #2440 - PEtab: fix KeyErrors for missing parameters in
fill_in_parameters
(default values are now used if only a subset of parameters is provided) by @dweindl in #2449 - CMake: Fix Intel MKL detection when not using environment modules by @dweindl in #2443
- CMake: Fix some issues with multi-config generators by @dweindl in #2445
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.25.0...v0.25.1
This release requires Python >= 3.10.
Fixes
- Fixed a bug in event handling that could lead to incorrect simulation results for models with events that assign to compartments and have additional event assignments by @dweindl in #2428
- SBML import: handle
useValuesFromTriggerTime
attribute on events. This attribute was previously ignored. It is possible that now AMICI fails to import models that it previously imported successfully. For cases whereuseValuesFromTriggerTime=True
made a difference, AMICI might have produced incorrect results before. by @dweindl in #2429 - Faster code generation for models with events if they don't have state-dependent triggers by @dweindl in #2417
- Most warnings now come with a more informative code location by @dweindl in #2421
amici.ExpData
was changed so thatisinstance(edata, amici.ExpData)
works by @dweindl in #2396
Features
- Event-assignments to compartments are now supported. Previously, this only worked for compartments that were rate rule targets. by @dweindl in #2425
- Releases are now deployed to Docker Hub by @dweindl in #2413
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.24.0...v0.25.0
This will be the last release supporting Python 3.9. Future releases will require Python>=3.10.
Fixes
- Fix cmake error
cannot create directory: /cmake/Amici
during model import in cases where BLAS was not found viaFindBLAS
by @dweindl in #2389 - Added status code
AMICI_CONSTR_FAIL
by @dweindl in #2379 - Fixed certain initial state issues with PEtab by @dweindl in #2382
- Fixed Solver
operator==
and copyctor (constraints were not copied correctly) by @dweindl in #2388 - Avoid confusing warnings about non-finite timepoints by @dweindl in #2395
- Fixed incorrect exception types / messages for
IDASolver
by @dweindl in #2398 - cmake: set SUNDIALS path hint for python package to help CMake find the correct SUNDIALS installation by @dweindl in #2397
Features
- Optionally include measurements in
plot_observable_trajectories
by @dweindl in #2381 - Improved type annotations in swig-wrappers by @dweindl in #2401
- Additional attributes are accessible directly via
ReturnDataView
andExpDataView
, e.g.ReturnDataView.ny
,ReturnDataView.nx
by @dweindl in #2405 - Allow subselection of state variables for convergence check during
steady-state simulations via
Model.set_steadystate_mask([1, 0, ..., 1])
(positive value: check; non-positive: don't check). by @dweindl in #2387
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.23.1...v0.24.0
- Fixes installation issues related to building SuiteSparse on some systems by @dweindl in #2375
Features
-
SBML
InitialAssignment
are no longer absorbed into other model expressions, but are available as parameters or expressions (w
) in the amici model by @dweindl in #2304, #2305, #2345, #2359 -
Upgraded to SuiteSparse 7.6 by @dweindl in #2316
-
Model expressions
w
are now split into static and dynamic expressions, and only evaluated as needed by @dweindl in #2303 -
Exposed additional solver settings:
Solver.setMaxConvFails()
: maximum number of non-linear solver convergence failuresSolver.setMaxNonlinIters()
: maximum number of non-linear solver iterationsSolver.setMaxStepSize()
: maximum step sizeSolver.setConstraints()
: for setting (non)negativity/positivity constraints on state variables
-
Improved output for debugging simulation failures:
ReturnData.{xdot,J}
now contain the respective values from the timepoint of failure, not the last output timepoint. NaN/Inf warnings now always include the timepoint at which the issue occurred. Note that C++ stacktraces are now only logged for debug builds. by @dweindl in #2349, #2347, #2366 -
Updated dataframes import/export to include parameter values and scales by @FFroehlich in #2351
Fixes
- CMake: Updated BLAS detection and some minor fixes by @dweindl in #2318 and #2357
- Deterministic ordering of source files in generated
CMakeLists.txt
by @dweindl in #2322 - Fixed size check in
Model::setStateIsNonNegative
by @dweindl in #2332 - Fixed uncaught C++ exception in
runAmiciSimulation
that may crash Python in case of invalid values for standard deviations by @dweindl in #2338 - Fixed missing import in
amici/petab/petab_import.py
by @plakrisenko in #2342 - Fixed
ReturnDataView
AttributeError: messages
by @dweindl in #2341 - Added a missing return code constant
LSETUP_FAIL
by @dweindl in #2353 - Fixed in-place building of model wheels by @dweindl in #2352
- Made is-zero-checks compatible with the upcoming sympy>1.12 by @dweindl in #2350
- Fixed issues with paths containing blanks for sundials by @dweindl in #2361
- Added
amici.petab.conditions
to the API documentation by @PaulJonasJost in #2364 - Improved type annotations in swig-wrappers by @dweindl in #2344, #2365
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.22.0...v0.23.0
Features
-
PEtab import: User option to fail if model needs to be compiled by @dilpath in #2289
The
force_compile
argument is now deprecated. Usecompile_
instead. -
Model import now adds a
.gitignore
file to the model output directory by @dweindl in #2301
Fixes
- Fixed a bug that may have caused wrong simulation results for certain
SBML models that contain
rateOf
-expressions by @dweindl in #2291 - More informative error message for
ReturnDataView.by_id
by @dweindl in #2295 - Fixed
ENABLE_AMICI_DEBUGGING=TRUE
not working with MSVC by @dweindl in #2296 - Fixed MANIFEST.in warning by @dweindl in #2297
- (performance) Skip unnecessary toposorting in
DEModel._collect_heaviside_roots
by @dweindl in #2299 - (performance) Fix redundant calls to
Model::fdwdx
fromModel_ODE::fJ
(only relevant for dense and banded solvers) by @dweindl in #2298
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.21.2...v0.22.0
- Fixed
Solver
copyctor issues with swig4.2 that resulted in installation errors by @dweindl in #2276 - Fixed error when calling
amici.ExpData()
by @dweindl in #2280 - Fixed invalid-type-error when loading an antimony model from file by @dweindl in #2281
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.21.1...v0.21.2
Fixed package configuration for PyPI upload. No further changes.
Deprecations
- Moved PEtab-related functionality from
amici.petab_*
to the petab-subpackageamici.petab.*
. The old public functions are still available but will be removed in a future release. by @dweindl in #2205, #2211, #2252
Features
- Handle events occurring at fixed timepoints without root-finding. This avoids event-after-reinitialization errors in many cases a brings a slight performance improvement. by @dweindl in #2227
- Added
PetabProblem
class for handling PEtab-defined simulation conditions, making it easier to perform customized operations based on PEtab-defined simulation conditions. by @dweindl in #2255 - code-gen: Simplified
switch
statements, leading to reduced file sizes and faster compilation for certain models. by @dweindl in #2240 - Made
Model
andModelPtr
deepcopyable by @dweindl in #2247 - Made
Solver
andSolverPtr
deepcopyable by @dweindl in #2245 - Added a debugging helper
get_model_for_preeq
for debugging simulation issues during pre-equilibration. by @dweindl in #2250 - Added
SwigPtrView
fields todir()
outputs by @dweindl in #2244 - Use proper labels for in plotting functions if IDs are available in
ReturnData
. by @dweindl in #2249 - Added
ExpData::clear_observations
to set all measurements/sigmas to NaN by @dweindl in #2258
Fixes
- Fixed AMICI hiding all warnings. Previously, importing
amici
resulted in all warnings being hidden in the rest of the program. by @dweindl in #2243 - CMake: Fixed model debug builds by @dweindl in #2222
- Fixed CMake potentially using incorrect Python library for building model extension by @dweindl in #2220
- CMake: fixed cxx flag check by @dweindl in #2225
- Fixed potential out-of-bounds read in
Model::checkFinite
for matlab-imported models by @dweindl in #2232 - Fixed piecewise/Heaviside handling by @dweindl in #2234
- Deterministic order of event assignments by @dweindl in #2242
- Proper error message in case of unsupported state-dependent sigmas by @dweindl in #2239
- Fixed swig shadow warning + other linting issues by @dweindl in #2261
- Fixed
SwigPtrView.__getattr__
by @dweindl in #2259 simulate_petab
: Avoid warning when simulating with default parameters by @dweindl in #2265
Documentation
- Updated Python package installation instructions for Arch Linux by @willov in #2212
- Updated
ExpData
documentation by @dweindl in #2254 - Documented simulation starting time
t0
by @dweindl in #2263 - Updated PEtab example by @dweindl in #2255
...
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.20.0...v0.21.0
Fixes
- Fixed CMake
cmake_minimum_required
deprecation warning by @dweindl in #2183 - Fixed misleading preequilibration failure messages by @dweindl in #2181
- Removed setuptools<64 restriction by @dweindl in #2180
- Fixed ExpData equality operator for Python by @dweindl in #2194
- Enabled deepcopy for ExpData(View) by @dweindl in #2196
- Allowed subsetting simulation conditions in simulate_petab by @dweindl in #2199
- Set CMake CMP0144 to prevent warning by @dweindl in #2209
Features
- Possibility to evaluate and plot symbolic expressions based on simulation results by @dweindl in #2152
- Easier access to timepoints via ExpDataView by @dweindl in #2193
- Nicer
__repr__
for ReturnDataView by @dweindl in #2192
Documentation
- Added installation instructions for Arch Linux by @stephanmg in #2173
- Updated reference list by @dweindl in #2172
- Installation guide: optional requirements by @dweindl in #2207
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.19.0...v0.20.0
Features
-
SBML import now supports
rateOf
by @dweindl in #2120 -
Added
Model.{get,set}SteadyStateComputationMode
(analogous toSteadyStateSensitivityMode
) which allows to choose how steady state is computed. by @plakrisenko in #2074Note: The default
SteadyStateSensitivityMode
changed fromnewtonOnly
tointegrateIfNewtonFails
. -
SBML import: Allow hardcoding of numerical values by @dweindl in #2134
-
Added
antimony2amici
for more convenient antimony import (simplifies working with raw ODEs, see documentation) by @dweindl in #2142 -
Added
AMICI_TRY_ENABLE_HDF5
environment variable to control whether to search for HDF5 or not by @dweindl in #2148
Fixes
- Fixed SBML import for events with trigger functions depending on parameters that are initial assignment targets by @dweindl in #2145
- Fixed SBML import for event-assigned parameters with non-float initial assignments by @dweindl in #2156
- Fixed
unistd.h
dependency ofhdf5.cpp
that led to compilation failures on Windows by @dweindl in #2154 - Set CMake policies for cmake 3.27 by @dweindl in #2162
- Fixed a
lib/
vslib64/
issue, leading toSUNDIALSConfig.cmake
-not-found issues on some systems by @dweindl in #2165 - CMake: fixed scope of
-DHAS_BOOST_CHRONO
which may have lead to a mix ofboost::chrono::thread_clock
andstd::clock
being used in programs using amici, and potentially segmentation faults by @dweindl in #2163
Performance:
-
Combined code for sparse model functions and their index files for slightly faster compilation of small models by @dweindl in #2159
-
Removed complex / complex long KLU functions for slightly faster amici package installation by @dweindl in #2160
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.18.1...v0.19.0
Fixes:
- Fixed pysb pattern matching during PEtab import by @FFroehlich in #2118
- Fixed
sp.Matrix
errors withnumpy==1.25
by @dweindl in #2124 - Readme: added info containers by @dweindl in #2125
- Fixed deprecation warnings by @dweindl in #2122 #2131
- Fixed logging typo in SBML import by @dilpath in #2126
- Added minimum version for
pandas
by @dilpath in #2129
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.18.0...v0.18.1
Features:
- More efficient handling of splines in SBML models by @paulstapor, @lcontento, @dweindl in #1515
- Partial support of current PEtab2.0 draft, including support for PySB models by @dweindl, @FFroehlich in #1800
Fixes:
- Fixed incorrect forward sensitivities for models with events with state-dependent trigger functions by @dweindl in #2084
- Model import: Don't create spl.h and sspl.h for models without splines by @dweindl in #2088
- SBML import - faster processing of SpeciesReference IDs by @dweindl in #2094
- Update swig ignores by @FFroehlich in #2098
- CMake: Fixed choosing SWIG via
SWIG
env variable by @dweindl in #2100 - CMake: Try FindBLAS if no other information was provided by @dweindl in #2104
- Fixed cblas error for models without solver states in combination with forward sensitivities by @dweindl in #2108
- Fixed compilation error for models with events and xdot=0 by @dweindl in #2111
- Fixed import error for models with events and 0 states by @dweindl in #2112
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.17.1...v0.18.0
This release fixes two bugs:
- One bug introduced in v0.17.0, that causes an
ImportError
on macOS (#2075). - An AttributeError in petab_import_pysb with petab>=0.2.0 #2079
AMICI v0.17.0 requires Python>=3.9 and a C++17 compatible compiler
Features
- DAE support in SBML by @FFroehlich in #2017
- SBML import: flatten SBML-comp models by @dweindl in #2063
- Added sllh computation back to
petab_objective.simulate_petab
by @dilpath in #1548 - CMake-based Python extension builds by @dweindl in #1992
Fixes
- Fixed CPU time tracking with multi-threading (partially) by @dweindl in #2023
- Fixed HDF5 ambiguous overload by @dweindl in #2031
- Fixed varying cmake libdir lib(64)/ by @dweindl in #2033
- Fixed Equilibration cpu time computation by @plakrisenko in #2035
- CMake: add header files to library sources for generated models by @dweindl in #2047
- CMake: Handle header-dependency of swig files by @dweindl in #2046
- Don't try to detect conservation laws for models with Species-AssignmentRules by @dweindl in #2056
- Smith benchmark and SBML initialization fix by @FFroehlich in #2034
- SBML import: Fixed check for required packages by @dweindl in #2064
- Nan observables by @FFroehlich in #2065
- Fixed check for discontinuities for conservation law computation by @dweindl in #2068
- Specify visualization dependencies by @dweindl in #2070
- Fixed sympy symbol name clashes during PEtab import by @dweindl in #2069
- Fixed ReturnData::{preeq_wrms,posteq_wrms} with FSA and check_sensi_steadystate_conv_=True by @dweindl in #2071
Extended / updated documentation, for example:
- Jax example notebook by @FFroehlich in #1996
- Updated Windows/MSVC installation instructions by @Podde1 in #2053
New Contributors
- @Podde1 made their first contribution in #2053
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.16.1...v0.17.0
Fixes:
- Additional package names for finding blas via pkg-config by @dweindl in #1959
- Changed default interpolation type from hermite to polynomial by @dweindl in #1960
- PySB import: Change default simplify to work with multiprocessing by @dweindl in #1961
- Add --no-validate to amici_import_petab @dweindl in #1963
- Fix get_model for direct import of swig interface by @dweindl in #1969
- Fix PytestReturnNotNoneWarning in test_conserved_quantities_demartino.py by @dweindl in #1968
- Fix MSVC builds / remove -W* flags by @dweindl in #1972
- Add option to use IDs when plotting trajectories by @dweindl in #1974
- Fix assignmentRules2observables - skip non-assignment-rule targets by @dweindl in #1973
- Use std::clock for measuring solver time by @dweindl in #1982 (Note that this uses cpu-time consumed by all threads)
- Fix narrowing-conversion-warning by @dweindl in #1983
- PEtab import: allow specifying default values for output parameters by @dweindl in #1987
- Print stacktraces only with debug logging level by @dweindl in #1985
- Change default ReturnData::status to AMICI_NOT_RUN by @dweindl in #1984
- Reduce time-tracking overhead by @dweindl in #1988
- Fix equilibraton status discrepancy by @plakrisenko in #1991
- Pass model_name to _create_model_output_dir_name by @FFroehlich in #1994
- CMake: Build with OpenMP support if available by @dweindl in #2000
- Fix SuiteSparse Makefiles for compiler-paths by @dweindl in #2003
- CMake: Build with HDF5 support if available by @dweindl in #1999
- CMake: Fix reading version file on Windows by @dweindl in #2001
- CMake: raise minimum required version to 3.15 by @dweindl in #2002
- Fix/extend runtime logging by @FFroehlich in #2005
- Fix error logging in steadystate solver by @FFroehlich in #2008
- Don't pass
-py3
to swig after 4.1.0 by @FFroehlich in #2010 - SWIG __repr__s for different templated vector classes by @FFroehlich in #2009
- Matlab: If mex fails, print mex arguments by @dweindl in #2013
- Simplify OpenBLAS installation on Windows by @dweindl in #2016
- Remove model name prefix in generated model files by @dweindl in #2015
- ...
Documentation:
- Restructure sphinx doc by @dweindl in #1978
- Instructions for AMICI with singularity by @dweindl in #1964
- Illustrate options for potentially speeding up model import/simulation by @dweindl in #1965
- ...
Dependencies:
- Updated SuiteSparse to v7.0.1 by @dweindl in #2018
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.16.0...v0.16.1
Features
- Python 3.11 compatibility by @dweindl in #1876
- AMICI now runs on binder (https://mybinder.org/v2/gh/AMICI-dev/AMICI/develop?labpath=binder%2Foverview.ipynb) by @dweindl in #1935, #1937, #1939
- More informative
Solver.__repr__
andExpData.__repr__
by @dweindl in #1928 and @FFroehlich in #1948 simulate_petab
returns the generated/usedExpData
s by @dweindl in #1933- Model module is now accessible from model instance by @dweindl in #1932
- Added
plot_jacobian
by @dweindl in #1930 - Now logs all nested execution times as debug by @FFroehlich in #1947
- Always check for finite initial states, not only with
Model.setAlwaysCheckFinite(True)
by @dweindl in #1955
Fixes
ReturnDataView.status
now returnsint
instead offloat
by @dweindl in #1929- Updated simulation status codes by @dweindl in #1931
- Skip irrelevant frames in stacktraces by @dweindl in #1934
- Fixed compiler warning (matlab) by @dweindl in #1954
Documentation:
- Added a notebook demonstrating common simulation failures and show how to analyze / fix them by @dweindl in #1946
- various minor fixes / updates
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.15.0...v0.16.0
Features
-
Improved logging by @dweindl in #1907
For Python: Don't print messages to stdout, but collect them in ReturnData and forward them to python logging, making it easier to filter specific messages or to disable output completely. Messages are also available via
ReturnData.messages
.breaking change for C++ interface: Messages aren't printed to stdout by default, but are collected in
ReturnData
. The user has to decide what to do with them. -
MultiArch docker build by @FFroehlich in #1903
-
Added cmake target for cmake-format by @dweindl in #1909
-
Updated clang-format style, fixed clang-format target by @dweindl in #1908
-
Subsetting
ReturnData
fields by ID viaReturnDataView.by_id
by @dweindl in #1911 #1916
Fixes
- PEtab import: fixed handling of fixed parameters for rule targets by @dweindl in #1915
- Fixed compiler warnings for matlab interface by @dweindl in #1919
- Fixed pandas DeprecationWarning for Series.iteritems() by @dweindl in #1921
- Fixed circular import in amici.petab_import_pysb by @dweindl in #1922
- Fixed 'operator ==' swig warning by @dweindl in #1923
- Prevent swig4.0.1 segfault by @dweindl in #1924
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.14.0...v0.15.0
-
Added optional functionality to apply C99 math optimization to generated C++ code by @dweindl and @lcontento in #1377, #1878
-
Added option to treat fixed parameters as constants in PEtab import
by @dweindl in #1877
-
Added equality operator for ExpData
by @dweindl in #1881
-
Updated base image for Dockerfile to Ubuntu 22.04/Python 3.10
by @dweindl in #1896
-
Fixes/updates to GitHub actions by @dweindl in #1885, #1893, #1889, #1891
-
Added hdf5 search directories for arm64 architecture (M1/M2 macs)
by @Doresic in #1894
-
Fixed missing return in generated non-void functions
by @dweindl in #1892
-
Fixed import failure for pre-compiled models
by @dweindl in #1897
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.13.0...v0.14.0
- Fixed extraction of common subexpressions by @dweindl in #1865
- Added function to convert
ReturnData::status
flags to string by @dweindl in #1864
And further contributions by @dweindl, @FFroehlich
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.12.0...v0.13.0
Features:
- Support for event observables via the Python interface by @FFroehlich in #1845
- Treat non-estimated parameters as constants during SBML-PEtab import by @dweindl in #1810
- Updated SUNDIALS to v5.8.0 by @dweindl in #1836
- Option to extract common subexpressions by @dweindl in #1852, #1856 not available in this release, use v0.13.0
- Parallelize matrix simplification by @dweindl in #1778
- Validate PEtab problems before attempting import by @dweindl in #1842
- Improved type annotations for the swig interface by @dweindl in #1860
Fixes:
- Fixed an issue with potentially infinite loops during conservation law processing by @FFroehlich in #1833
- Fixed potential deadlocks during parallel simplification by @dweindl in #1844
- Fix resetting
ReturnData::numstepsB
when re-using Solver by @dweindl in #1841
And further contributions by @dilpath, @dweindl, @FFroehlich
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.11.32...v0.12.0
Fixes:
- Fixed
ImportError
s during package installation with recent setuptools by @dweindl in #1830
Fixes:
- Fixed
ParameterMapping.__getitem__
to either return aParameterMappingForCondition
or a newParameterMapping
, but not a list by @dweindl in #1826
Features:
- Allow overriding model name during PySB import by @dweindl in #1801
- Added repr for parameter mapping classes by @dweindl in #1799
- More informative warning messages for NaNs/Infs by @dweindl in #1798
- Moved
sim_steps
increment by @plakrisenko in #1806 - Re-arranged application of parameters from
ExpData
to avoid initial sensitivities error by @dilpath in #1805 - Checking for unused parameters in
simulate_petab
by @dweindl in #1816 - Add
create_parameter_mapping
kwarg forwarding by @dweindl in #1820
Other
- Remove
constant_species_to_parameters
by @dweindl in #1809
Fixes
- Fixed handling of SBML models given as
pathlib.Path
in `petab_import.import_model_sbml by @dweindl in #1808 - Fixed missing CPU time reset by @dweindl in #1814
- Raise in
simulate_petab
withscaled_parameters=True
problem_parameters=None
by @dweindl in #1819
...
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.11.29...v0.11.30
Features:
- Performance: Limit newton step convergence check by @FFroehlich in #1780
- More informative NaN/Inf warnings by @dweindl in #1640
- SBML import can now handle initial events by @FFroehlich in #1789
Fixes:
- Avoid error if no measurements in PEtab problem; fixed type handling in PEtab parameter mapping by @dilpath in #1783
- Fixed substitution of expressions in root and stau by @dilpath in #1784
- Workaround for PEtab problems with state-dependent noise models by @dweindl in #1791
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.11.28...v0.11.29
New features:
-
Added
Solver.setSteadyStateToleranceFactor
andSolver.setSteadyStateSensiToleranceFactor
to specify a steady state tolerance factor by @dilpath in #1758NOTE: This also relaxed the default steady state (sensitivity) tolerances by a factor of 100.
-
Added support for
pathlib.Path
by @dweindl in #1769 -
Allow specifying initial timepoint with
ExpData
by @dilpath in #1776
Performance:
- Speedup for models with conservation laws by @FFroehlich in #1765
- Improved efficiency of newton step convergence check by @FFroehlich in #1775
Fixes:
- Fixed deprecation warning for pandas.DataFrame.append in
rdatas_to_measurement_df
by @dweindl in #1770 - Fixed Rule-target handling in PEtab import by @dweindl in #1753
Removed functionality:
- Removed long deprecated
sbml2amici
argumentsmodelName
andconstantParameters
by @dweindl in #1774
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.11.27...v0.11.28
New features:
- Checking condition number when computing sensitivities via Newton by @FFroehlich in #1730
- Removed SPBCG solver by @FFroehlich in #1729
- Added Newton step convergence checks to steadystate solver by @FFroehlich in #1737
- Removed legacy options/members
amioption.newton_preeq
and `Solver::r… by @dweindl in #1744 - Added
ReturnData::cpu_time_total
to track total time spent inrunAmiciSimulation
by @dweindl in #1743 - SBML import: Alternative algorithm for identifying conservation laws by @dweindl in #1748
- Use
amici.AmiciVersionError
to indicate version mismatch by @dweindl in #1764
Performance:
- Optional parallel computation of derivatives during model import by @dweindl in #1740
- Sparsify jacobian by @FFroehlich in #1766
- Speedup conservation law computation by @FFroehlich in #1754
- Exploit stoichiometric matrix in pysb import by @FFroehlich in #1761
- Speedup edata construction from petab problems by @FFroehlich in #1746
Fixes:
- Fixed
get_model_settings
that would to setting incorrect initial states and initial state sensitivities for models with parameter-dependent initial states by @dweindl in #1751 - Use correct tolerances for convergence check in Newton solver by @FFroehlich in #1728
- Harmonized convergence checks by @FFroehlich in #1731
- Made sundials' KLU_INDEXTYPE match actual klu index type by @dweindl in #1733
- Fixed
Model::setStateIsNonNegative
logic that would raise exceptions in cases where it shouldn't by @dweindl in #1736 - Fixed undefined reference to dladdr by @kristianmeyerr in #1738
- Fixed HDF5 OSX intermediate group creation errors by @dweindl in #1741
- Fixed recent cmake-based build issues due to changed sundials library directory by @dweindl in #1756
- Updated Windows installation instructions by @paulflang in #1763
... and other contributions by @FFroehlich, @dweindl
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.11.26...v0.11.27
New features:
- Import of BioNetGenLanguage (BNGL) models by @FFroehlich in #1709
- Added support for observable-dependent sigmas by @dweindl, @FFroehlich in #1692
- Added support for pysb local functions by @FFroehlich in #1666
- Added experimental support for conservation laws for non-constant species to
SBML import: conservation laws for non-constant species
by @stephanmg, @dweindl in #1669
Enable this feature by setting environment variable
AMICI_EXPERIMENTAL_SBML_NONCONST_CLS
to any value - Added optional caching for symbolic simplifications in ODE export by @dilpath in #1672
- Added CLI option
--no-sensitivities
toamici_import_petab
by @dweindl in #1688
Fixes:
- SBML import: Raise in case of nested observables by @dweindl in #1690
- Sympy 1.10 compatibility by @dweindl in #1694
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.11.25...v0.11.26
- Fixed a bug
where
Model::setStateIsNonNegative(Model::getStateIsNonNegative())
would raise an exception in case conservation laws were enabled - by @dweindl in #1648 - Fixed a bug where
Model::setStateIsNonNegative
would be ignored in certain model expressions - by @FFroehlich in #1650 - Fixed a bug where special function parsing inside
min()
andmax()
would not be parsed correctly - by @dweindl in #1655 - Fixed a numpy dependency issues for Mac+ARM systems - by @dweindl in #1657
- Fixed convergence check in Newton method - by @plakrisenko in #1663
- Add
AMICI_CXX_OPTIONS
to pass libamici-specific compiler options during CMake-based builds - by @dweindl in #1664 - Fixed various warnings and updated documentation - by @dweindl
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.11.24...v0.11.25
Features:
- Introduced environment variable
AMICI_DLL_DIRS
to control DLL directories on Windows (useful for setting BLAS library directory, as required by Python>=3.8) by @dweindl in #1637 - Dropped Python3.7 support by @dweindl in #1635
- Include header files in CMake targets for better IDE integration by @dweindl in #1639
Fixes:
- Fixed an issue in PEtab import where all-integer parameters would previously result in a TypeError by @stephanmg in #1634
- Fixed tempdir deletion issues for test suite on Windows by @dweindl in #1636
- Added functions to provide state IDs/names for x_solver by @dweindl in #1638
- Fixed docs on RTD by @dweindl in #1643
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.11.23...v0.11.24
Features:
- Added overload for Model::setParameterScale with vector by @dilpath in #1614
- Removed assert_fun argument from gradient checking, improve output by @dweindl, @FFroehlich in #1609
- Added get_expressions_as_dataframe by @dweindl in #1621
- Added
id
field to ExpData and ReturnData by @dweindl in #1622 - Included condition id in dataframes by @dweindl in #1623
Fixes:
- C++: Fixed SUNMatrixWrapper ctor for size 0 matrices by @dweindl in #1608
- Python: Handle TemporaryDirectory cleanup failures on Windows by @dweindl in #1617
- Python: pysb.Model.initial_conditions throws a DeprecationWarning by @PaulJonasJost in #1620
- Fixed wrong array size in warnings by @dweindl in #1624
NOTE: AMICI 0.11.23 requires numpy<1.22.0
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.11.22...v0.11.23
- *Require sympy>=1.9,pysb>=1.13.1 by @FFroehlich, @dweindl in #1599
- Fixed sympy deprecation warning by @dweindl in #1600
- Updated Windows installation instructions for Python>=3.8 by @dweindl in #1597
- Fixed plot labels by @dweindl in #1598
Full Changelog: https://github.com/AMICI-dev/AMICI/compare/v0.11.21...v0.11.22
Fixes:
- Fixed a bug in recursion depth computation for model expressions. This may have resulted in incorrect sensitivities for models with expressions nested more than 2 levels. (#1595)
- Fixed improper handling of Piecewise functions in PySB import which may have produced incorrect simulation results. (#1594)
- Fixed changed googletest reference which broke the CMake-based build if tests were enabled (#1592)
New:
- It's now possible to build AMICI using Ninja (#1593)
New:
- Changed parameter mappings such that unassigned values have non-nan default values. This fixes erroneous evaluation of
llh
asNaN
in some situations (#1574) - Added support for Python 3.10 (#1555)
Fixes:
- Fixed a bug when simulation start time was not transferred when copying a solver instance (#1573)
- Fixed a bug which led to incorrect sensitivies for models with multiple assignment rules or rate rules (#1584)
Other:
- Update CI and documentation settings (#1569, #1527, #1572, #1575, #1579, #1580, #1589, #1581)
- Extend set of validated benchmark models that is checked during CI (#1571, #1577)
- Fixed string formatting in derivative checks (#1585)
- Added helper methods to save and restore model instance-only settings (#1576)
New:
- Added support for observable transformations (lin/log/log10) (#1567). Thereby supporting additional noise distributions in combination with least squares solvers.
Fixes:
- Fixed a bug when Newton sensitivity computation was activated despite specifying newton_steps == 0. The error occurs when simulation converges to a steadystate but simulation sensitivities are not converged according to convergence criteria. In that case simulation returned failure, but the newton rootfinding "finds" a steadystate even before the iteration check, leading to the erroneous computation of sensitivities via Newton/IFT. For singular jacobians this means the overall simulation still fails, but a different, more informative error message is displayed. (#1541)
- Fixed a bug where argument "outdir" in ODEExporter.init would not be used (#1543)
Other:
- Improve checking support for SBML extensions (#1546)
- SBML import: Use more descriptive IDs for flux expressions (#1551)
- Optimized SUNMatrixWrapper functions (#1538)
- C++: Changed test suite from CppUTest to gtest (#1532)
- Add CITATION.cff (#1559)
- Updated documentation (#1563, #1554, #1536)
- Removed distutils dependency (#1557)
- Require sympy<1.9
New:
- Allow specifying maximum integration time via
amici::Solver::setMaxTime()
(#1530) - Py: Add
failfast
andnum_threads
argument tosimulate_petab
(#1528, #1524) - Enable typehints / static type checking for AMICI-generated model modules
(#1514) (
amici.ModelModule
, available with Python>=3.8)
Fixes:
- Python: Fix unused argument
generate_sensitivity_code
inpysb2amici
(#1526) - Python: Fix C(++) stdout redirection which could have let to deadlocks in exotic situations (#1529)
- Py: Fixed deprecation warning (#1525)
- Py: Proper typehints for SWIG interface (#1523), enabling better static type checking and IDE integration (available with Python>=3.9)
- C++: Fixed clang compiler warning (#1521)
- C++: Fix inherited variadic ctor in exception class (#1520)
- PEtab: More informative output for unhandled species overrides (#1519)
- Return SbmlImporter from PEtab model import (#1517)
Fixes:
- Fix "maybe-uninitialized" compiler warning (#1495)
- Fix substitution of expressions in `drootdt_total (#1512)
- C++: Fix serialization and == operator (#1493)
- C++: Avoid
w
inroot
andstau
headers (refactor) (#1503)
Documentation:
- Updated OpenBLAS Windows installation instructions (#1496)
- Updated how-to-cite to Bioinformatics paper (#1499)
- Updated list of papers using AMICI (#1509)
Other:
- Remove sllh computation from
petab_objective.simulate_petab
(#1498) - Add main.py to Python package to provide info on AMICI installation
via
python -m amici
(#1500)
Fixes:
- Fixed serialization bug (#1490)
New:
- Construction of condition specific plist for parameter mappings (#1487, #1488)
- Add support for error residuals (#1489)
Fixes:
- Fixed initial state sensitivities in adjoint preequilibration (#1468)
- Fixed various model import / parameter mapping issues (#1469, #1473, #1475)
New:
- New AMICI releases will automatically trigger releases at https://biosimulators.org/simulators/amici/latest
- Transparent logo
New features:
- Python import now supports SBML Events (#1443)
- Implement support for compilation without sensitivities (#1454)
Fixes:
- Issue #1446: Check whether constant parameters are valid targets (#1450)
- Issue #1422: Fix Steadystate solver failing if preequilibration starts in steadystate (#1461)
- Issue #1401: Ensure diagnostics variables in ReturnData are always of expected length (#1438, #1447)
- Fix FIM approximation for parameter dependent sigma (#1441)
- Fix invalid SBML in PEtab/PySB import (#1433)
- Fix: No context for inspect.getouterframes (#1432)
Documentation:
- Added this CHANGELOG
- Added feature request issue template (#1437)
- Updated reference list (#1430)
- Overhauled experimental conditions notebook (#1460)
CI:
- Test Matlab interface on GHA (#1451)
- Include componentTags in SBML test suite output (#1462)
- Split SBML semantic test suite into multiple jobs (#1452)
- Fix Crauste ref val, fixes #1458 (#1459)
Misc:
- Various cleanup (#1465, #1448, #1455)
- Micro-optimize SUNMatrixWrapper::transpose (#1439)
- Remove constant triggers from roots in Heaviside (#1440)
Breaking changes:
- AMICI requires Python>=3.7
- Updated package installation (PEP517/518): Creating source distributions requires https://github.com/pypa/build (#1384) (but now handles all package building dependencies properly)
Features:
- More flexible state reinitialization (#1417)
Updated dependencies:
- Upgrade to sundials 5.7.0 (#1392)
Fixes:
- Python: account for heaviside functions in expressions (#1382)
- Python: allow loading of existing models in import_petab_problem (#1383)
- Python: Don't override user-provided compiler/linker flags (#1389)
- Python: PEtab import reinitialization fixes (#1417)
- Python: Fix PEtab observables for pysb models (#1390)
- Python: Substitute expressions in event condition expressions (#1404)
- Python: Unspecified initial states in PEtab conditions table default to SBML initial value (#1397)
- C++: Fix timepoint out of bounds access (#1402)
- C++: Fix exported CMake config (#1388)
- Fixed Dockerfile: add python3-venv (#1398, #1408)
Other:
- Slim exported swig interface (#1425)
- Updated documentation
- Getting started tutorial (#1423)
- List supported SBML test tags (#1428)
- Add AMICI C++/Python/Matlab feature comparison (#1409)
- ...
- Various minor CI improvements
- ...
Features:
- Add expression IDs and names to generated models (#1374)
Fixes:
- Raise minimum sympy version to 1.7.1 (Closes #1367)
- Fix species assignment rules in reactions (#1372)
- Fix id vector for DAEs (#1376)
Docs:
- Update how-to-cite (#1378)
- Restore support for species references (#1358)
- Add support for noise models in pysb (#1360)
- Proper handling of discontinuities in the ODE rhs (#1352)
- Fix directly calling AMICI from snakemake (#1348 )
- Extend mathml function support, particularly for numerical arguments (#1357)
- Restore support for sympy 1.6 (#1356)
- Fix some compiler related warnings (#1349, #1362 )
- Fix a rare segfault for adjoint sensitivities (#1351)
- Move windows tests to GHA (#1354)
- Pin breathe to 4.24.1
- Update ubuntu to 20.04
Bugfix release that restores compatibility with sympy 1.7
- General improvements to SBML import (#1312, #1316, #1315, #1322 , #1324 #1333, #1329)
- Small bugfixes and improvements (#1321 )
- Improve derivative computation for instances of
power
(#1320 )
- Fix FIM and residual computation for models with parameter dependent sigma. (#1338)
- Disable chi2/residual/FIM computation for non-gaussian objective functions. (#1338)
- Bugfix for integration failure during adjoint solve (#1327)
- Update references (#1331, #1336)
- Update OpenBLAS for windows (#1334)
- Fix pysb-petab support (#1288)
- Fix ExpData constructor overloading (#1299)
- Fix support for positivity enforcement (#1306)
- Refactor SBML import, adds support for parameter rate rules and initial assignments (#1284, #1296, #1304)
- Improve model generation for models with many parameters (#1300)
- Add support for PEtab based synthetic data generation (#1283)
- Make HDF5 an optional dependency (#1285)
- General Improvements to Documentation (#1289, #1291, #1292, #1293, #1294, #1286, #1277, #1281)
- Add python 3.9 support test (#1282)
- Allow manual triggering of GitHub actions (#1287)
- Remove appveyor config (#1295)
- Update GHA env and path management (#1302)
- Improve and extend available objective functions (#1235)
- Fix processing of compartment definitions (#1223)
- Fix replacement of reserved symbols (#1265)
- Use Hierarchical Derivatives for Expressions (#1224, #1246)
- Fix duplicate running of swig (#1216)
- Overload python interface functions for amici.{Model,Solver,ExpData} and amici.{Model,Solver,ExpData}Ptr (#1271)
- Fix and extend use of sparse matrix operations (#1230, #1240, #1244, #1247, #1271)
- Fix application of maximal number of steps, MaxNumStep parameter now limit total number of steps, not number of steps between output times. (#1267)
- Move all Documentation to RTD (#1229, #1241)
- General Improvements to Documentation (#1225, #1227, #1219, #1228, #1232, #1233, #1234, #1237, #1238, #1239, #1243, #1253, #1255, #1262)
- Better check for doc building (#1226)
- Add more gradient checks (#1213)
- Update GHA to Ubuntu 20.04 (#1268)
- Bugfix for piecewise functions (#1199)
- Refactor swigging - generate one single wrapper (#1213)
- Fix warnings: account for zero indexing in nan/inf error (#1112)
- Update Windows build instructions (#1200, #1202)
- Update README: Projects using AMICI (#1209)
- Add CODE_OF_CONDUCT.md (#1210)
- Update documentation for Python interface (#1208)
- Create sdist on GHA using swig4.0.1 (#1204) (Fixing broken pypi package)
- Fix links after repository move
- Speed-up swig build: disable all languages except python (#1211)
- Fix doc generation on readthedocs (#1196)
- Move repo to new organization (#1193)
- Update Bibliography
- Fix bug for energyPySB models (#1191)
- Fix release deployment (#1189)
- Skip unnecessary expressions in pysb models (#1185)
- MSVC compiler support (this time for real... #1152)
- Implement MSVC tests (#1152)
- Rename and group GitHub actions (#1186)
- Fix release deployment (#1186)
- Fix simplification for pysb models (#1168)
- Pass verbosity flags to pysb network generation (#1173)
- Enable experimental pysb-petab support (#1175)
- Add installation instructions for Fedora (#1177)
- Implement support for SBML rate-references (#1180)
- Refactoring (#1162, #1163)
- Move majority of tests to Github Actions (#1166, #1160)
- Improve reporting of skipped tests in SBML testsuite (#1183)
- Speed up model import, compilation (#1123, #1112)
- Improve/Add steady-state solver documentation (#1102)
- Improve extension import (#1141)
- Bugfixes SBML import (#1135, #1134, #1145, #1154)
- Fixed issue that prevented simplification (#1158)
- Bugfixes (#1121, #1125, #1131, #1132, #1136)
- Enable openMP by default (#1118)
- Improve memoy footprint for simulations with replicates (#1153)
- Improve steady-state solver and add option to to adjoint-steadystate hybrid (#1143, #1099, #1129, #1146)
- Store build artifacts from github actions (#1138)
- Upgrade to sympy 1.6.0, which is now required minimum version (#1098, #1103)
- Speed up model import
- Speed-up computation of sx0, reduce file size (#1109)
- Replace terribly slow sympy.MutableDenseMatrix.is_zero_matrix by custom implementation (#1104)
- speedup dataframe creation in
get*AsDataFrame
(#1088) - Allow caching edatas for simulate_petab (#1106)
- Fix wrong deprecation warning (Fixes #1093)
- Fix segmentation faults in NewtonSolver under certain conditions (#1089, #1090, #1097)
- fix wrong power function call in
unscale_parameter
(#1094) - Fix MathML conversion (#1086)
- Fix deepcopy of SymPy objects (#1091)
- handle empty rdata->{pre|post}eq_numlinsteps (Closes #1113), which previously made the matlab interface unusable
- Fix generation of compileMexFile.m for matlab compilation of python code (#1115)
- Reduce memory requirements and speedup compilation of large models (#1105)
- Place generated code into own namespace (#937) (#1112)
- Fix several msvc compiler warnings (#1116) (Note that MSVC support is still experimental) breaking change for users of C++ interface
- Fix swig warning: ensure base class ContextManager is known before use (Fixes #1092) (#1101)
- Don't install/run valgrind on travis CI (done with github actions… (#1111)
Python:
- Implement support for variable compartments (#1036)
- Better handling of constant species (#1047)
- Better handling of C++ enums, this makes
amici.SensitivityMethod_forward
available asamici.SensitivityMethod.forward
(#1042) - Improve installation routines (#1055, #1056, #1058, #1076)
- Add option to reduce memory usage (#1044)
- Fix handling of symbolic expressions in nested rules (#1081, 1069)
Library:
- Update Sundials to 5.2.0 (#1039)
- Update SuiteSparse to 5.4.0 (#1040)
- Refactor use of ReturnData, now completely created post-hoc (#1002)
- Fix propagation of reinitialization in ExpData constructor (#1041)
- Fix issue where InternalSensitivityParameter was sometimes not set (#1075)
- Fix or disable certain combinations of equilibraition, presimulation and adjoint sensitivity analysis
CI:
- Move from Codacy to Sonarcloud (#1065)
- Run SBML Testsuite when appropriate (#1058)
Library:
- Fix: Handle paths with blanks in build scripts
- Feature: Add function to write amici::Solver settings to HDF5 (#1023)
- Fix: typehints (#1018, #1022)
- Refactor: Move creation of parameter mapping for objective<->simulation to classes (#1020)
CI:
- Refactor: Cleanup and reorganize tests (#1026)
- Fix: benchmark problem test should fail on missing files (Closes #1015)
- Fixed (re)initialization of sensitivities if ExpData::fixedParametersPreequilibration is set (#994)
- Fixed sensitivities for parameters in sigma expressions for Python/SBML in case provided expression was not just a single parameter ID
- Enable parallel compilation of model files from Python (#997) based on AMICI_PARALLEL_COMPILE enviroment variable
- Fixed computation of log-likelihood for log10-normal distributed noise
- Added
reinitializeFixedParameterInitialStates
to ExpData (#1000) (breaking change: overrides settings inamici::Model
) - Python model import now verifies that chosen model name is a valid identifier (Closes #928)
- Made w available in ReturnData (Closes #990) (#992)
- Fixed setting of log level when passing boolean values to verbose (#991)
- Documentation now on ReadTheDocs https://amici.readthedocs.io/en/
- Use proper state/observable names in plotting functions (#979)
- PEtab support:
- Adapt to most recent PEtab (0.1.5)
- Extended support for import of PEtab models
- Added support for computing cost function based on PEtab problem
- Implemented handling of species in condition table
- petab_import.import_model now provides reproducible parameter list (Closes #976)
- Fix python import error in import_petab_problem: Add absolute paths to python path, invalidate caches and reload (#970)
- Added example notebook
- CI: PEtab test suite integrated in CI workflow
- Added AMICI dockerfile and image deployment to dockerhub (#948)
- Removed mention of 'mex' in warning/error ids (#968)
- More informative errors on SWIG interface import failures (#959)
Python:
- Fix logo display on pypi
- Fix deadlocks in multithreaded python environments when using openMP parallelization
Matlab:
- Fix compilation errors due to switch to C++14
General:
- AMICI now comes with a logo
- implement getName function for models
- Updated documentation / examples
Python:
- Enable MSVC compilation of Python extensions (#847)
- Always recompile clibs and extensions (Closes #700)
- Extended PEtab support (Running
- enable multithreading in swig (#938)
- Fixes pysb (#902) (#907)
C++
- Build optimized AMICI and sundials by default (Closes #934)
Matlab:
- Fix(matlab) Compile CalcMD5 on demand (Fixes #914)
- Don't pass empty include strings to mex
- Fix Matlab compilation error if AMICI or model path contains blanks
CI:
- Running additional test models
... and various minor fixes/updates
- added python 3.8 support, dropped python 3.6 support (#898)
- Added logging functionality (#900)
- Fixes PySB import (#879, #902)
- Fixes symbolic processing (#899)
- Improved build scripts (#894,
- Improved petab support (#886, #888, #891)
- CI related fixes (#865, #896)
- Sparsify dwdp to reduce computation time for adjoints (#858)
- Fix(matlab) update example name example_dae_events->example_calvetti (Closes #866)
- Fix nullptr deferencing for simulations with events when no measurements are provided (Fixes #866)
- Fix accessing empty vector during adjoint state event update (Closes #866)
- Fix pysb_import (fixes #878)
Bugfix release due to incorrect sensitivities w.r.t. sigmas introduced in 0.10.14.
No other changes.
NOTE: For Python-imported SBML-models this release may compute incorrect sensitivities w.r.t. sigma. Bug introduced in 0.10.14, fixed in 0.10.15.
Python:
- Don't require use of ModelPtr.get to call ExpData(Model)
- Fix import in generated model Python package
- Setup AMICI standalone scripts as setuptools entrypoints
- Simplify symbolic sensitivity expressions during Python SBML import Fixes Infs in the Jacobian when using Hill-functions with states of 0.0.
- Extended Newton solver #848 The changes that allow performing Newton tests from the paper: G. T. Lines, Ł. Paszkowski, L. Schmiester, D. Weindl, P. Stapor, and J. Hasenauer. Efficient computation of steady states in large-scale ODE models of biochemical reaction networks. accepted for Proceedings of the 8th IFAC Conference on Foundations of Systems Biology in Engineering (FOSBE), Valencia, Spain, October 2019.
- Use SWIG>=4.0 on travis to include PyDoc in sdist / pypi package (#841)
- Fix choice of likelihood formula; failed if observable names were not equal to observable IDs
- Fix(sbml-import) Compartment IDs in right-hand side of Rules are not replaced and lead to undefined identifiers in c++ files
- Fix invalid logging level
- Speed up sympy simplification (#871)
C++:
- Performance: Avoid unnecessary repeated function calls for SUNMatrixWrapper dimensions
- Add AmiciApplication class as context for handling so far global settings. This allows for example setting custom logging functions for concurrent AMICI runs, e.g. in multi-thread applications (Closes #576).
Misc:
- Setup performance test on github actions (#853)
- Update documentation and FAQ for CBLAS requirement and others
- Update reference list
- BREAKING CHANGE: Renaming {get|set}tNewtonPreequilibration to {get|set}Preequilibration (Closes #720)
- Make wurlitzer non-optional requirement for AMICI python package (Fixes missing AMICI errors when running from jupyter notebooks)
- Compute initial state for Model::getInitialStates if not already set (Fixes #818)
- Make swig generate pydoc comments from doxygen comments #830 (Closes #745) to provide Python docstrings for C++ wrapper functions
- feature(cmake) Add option to disable compiler optimizations for wrapfunctions.cpp (Fixes #828) (#829)
- Change SBML test suite to pytest to allow for parallel test execution… (#824)
- Fix(cmake): -E option is not available in all sed versions. Neither is the equivalent -r. Use --regexp-extended instead (Closes #826)
- Refactor(python) Move PEtab import code from command line script… (#825)
- Fix(core) Fix regular expressions for intel compiler (Closes #754) (#822)
- Update workflow figure to include PySB (Closes #799)
- Fix compiler warnings
- Fix handling of species specified in PEtab condition table (#813)
- Fix some Visual C++ issues, update cppcheck handling, cleanup (VisualC++ still not fully supported)
- Minor fixups (#801)
- Create SBML test suite result files for upload to http://sbml.org/Facilities/Database/ (#798)
- Fixed setting initial conditions for preequilibration (#784)
- Fixed species->parameter conversion during PEtab import (#782)
- Set correct Matlab include directories in CMake (#793)
- Extended and updated documentation (#785, #787)
- Fix various SBML import issues
- Run SBML test suite using github actions instead of travisCI (#789)
-
Simplify/fix AMICI installation
-
If available use environment modules to detect dependencies
-
Add SWIG installation script
-
-
Update list of publication
-
Update documentation
- Update doc for SWIG build and custom SWIG location.
- Add AMICI interface overview / workflow figure and show in README
- Document environment variables for model/core compilation (Closes #737)
-
Added handling of abs function, since there seem to be problems with case sensitivity (#713) Closes #770
Detaills: * cmake: Use package_ROOT environment variables * fix(cmake) Fix finding version.txt * cmake: Auto-detect loaded MKL environment module * cmake: Use new FindPython3 modules where possible * fix(python) Restore python3.6 compatibility * Inside venv, use pip instead of pip3 which should point to the correct version * fix(python) Workaround for missing ensurepip during venv creation [ci skip] * feature(python) Use MKL from environment modules to provide cblas * fix(python) Fix define_macros not being passed to setuptools for Extension * fix(python) Fix define_macros not being passed to setuptools for clibs * Do not always add 'cblas' library since users may want to override that by a cblas-compatible library with a different name (closes #736) * Update HDF5 path hints; use shared library if static is not available. * Check for HDF5_BASE from environment module * Fix system-dependent sundials library directory (Fixes #749) (#750) * Handle OSTYPE==linux in scripts/buildBNGL.sh (Fixes #751) * Add SWIG download and build script * Improve finding swig executable and allow user override via SWIG environment variable * Provide installation hints if no SWIG found (Closes #724) * Allow overriding cmake executable with environment variables in build scripts (Closes #738)
Fixup for missing version bump in v0.10.8 release. No code changes compared to v0.10.8.
Changes in this release:
All:
- Updated / extended documentation
- Fix reuse of
Solver
instances (#541)
C++:
- Check for correct AMICI version for model in CMake
- Add reporting of computation times (#699)
Python:
- Fix manifest file (#698)
- Fix initial amounts/concentrations in SBML import
... and various other minor fixes/improvements
Python
- fix unset noise distribution when automatically generating observables in case None are passed (#691)
C++
- Add SuperLUMT support (#681)
- Sparsified dJydy (#686)
- Enabled support of impulse-free events for DAE models (#687) - thanks to Sebastien Sten for providing a testcase for this
Python
- Enabled support for piecewise functions in SBML import (#662)
- Fix numeric type when constructing ExpData from Dataframes (#690)
- Fix dynamic override in PETab
Bugfix release
Doc
- Update documentation of Windows installation
C++
- Fix missing source files in CMakeLists.txt (#658)
- Set CMake policies to prevent warnings (Closes #676) (#677)
- Start using gsl::span instead of raw pointers (#393) (#678)
Python
- PySB parsing fix (#669)
- Fix failure to propagate BLAS_LIBS contents (#665)
- Require setuptools at setup (#673)
- Updated PEtab import to allow for different noise models
Features / improvements:
- Implement ReturnData and ExpData wrappers as more efficient views (#657)
- Add list of references using AMICI (#659)
- Custom llh (normal/laplace, lin/log/log10) (#656)
Bugfixes:
- Speedup and fix travis build
Features / improvements:
- adds the option for early termination on integration failures for runAmiciSimulations
- improve runtime of
SUNMatrixWrapper::mutliply
- expose finite difference routines in public API
- enable parallel compilation of clib source files
Bugfixes:
- fixed symbolic processing for unreleased pysb features
Features / improvements:
- extended
ExpData
interface to allow for condition specific parameters, parameter scales, parameter lists, initial conditions and initial condition sensitivities.
Bugfixes:
- fixed output values of
ReturnData::x_ss
andReturnData::sx_ss
- travis-ci.com migration
- fix problem with has{variable} functions
- allow to import sbml model from string, not only file
Features / improvements:
- updated sundials to 4.1.0
- updated SuiteSparse to 5.4.0
- added generic implementations for symbolic expressions that were sparse matrix vector products
Bugfixes:
- fixed return value of
rz
when no data is provided.
Features / improvements:
- allow python installations without compilation of c++ extension
- improvements to ExpData <-> pandas.DataFrame interface
- allow generation of matlab models from python
- implement CLI interface for PEtab
- improve computation time for conservation laws from pysb import
Bugfixes:
- Fix sign in undamped Newton step.
Maintenance:
- use newer CI images
Minor fixes only:
- fix(core) Get solver diagnostics for first(last) timepoint (#588) (Closes #586)
- fix(ci) Fix autodeploy (Closes #589)
CRITICAL FIXES
- fix(python) fix symbolic computations for adjoint (#583)
Features
- feature(python) Check for matching AMICI versions when importing model (Closes #556). Set exact AMICI version as model package requirement.
- feature(core) Add option to rethrow AMICI exception (Closes #552)
- feature(python) Redirect C/C++ output in stdout is redirected (e.g. in ipython notebooks) (Closes #456)
Minor fixes
- fix(python) Fix doc and rename sys_pipes to something more meaningful
- fix(ci) Fix premature exit of scripts/runNotebook.sh
- fix(deploy) Update pyenv shims to find twine
Bugfixes:
- fixes a critical bug in the newton solver
- fixes multiple bugs in sbml import for degenerate models, empty stoichiometry assignments and conversion factors
- improved error messages for sbml import
- #560
- #557
- #559
Features / improvements:
- make pure steadystate results available as
rdata['x_ss']
andrdata['sx_ss']
- add option to specify integration tolerances for the adjoint problem via
atolB
andrtolB
Bugfixes:
- improved conservation law identification to also account for constant species.
- fixed a bug where simulation results were written into results of the second newton solver attempt
- fixed an openMP related warning
Maintenance:
- attempt to fix automatic deploy to pypi via travis
Features / improvements:
- Allow computation and application of conservation laws for pysb importet models. This enables use of NewtonSolver for preequilibration for models where it was previously not possible.
- Use
klu_refactor
in the sparse Newton solver instead of always usingklu_factor
and only perform symbolic factorization once (#421) - Allow more detailed finiteness checks (#514)
Bugfixes:
- #491
Maintenance:
- Several improvements to travis log sizes and folding
- Use default copy constructor for Model by implementing class wrappers for sundials matrix types (#516)
- Reenable run of SBML testsuite
Features / improvements:
- Speedup symbolic processing for ODE generation in python
Bugfixes:
- Fix(python) Add missing deepcopy (introduced in 6847ba675f2088854db583199b8754aaa6e01576)
- Fix(core) Prevent parameter scaling length mismatch (#488)
- Fix(python) Set distutils dependency to current version to fix </usr/lib/python3.6/distutils/dist.py:261: UserWarning: Unknown distribution option: 'long_description_content_type'>
- fix(python) add symlink to version.txt to be included in package
Backwards-compatibility:
- Replace 'newline' by literal to restore <R2016b compatibility (Fixes #493)
Maintenance:
- Remove obsolete swig library build via cmake and related file copying
- Provide issue template for bug reports
- Providing valid SBML document to import is not optional anymore
- Update documentation and tests
- Add python version check and raise required version to 3.6 to prevent cryptic error messages when encountering f-strings
- [all] critical Fix long standing bugs in solving steadystate problems (including preequilibration) (#471)
- [all] Fix AmiVector constructor from std::vector (#471)
- [python] Reenable Solver and Model copy constructors
- Update documentation
- replaced symengine by sympy for symbolic processing in python which fixes several critical bugs that were due to bugs in symengine (#467)
- fixes a critical bug in objective function computation for models compiled using
sbml2amici
andpysb2amici
that was introduced in v0.7.12 - fixes a critical bug in sensitivity computation when
model.reinitializeFixedParameterInitialStates
was set to true - readds the python interface to the ExpData copy constructor that was inadvertently removed in 0.7.12 and streamlines the respective convenience wrapper to provide access to the full range of constructors.
- fixed a critical bug in
amici.constructEdataFromDataFrame
- enabled multithreaded simulation of multiple experiments (requires compilation with openMP)
- modularized sbml import and added pysb import
- [python] Added numpy and python wrappers that provide a more user friendly python API
- [python] Enable import of SBML models with non-float assignment rules
- [python] Enable handling of exceptions in python
- [python] Enable nativ python access to std::vector data-structures
- [core] Provide an API for more fine-grained control over sensitivity tolerances and steady-state tolerances
- [core] Provide an API to specify non-negative state variables (this feature is still preliminary)
- Fixed python SBML import
log()
issues (#412)
- fixes MATLAB compilation of models
- adds option to perform steady state sensitivity analysis via FSA
- condition dependent intitial conditions are now newly set after preequilibration is done
- bugfixes for the ExpData interface
- created build configuration that enables debugging of c++ extensions on os x
- fixed python sbml import when stoichiometry is empty
Fixes a couple of bugs just introduced in v0.7.6
Important: Use AMICI v0.7.7 due to https://github.com/ICB-DCM/AMICI/pull/403/commits/3a495d3db2fdbba70c2b0d52a3d4655c33c817a2
Bug fixes:
- Python import: Fix log10 issues in observables (#382)
- Matlab: Fix broken model compilation (#392)
- Fixed simulation for models without observables (#390)
- Fixed potential matlab memory leaks (#392)
Breaking C++ API changes:
- Revised ExpData interface (#388)
Features/enhancements:
- Add computation of residuals, residuals sensitivity, Fisher information matrix (#223)
- More efficient conversion of std::vector to numpy ndarray (#375)
- Allow specifying timepoints in ExpData (#370)
Minor fixes:
- Condition parameters in ExpData now only temporarily override Model parameters (#371)
- Ensure non-negative states for Newton solver (#378)
Features/enhancements:
- Check SBML model validity (#343)
- Allow per-parameter setting of amioptions::pscale from matlab interface (#350)
- Documentation
Major fixes:
- Don't compile main.cpp into python model module which broke modules if amici was compiled without libhdf5 (#363)
Minor fixes:
- Fix compiler warnings (#353)
- Plotting, SBML example mode, ...
Features:
- Added symbol names to python-wrapped models and make available via Model.getParameterNames(), model.getStateNames(), ...
- Extended Python interface example
Python package available via pypi: https://pypi.org/project/amici/0.7.3/
Features:
- Python package: more flexible HDF5 library localization
- Extended CI: python tests, preequilibration tests, run in venv
Major bugfixes:
- Fix python sbml model import / compilation error (undefined function)
- Fix model preequilibration
Minor fixes:
- Various fixes for mingw compilation of python source distribution
- Cmake compatibility with < 3.8 restored
Features:
- Allow specifying sigma-parameters from Python interface
Major bugfixes:
- Fix dsigma_y/dp and downstream sensitivity errors
- Major revision of documentation
- Improved Python interface
- More comprehensive Python interface example
- Fixed sensitivity computation in Python-generated models
- Various other bug fixes
WARNING:
- For models with sigma-parameters and dsigma_y/dp != 0, dsigma_y/dp was computed incorrectly. This propagates to all dependent sensitivities. This applies also to some older releases and has been fixed in v0.7.1.
Implement experimental support for python via swig. Python interface is now usable, but API will still receive some updates in the future.
WARNING:
- There is a bug in sensitivity computation for Python-generated models
- Matlab C++ compilation will fail due to undefined M_PI -> Please use v0.7.0
Main new features are:
-
Reimplemented support for DAE equations
-
Added newton solver for steady state calculation and preequilibration
-
Better caching for recompilation of models
-
Blas support to allow compilation of large models with many observables
-
Improved SBML support
-
Added c++ interface
-
Added support for second order adjoint computation
-
Rewrote large parts of the code as proper c++11 code to allow easier code maintanance
-
Substantially extended testing in continuous integration to assure code quality
- First citable version of AMICI (via zenodo integration).
- Better support for standalone compilation
- Improved SBML import scripts
- General Code Cleanup
This update comes with many improvements, bug fixes and several new features. Most notably:
- AMICI should now run on older versions of MATLAB (back to R2014a)
- AMICI now compiles using older versions of Visual Studio
- AMICI now also supports second order adjoint sensitivities (full (via the o2flag = 1 and as a vector product via o2flag = 2)
- AMICI now supports more SBML, SBML v2 and rate rules
Bugfix release. This release also includes some changes that should improve the performance on the new R2016a release of MATLAB.
This update comes with many improvements to the computation time for both compilation and simulation. Moreover several new features were included:
- Hessian Vector products for second order forward sensitivities
- Correct treatment of parameter/state dependent discontinuities for both forward and adjoint sensitivities
This is the initial release of the toolbox