CHANGELOG.md

Unreleased

v23.9rc0 - 2023-10-31

Features

• The parameter "Ambient temperature [K]" can now be given as a function of position (y,z) and time t. The "edge" and "current collector" heat transfer coefficient parameters can also depend on (y,z) (#3257)
• Spherical and cylindrical shell domains can now be solved with any boundary conditions (#3237)
• Processed variables now get the spatial variables automatically, allowing plotting of more generic models (#3234)
• Numpy functions now work with PyBaMM symbols (e.g. np.exp(pybamm.Symbol("a")) returns pybamm.Exp(pybamm.Symbol("a"))). This means that parameter functions can be specified using numpy functions instead of pybamm functions. Additionally, combining numpy arrays with pybamm objects now works (the numpy array is converted to a pybamm array) (#3205)
• Half-cell models where graphite - or other negative electrode material of choice - is treated as the positive electrode (#3198)
• Degradation mechanisms SEI, SEI on cracks and lithium plating can be made to work on the positive electrode by specifying the relevant options as a 2-tuple. If a tuple is not given and working electrode is set to both, they will be applied on the negative electrode only. (#3198)
• Added an example notebook to demonstrate how to use half-cell models (#3198)
• Added option to use an empirical hysteresis model for the diffusivity and exchange-current density (#3194)
• Double-layer capacity can now be provided as a function of temperature (#3174)
• pybamm_install_jax is deprecated. It is now replaced with pip install pybamm[jax] (#3163)
• Implement the MSMR model (#3116)
• Added new example notebook rpt-experiment to demonstrate how to set up degradation experiments with RPTs (#2851)

Bug fixes

• Fixed a bug where empty lists passed to QuickPlot resulted in an IndexError and did not return a meaningful error message (#3359)
• Fixed a bug where there was a missing thermal conductivity in the thermal pouch cell models (#3330)
• Fixed a bug that caused incorrect results of “{Domain} electrode thickness change [m]” due to the absence of dimension for the variable electrode_thickness_change(#3329).
• Fixed a bug that occured in check_ys_are_not_too_large when trying to reference y-slice where the referenced variable was not a pybamm.StateVector (#3313
• Fixed a bug with _Heaviside._evaluate_for_shape which meant some expressions involving heaviside function and subtractions did not work (#3306)
• Attributes of pybamm.Simulation objects (models, parameter values, geometries, choice of solver, and output variables) are now private and as such cannot be edited in-place after the simulation has been created (#3267
• Fixed bug causing incorrect activation energies using create_from_bpx() (#3242)
• Fixed a bug where the "basic" lithium-ion models gave incorrect results when using nonlinear particle diffusivity (#3207)
• Particle size distributions now work with SPMe and NewmanTobias models (#3207)
• Attempting to set working electrode to negative now triggers an OptionError. Instead, set it to positive and use what would normally be the negative electrode as the positive electrode. (#3198)
• Fix to simulate c_rate steps with drive cycles (#3186)
• Always save last cycle in experiment, to fix issues with starting_solution and last_state (#3177)
• Fix simulations with starting_solution to work with start_time experiments (#3177)
• Fix SEI Example Notebook (#3166)
• Thevenin() model is now constructed with standard variables: Time [s], Time [min], Time [h] (#3143)
• Error generated when invalid parameter values are passed (#3132)
• Parameters in Prada2013 have been updated to better match those given in the paper, which is a 2.3 Ah cell, instead of the mix-and-match with the 1.1 Ah cell from Lain2019 (#3096)
• The OneDimensionalX thermal model has been updated to account for edge/tab cooling and account for the current collector volumetric heat capacity. It now gives the correct behaviour compared with a lumped model with the correct total heat transfer coefficient and surface area for cooling. (#3042)

Optimizations

• Improved how steps are processed in simulations to reduce memory usage (#3261)
• Added parameter list support to JAX solver, permitting multithreading / GPU execution (#3121)

Breaking changes

• Dropped support for i686 (32-bit) architectures on GNU/Linux distributions (#3412)
• The class pybamm.thermal.OneDimensionalX has been moved to pybamm.thermal.pouch_cell.OneDimensionalX to reflect the fact that the model formulation implicitly assumes a pouch cell geometry (#3257)
• The "lumped" thermal option now always used the parameters "Cell cooling surface area [m2]", "Cell volume [m3]" and "Total heat transfer coefficient [W.m-2.K-1]" to compute the cell cooling regardless of the chosen "cell geometry" option. The user must now specify the correct values for these parameters instead of them being calculated based on e.g. a pouch cell. An OptionWarning is raised to let users know to update their parameters (#3257)
• Numpy functions now work with PyBaMM symbols (e.g. np.exp(pybamm.Symbol("a")) returns pybamm.Exp(pybamm.Symbol("a"))). This means that parameter functions can be specified using numpy functions instead of pybamm functions. Additionally, combining numpy arrays with pybamm objects now works (the numpy array is converted to a pybamm array) (#3205)
• The SEI, SEI on cracks and lithium plating submodels can now be used on either electrode, which means the __init__ functions for the relevant classes now have domain as a required argument (#3198)
• Likewise, the names of all variables corresponding to those submodels now have domains. For example, instead of SEI thickness [m], use Negative SEI thickness [m] or Positive SEI thickness [m]. (#3198)
• If options["working electrode"] == "both" and either SEI, SEI on cracks or lithium plating are not provided as tuples, they are automatically made into tuples. This directly modifies extra_options, not default_options to ensure the other changes to default_options still happen when required. (#3198)
• Added option to use an empirical hysteresis model for the diffusivity and exchange-current density (#3194)
• Double-layer capacity can now be provided as a function of temperature (#3174)
• pybamm_install_jax is deprecated. It is now replaced with pip install pybamm[jax] (#3163)
• PyBaMM now has optional dependencies that can be installed with the pattern pip install pybamm[option] e.g. pybamm[plot] (#3044)

v23.5 - 2023-06-18

Features

• Idaklu solver can be given a list of variables to calculate during the solve (#3217)
• Enable multithreading in IDAKLU solver (#2947)
• If a solution contains cycles and steps, the cycle number and step number are now saved when solution.save_data() is called (#2931)
• Experiments can now be given a start_time to define when each step should be triggered (#2616)

Optimizations

• Test JaxSolver's compatibility with Python 3.8, 3.9, 3.10, and 3.11 (#2958)
• Update Jax (0.4.8) and JaxLib (0.4.7) compatibility (#2927)
• Migrate from tox=3.28 to nox (#3005)
• Removed importlib_metadata as a required dependency for user installations (#3050)

Bug fixes

• Realign 'count' increment in CasadiSolver._integrate() (#2986)
• Fix pybamm_install_odes and update the required SUNDIALS version (#2958)
• Fixed a bug where all data included in a BPX was incorrectly assumed to be given as a function of time.(#2957)
• Remove brew install for Mac from the recommended developer installation options for SUNDIALS (#2925)
• Fix bpx.py to correctly generate parameters for "lumped" thermal model (#2860)

Breaking changes

• Deprecate functionality to load parameter set from a csv file. Parameter sets must now be provided as python dictionaries (#2959)
• Tox support for Installation & testing has now been replaced by Nox (#3005)

v23.4.1 - 2023-05-01

Bug fixes

• Fixed a performance regression introduced by citation tags (#2862). Citations tags functionality is removed for now.

v23.4 - 2023-04-30

Features

• Added verbose logging to pybamm.print_citations() and citation tags for the pybamm.Citations class so that users can now see where the citations were registered when running simulations (#2862)
• Updated to casadi 3.6, which required some changes to the casadi integrator (#2859)
• PyBaMM is now natively supported on Apple silicon chips (M1/M2) (#2435)
• PyBaMM is now supported on Python 3.10 and 3.11 (#2435)

Optimizations

• Fixed deprecated interp2d method by switching to xarray.DataArray as the backend for ProcessedVariable (#2907)

Bug fixes

• Initial conditions for sensitivity equations calculated correctly (#2920)
• Parameter sets can now contain the key "chemistry", and will ignore its value (this previously would give errors in some cases) (#2901)
• Fixed keyerror on "all" when getting sensitivities from IDAKLU solver(#2883)
• Fixed a bug in the discretisation of initial conditions of a scaled variable (#2856)

Breaking changes

• Made Jupyter a development only dependency. Now Jupyter would not be a required dependency for users while installing PyBaMM. (#2846)

v23.3 - 2023-03-31

Features

• Added option to limit the number of integrators stored in CasadiSolver, which is particularly relevant when running simulations back-to-back #2823
• Added new variables, related to electrode balance, for the ElectrodeSOH model (#2807)
• Added method to calculate maximum theoretical energy. (#2777) and add to summary variables (#2781)
• Renamed "Terminal voltage [V]" to just "Voltage [V]". "Terminal voltage [V]" can still be used and will return the same value as "Voltage [V]". (#2740)
• Added "Negative electrode surface potential difference at separator interface [V]", which is the value of the surface potential difference (phi_s - phi_e) at the anode/separator interface, commonly controlled in fast-charging algorithms to avoid plating. Also added "Positive electrode surface potential difference at separator interface [V]". (#2740)
• Added "Bulk open-circuit voltage [V]", which is the open-circuit voltage as calculated from the bulk particle concentrations. The old variable "Measured open circuit voltage [V]", which referred to the open-circuit potential as calculated from the surface particle concentrations, has been renamed to "Surface open-circuit voltage [V]". (#2740) "Bulk open-circuit voltage [V]" was briefly named "Open-circuit voltage [V]", but this was changed in (#2845)
• Added an example for plot_voltage_components, explaining what the different voltage components are. (#2740)

Bug fixes

• Fix non-deteministic outcome of some tests in the test suite (#2844)
• Fixed excessive RAM consumption when running multiple simulations (#2823)
• Fixed use of last_state as starting_solution in Simulation.solve() (#2822)
• Fixed a bug where variable bounds could not contain InputParameters (#2795)
• Improved model.latexify() to have a cleaner and more readable output (#2764)
• Fixed electrolyte conservation in the case of concentration-dependent transference number (#2758)
• Fixed plot_voltage_components so that the sum of overpotentials is now equal to the voltage (#2740)

Optimizations

• Migrated to Lychee workflow for checking URLs (#2734)

Breaking changes

• ElectrodeSOH.solve now returns a {str: float} dict instead of a pybamm.Solution object (to avoid having to do .data[0] every time). In any code that uses sol = ElectrodeSOH.solve(), sol[key].data[0] should be replaced with sol[key]. (#2779)
• Removed "... cation signed stoichiometry" and "... electrons in reaction" parameters, they are now hardcoded. (#2778)
• When using solver.step(), the first time point in the step is shifted by pybamm.settings.step_start_offset (default 1 ns) to avoid having duplicate times in the solution steps from the end of one step and the start of the next. (#2773)
• Renamed "Measured open circuit voltage [V]" to "Surface open-circuit voltage [V]". This variable was calculated from surface particle concentrations, and hence "hid" the overpotential from particle gradients. The new variable "Bulk open-circuit voltage [V]" is calculated from bulk particle concentrations instead. (#2740)
• Renamed all references to "open circuit" to be "open-circuit" instead. (#2740)
• Renamed parameter "1 + dlnf/dlnc" to "Thermodynamic factor". (#2727)
• All PyBaMM models are now dimensional. This has been benchmarked against dimensionless models and found to give around the same solve time. Implementing dimensional models greatly reduces the barrier to entry for adding new models. However, this comes with several breaking changes: (i) the timescale and length_scales attributes of a model have been removed (they are no longer needed) (ii) several dimensionless variables are no longer defined, but the corresponding dimensional variables can still be accessed by adding the units to the name (iii) some parameters used only for non-dimensionalization, such as "Typical current [A]", have been removed (#2419)

v23.2 - 2023-02-28

Features

• Added an option for using a banded jacobian and sundials banded solvers for the IDAKLU solve (#2677)
• The "particle size" option can now be a tuple to allow different behaviour in each electrode (#2672).
• Added temperature control to experiment class. (#2518)

Bug fixes

• Fixed current_sigmoid_ocp to be valid for both electrodes (#2719).
• Fixed the length scaling for the first dimension of r-R plots (#2663).

v23.1 - 2023-01-31

Features

• Changed linting from flake8 to ruff (#2630).
• Changed docs theme to pydata theme and start to improve docs in general (#2618).
• New contact resistance option, new parameter Contact resistance [Ohm] and new variable Contact overpotential [V] (#2598).
• Steps in Experiment can now be tagged and cycle numbers be searched based on those tags (#2593).

Bug fixes

• Fixed a bug where the solid phase conductivity was double-corrected for tortuosity when loading parameters from a BPX file (#2638).
• Changed termination from "success" to "final time" for algebraic solvers to match ODE/DAE solvers (#2613).

v22.12 - 2022-12-31

Features

• Added functionality to create pybamm.ParameterValues from a BPX standard JSON file (#2555).
• Allow the option "surface form" to be "differential" in the MPM (#2533)
• Added variables "Loss of lithium due to loss of active material in negative/positive electrode [mol]". These should be included in the calculation of "total lithium in system" to make sure that lithium is truly conserved. (#2529)
• initial_soc can now be a string "x V", in which case the simulation is initialized to start from that voltage (#2508)
• The ElectrodeSOH solver can now calculate electrode balance based on a target "cell capacity" (requires cell capacity "Q" as input), as well as the default "cyclable cell capacity" (requires cyclable lithium capacity "Q_Li" as input). Use the keyword argument known_value to control which is used. (#2508)

Bug fixes

• Allow models that subclass BaseBatteryModel to use custom options classes (#2571)
• Fixed bug with EntryPoints in Spyder IDE (#2584)
• Fixed electrolyte conservation when options {"surface form": "algebraic"} are used
• Fixed "constant concentration" electrolyte model so that "porosity times concentration" is conserved when porosity changes (#2529)
• Fix installation on Google Colab (pybtex and Colab issue) (#2526)

Breaking changes

• Renamed "Negative/Positive electrode SOC" to "Negative/Positive electrode stoichiometry" to avoid confusion with cell SOC (#2529)
• Removed external variables and submodels. InputParameter should now be used in all cases (#2502)
• Trying to use a solver to solve multiple models results in a RuntimeError exception (#2481)
• Inputs for the ElectrodeSOH solver are now (i) "Q_Li", the total cyclable capacity of lithium in the electrodes (previously "n_Li", the total number of moles, n_Li = 3600/F * Q_Li) (ii) "Q_n", the capacity of the negative electrode (previously "C_n"), and "Q_p", the capacity of the positive electrode (previously "C_p") (#2508)

v22.11.1 - 2022-12-13

Bug fixes

• Fixed installation on Google Colab (pybtex issues) (#2547)

v22.11 - 2022-11-30

Features

• Updated parameter sets so that interpolants are created explicitly in the parameter set python file. This does not change functionality but allows finer control, e.g. specifying a "cubic" interpolator instead of the default "linear" (#2510)
• Equivalent circuit models (#2478)
• New Idaklu solver options for jacobian type and linear solver, support Sundials v6 (#2444)
• Added scale and reference attributes to Variable objects, which can be use to make the ODE/DAE solver better conditioned (#2440)
• SEI reactions can now be asymmetric (#2425)

Bug fixes

• Switched from pkg_resources to importlib_metadata for handling entry points (#2500)
• Fixed some bugs related to processing FunctionParameter to Interpolant (#2494)

Optimizations

• ParameterValues now avoids trying to process children if a function parameter is an object that doesn't depend on its children (#2477)
• Implemented memoization via cache and cached_property from functools (#2465)
• Added more rules for simplifying expressions, especially around Concatenations. Also, meshes constructed from multiple domains are now cached (#2443)
• Added more rules for simplifying expressions. Constants in binary operators are now moved to the left by default (e.g. x*2 returns 2*x) (#2424)

Breaking changes

• Interpolants created from parameter data are now "linear" by default (was "cubic") (#2494)
• Renamed entry point for parameter sets to pybamm_parameter_sets (#2475)
• Removed code for generating ModelingToolkit problems (#2432)
• Removed FirstOrder and Composite lead-acid models, and some submodels specific to those models (#2431)

v22.10.post1 - 2022-10-31

Breaking changes

• Removed all julia generation code (#2453). Julia code will be hosted at PyBaMM.jl from now on.

v22.10 - 2022-10-31

Features

• Third-party parameter sets can be added by registering entry points to pybamm_parameter_setpybamm_parameter_sets (#2396, changed in #2475)
• Added three-dimensional interpolation (#2380)

Bug fixes

• pybamm.have_julia() now checks that julia is properly configured (#2402)
• For simulations with events that cause the simulation to stop early, the sensitivities could be evaluated incorrectly to zero (#2337)

Optimizations

• Reformatted how simulations with experiments are built (#2395)
• Added small perturbation to initial conditions for casadi solver. This seems to help the solver converge better in some cases (#2356)
• Added ExplicitTimeIntegral functionality to move variables which do not appear anywhere on the rhs to a new location, and to integrate those variables explicitly when get is called by the solution object. (#2348)
• Added more rules for simplifying expressions (#2211)
• Sped up calculations of Electrode SOH variables for summary variables (#2210)

Breaking change

• Removed pybamm.SymbolReplacer as it is no longer needed to set up simulations with experiments, which is the only place where it was being used (#2395)
• Removed get_infinite_nested_dict, BaseModel.check_default_variables_dictionaries, and Discretisation.create_jacobian methods, which were not used by any other functionality in the repository (#2384)
• Dropped support for Python 3.7 after the release of Numpy v1.22.0 (#2379)
• Removed parameter cli tools (add/edit/remove parameters). Parameter sets can now more easily be added via python scripts. (#2342)
• Parameter sets should now be provided as single python files containing all parameters and functions. Parameters provided as "data" (e.g. OCP vs SOC) can still be csv files, but must be either in the same folder as the parameter file or in a subfolder called "data/". See for example Ai2020 (#2342)

v22.9 - 2022-09-30

Features

• Added function pybamm.get_git_commit_info(), which returns information about the last git commit, useful for reproducibility (#2293)
• Added SEI model for composite electrodes (#2290)
• For experiments, the simulation now automatically checks and skips steps that cannot be performed (e.g. "Charge at 1C until 4.2V" from 100% SOC) (#2212)

Bug fixes

• Arrhenius function for nmc_OKane2022 positive electrode actually gets used now (#2309)
• Added SEI on cracks to loop over all interfacial reactions (#2262)
• Fixed X-averaged SEI on cracks concentration so it's an average over x only, not y and z (#2262)
• Corrected initial state for SEI on cracks (#2262)

Optimizations

• Default options for particle mechanics now dealt with differently in each electrode (#2262)
• Sped up calculations of Electrode SOH variables for summary variables (#2210)

Breaking changes

• When creating a pybamm.Interpolant the default interpolator is now "linear". Passing data directly to ParameterValues using the [data] tag will be still used to create a cubic spline interpolant, as before (#2258)
• Events must now be defined in such a way that they are positive at the initial conditions (events will be triggered when they become negative, instead of when they change sign in either direction) (#2212)

v22.8 - 2022-08-31

Features

• Added CurrentSigmoidOpenCircuitPotential model to model voltage hysteresis for charge/discharge (#2256)
• Added "Chen2020_composite" parameter set for a composite graphite/silicon electrode. (#2256)
• Added new cumulative variables Throughput capacity [A.h] and Throughput energy [W.h] to standard variables and summary variables, to assist with degradation studies. Throughput variables are only calculated if calculate discharge energy is set to true. Time [s] and Time [h] also added to summary variables. (#2249)
• Added lipf6_OKane2022 electrolyte to OKane2022 parameter set (#2249)
• Reformated submodel structure to allow composite electrodes. Composite positive electrode is now also possible. With current implementation, electrodes can have at most two phases. (#2248)

Bug fixes

• Added new parameter Ratio of lithium moles to SEI moles (short name z_sei) to fix a bug where this number was incorrectly hardcoded to 1. (#2222)
• Changed short name of parameter Inner SEI reaction proportion from alpha_SEI to inner_sei_proportion, to avoid confusion with transfer coefficients. (#2222)
• Deleted legacy parameters with short names beta_sei and beta_plating. (#2222)
• Corrected initial SEI thickness for OKane2022 parameter set. (#2218)

Optimizations

• Simplified scaling for the exchange-current density. The dimensionless parameter C_r is kept, but no longer used anywhere (#2238)
• Added limits for variables in some functions to avoid division by zero, sqrt(negative number), etc (#2213)

Breaking changes

• Parameters specific to a (primary/secondary) phase in a domain are doubly nested. e.g. param.c_n_max is now param.n.prim.c_max (#2248)

v22.7 - 2022-07-31

Features

• Moved general code about submodels to BaseModel instead of BaseBatteryModel, making it easier to build custom models from submodels. (#2169)
• Events can now be plotted as a regular variable (under the name "Event: event_name", e.g. "Event: Minimum voltage [V]") (#2158)
• Added example showing how to print whether a model is compatible with a parameter set (#2112)
• Added SEI growth on cracks (#2104)
• Added Arrhenius temperature dependence of SEI growth (#2104)
• The "Inner SEI reaction proportion" parameter actually gets used now (#2104)
• New OKane2022 parameter set replaces Chen2020_plating (#2104)
• SEI growth, lithium plating and porosity change can now be set to distributed in SPMe. There is an additional option called x-average side reactions which allows to set this (note that for SPM it is always x-averaged). (#2099)

Optimizations

• Improved eSOH calculations to be more robust (#2192,#2199)
• The (2x2x2=8) particle diffusion submodels have been consolidated into just three submodels (Fickian diffusion, polynomial profile, and x-averaged polynomial profile) with optional x-averaging and size distribution. Polynomial profile and x-averaged polynomial profile are still two separate submodels, since they deal with surface concentration differently.
• Added error for when solution vector gets too large, to help debug solver errors (#2138)

Bug fixes

• Fixed error reporting for simulation with experiment (#2213)
• Fixed a bug in Simulation that caused initial conditions to change when solving an experiment multiple times (#2204)
• Fixed labels and ylims in plot_voltage_components(#2183)
• Fixed 2D interpolant (#2180)
• Fixes a bug where the SPMe always builds even when build=False (#2169)
• Some events have been removed in the case where they are constant, i.e. can never be reached (#2158)
• Raise explicit NotImplementedError if trying to call bool() on a pybamm Symbol (e.g. in an if statement condition) (#2141)
• Fixed bug causing cut-off voltage to change after setting up a simulation with a model (#2138)
• A single solution cycle can now be used as a starting solution for a simulation (#2138)

Breaking changes

• Exchange-current density functions (and some other functions) now take an additional argument, the maximum particle concentration for that phase (#2134)
• Loss of lithium to SEI on cracks is now a degradation variable, so setting a particle mechanics submodel is now compulsory (NoMechanics will suffice) (#2104)

v22.6 - 2022-06-30

Features

• Added open-circuit potential as a separate submodel (#2094)
• Added partially reversible lithium plating model and new OKane2022 parameter set to go with it (#2043)
• Added __eq__ and __hash__ methods for Symbol objects, using .id (#1978)

Optimizations

• Stoichiometry inputs to OCP functions are now bounded between 1e-10 and 1-1e-10, with singularities at 0 and 1 so that OCP goes to +- infinity (#2095)

Breaking changes

• Changed some dictionary keys to Symbol instead of Symbol.id (internal change only, should not affect external facing functions) (#1978)

v22.5 - 2022-05-31

Features

• Added a casadi version of the IDKLU solver, which is used for model.convert_to_format = "casadi" (#2002)
• Added functionality to generate Julia expressions from a model. See PyBaMM.jl for how to use these (#1942))
• Added basic callbacks to the Simulation class, and a LoggingCallback (#1880))

Bug fixes

• Corrected legend order in "plot_voltage_components.py", so each entry refers to the correct overpotential. (#2061)

Breaking changes

• Changed domain-specific parameter names to a nested attribute. param.n.l_n is now param.n.l (#2063)

v22.4 - 2022-04-30

Features

• Added a casadi version of the IDKLU solver, which is used for model.convert_to_format = "casadi" (#2002)

Bug fixes

• Remove old deprecation errors, including those in parameter_values.py that caused the simulation if, for example, the reaction rate is re-introduced manually (#2022)

v22.3 - 2022-03-31

Features

• Added "Discharge energy [W.h]", which is the integral of the power in Watts, as an optional output. Set the option "calculate discharge energy" to "true" to get this output ("false" by default, since it can slow down some of the simple models) (#1969))
• Added an option "calculate heat source for isothermal models" to choose whether or not the heat generation terms are computed when running models with the option thermal="isothermal" (#1958)

Optimizations

• Simplified model.new_copy() (#1977)

Bug fixes

• Fix bug where sensitivity calculation failed if len of calculate_sensitivities was less than inputs (#1897)
• Fixed a bug in the eSOH variable calculation when OCV is given as data (#1975)
• Fixed a bug where isothermal models did not compute any heat source terms (#1958)

Breaking changes

• Removed model.new_empty_copy() (use model.new_copy() instead) (#1977)
• Dropped support for Windows 32-bit architecture (#1964)

v22.2 - 2022-02-28

Features

• Isothermal models now calculate heat source terms (but the temperature remains constant). The models now also account for current collector heating when dimensionality=0 (#1929)
• Added new models for power control and resistance control (#1917)
• Initial concentrations can now be provided as a function of r as well as x (#1866)

Bug fixes

• Fixed a bug where thermal submodels could not be used with half-cells (#1929)
• Parameters can now be imported from a directory having "pybamm" in its name (#1919)
• scikit.odes and SUNDIALS can now be installed using pybamm_install_odes (#1916)

Breaking changes

• The domain setter and auxiliary_domains getter have been deprecated, domains setter/getter should be used instead. The domain getter is still active. We now recommend creating symbols with domains={...} instead of domain=..., auxiliary_domains={...}, but the latter is not yet deprecated (#1866)

v22.1 - 2022-01-31

Features

• Half-cell models can now be run with "surface form" (#1913)
• Added option for different kinetics on anode and cathode (#1913)
• Allow pybamm.Solution.save_data() to return a string if filename is None, and added json to_format option (#1909)
• Added an option to force install compatible versions of jax and jaxlib if already installed using CLI (#1881)

Optimizations

• The Symbol nodes no longer subclasses anytree.NodeMixIn. This removes some checks that were not really needed (#1912)

Bug fixes

• Parameters can now be imported from any given path in Windows (#1900)
• Fixed initial conditions for the EC SEI model (#1895)
• Fixed issue in extraction of sensitivites (#1894)

v21.12 - 2021-12-29

Features

• Added new kinetics models for asymmetric Butler-Volmer, linear kinetics, and Marcus-Hush-Chidsey (#1858)
• Experiments can be set to terminate when a voltage is reached (across all steps) (#1832)
• Added cylindrical geometry and finite volume method (#1824)

Bug fixes

• PyBaMM is now importable in Linux systems where jax is already installed (#1874)
• Simulations with drive cycles now support initial_soc (#1842)
• Fixed bug in expression tree simplification (#1831)
• Solid tortuosity is now correctly calculated with Bruggeman coefficient of the respective electrode (#1773)

v21.11 - 2021-11-30

Features

• The name of a parameter set can be passed to ParameterValues as a string, e.g. ParameterValues("Chen2020") (#1822)
• Added submodels for interface utilisation (#1821)
• Reformatted SEI growth models into a single submodel with conditionals (#1808)
• Stress-induced diffusion is now a separate model option instead of being automatically included when using the particle mechanics submodels (#1797)
• Experiments with drive cycles can be solved (#1793)
• Added surface area to volume ratio as a factor to the SEI equations (#1790)
• Half-cell SPM and SPMe have been implemented (#1731)

Bug fixes

• Fixed sympy operators for Arctan and Exponential (#1786)
• Fixed finite volume discretization in spherical polar coordinates (#1782)
• Fixed bug when using Experiment with a pouch cell model (#1707)
• Fixed bug when using Experiment with a plating model (#1707)
• Fixed hack for potentials in the SPMe model (#1707)

Breaking changes

• The chemistry keyword argument in ParameterValues has been deprecated. Use ParameterValues(chem) instead of ParameterValues(chemistry=chem) (#1822)
• Raise error when trying to convert an Interpolant with the "pchip" interpolator to CasADI (#1791)
• Raise error if Concatenation is used directly with Variable objects (concatenation should be used instead) (#1789)
• Made jax, jaxlib and the PyBaMM JaxSolver optional (#1767, #1803)

v21.10 - 2021-10-31

Features

• Summary variables can now be user-determined (#1760)
• Added all_first_states to the Solution object for a simulation with experiment (#1759)
• Added a new method (create_gif) in QuickPlot, Simulation and BatchStudy to create a GIF of a simulation (#1754)
• Added more examples for the BatchStudy class (#1747)
• SEI models can now be included in the half-cell model (#1705)

Bug fixes

• Half-cell model and lead-acid models can now be simulated with Experiments (#1759)
• Removed in-place modification of the solution objects by QuickPlot (#1747)
• Fixed vector-vector multiplication bug that was causing errors in the SPM with constant voltage or power (#1735)

v21.9 - 2021-09-30

Features

• Added thermal parameters (thermal conductivity, specific heat, etc.) to the Ecker2015 parameter set from Zhao et al. (2018) and Hales et al. (2019) (#1683)
• Added plot_summary_variables to plot and compare summary variables (#1678)
• The DFN model can now be used directly (instead of BasicDFNHalfCell) to simulate a half-cell (#1600)

Breaking changes

• Dropped support for Python 3.6 (#1696)
• The substring 'negative electrode' has been removed from variables related to SEI and lithium plating (e.g. 'Total negative electrode SEI thickness [m]' replaced by 'Total SEI thickness [m]') (#1654)

v21.08 - 2021-08-26

This release introduces:

• the switch to calendar versioning: from now on we will use year.month version number
• sensitivity analysis of solutions with respect to input parameters
• several new models, including many-particle and state-of-health models
• improvement on how CasADI solver's handle events, including a new "fast with events" mode
• several other new features, optimizations, and bug fixes, summarized below

Features

• Added submodels and functionality for particle-size distributions in the DFN model, including an example notebook (#1602)
• Added UDDS and WLTC drive cycles (#1601)
• Added LG M50 (NMC811 and graphite + SiOx) parameter set from O'Regan 2022 (#1594)
• pybamm.base_solver.solve function can take a list of input parameters to calculate the sensitivities of the solution with respect to. Alternatively, it can be set to True to calculate the sensitivities for all input parameters (#1552)
• Added capability for quaternary domains (in addition to primary, secondary and tertiary), increasing the maximum number of domains that a Symbol can have to 4. (#1580)
• Tabs can now be placed at the bottom of the cell in 1+1D thermal models (#1581)
• Added temperature dependence on electrode electronic conductivity (#1570)
• pybamm.base_solver.solve function can take a list of input parameters to calculate the sensitivities of the solution with respect to. Alternatively, it can be set to True to calculate the sensitivities for all input parameters (#1552)
• Added a new lithium-ion model MPM or Many-Particle Model, with a distribution of particle sizes in each electrode. (#1529)
• Added 2 new submodels for lithium transport in a size distribution of electrode particles: Fickian diffusion (FickianSingleSizeDistribution) and uniform concentration profile (FastSingleSizeDistribution). (#1529)
• Added a "particle size" domain to the default lithium-ion geometry, including plotting capabilities (QuickPlot) and processing of variables (ProcessedVariable). (#1529)
• Added fitted expressions for OCPs for the Chen2020 parameter set (#1526)
• Added initial_soc argument to Simualtion.solve for specifying the initial SOC when solving a model (#1512)
• Added print_name to some symbols (#1495, #1497)
• Added Base Parameters class and SymPy in dependencies (#1495)
• Added a new "reaction-driven" model for LAM from Reniers et al (2019) (#1490)
• Some features ("loss of active material" and "particle mechanics") can now be specified separately for the negative electrode and positive electrode by passing a 2-tuple (#1490)
• plot and plot2D now take and return a matplotlib Axis to allow for easier customization (#1472)
• ParameterValues.evaluate can now return arrays to allow function parameters to be easily evaluated (#1472)
• Added option to save only specific cycle numbers when simulating an Experiment (#1459)
• Added capacity-based termination conditions when simulating an Experiment (#1459)
• Added "summary variables" to track degradation over several cycles (#1459)
• Added ElectrodeSOH model for calculating capacities and stoichiometric limits (#1459)
• Added Batch Study class (#1455)
• Added ConcatenationVariable, which is automatically created when variables are concatenated (#1453)
• Added "fast with events" mode for the CasADi solver, which solves a model and finds events more efficiently than "safe" mode. As of PR #1450 this feature is still being tested and "safe" mode remains the default (#1450)

Optimizations

• Models that mostly use x-averaged quantities (SPM and SPMe) now use x-averaged degradation models (#1490)
• Improved how the CasADi solver's "safe" mode finds events (#1450)
• Perform more automatic simplifications of the expression tree (#1449)
• Reduce time taken to hash a sparse Matrix object (#1449)

Bug fixes

• Fixed bug with load_function (#1675)
• Updated documentation to include some previously missing functions, such as erf and tanh (#1628)
• Fixed reading citation file without closing (#1620)
• Porosity variation for SEI and plating models is calculated from the film thickness rather than from a separate ODE (#1617)
• Fixed a bug where the order of the indexing for the entries of variables discretised using FEM was incorrect (#1556)
• Fix broken module import for spyder when running a script twice (#1555)
• Fixed ElectrodeSOH model for multi-dimensional simulations (#1548)
• Removed the overly-restrictive check "each variable in the algebraic eqn keys must appear in the eqn" (#1510)
• Made parameters importable through pybamm (#1475)

Breaking changes

• Refactored the particle submodel module, with the models having no size distribution now found in particle.no_distribution, and those with a size distribution in particle.size_distribution. Renamed submodels to indicate the transport model (Fickian diffusion, polynomial profile) and if they are "x-averaged". E.g., FickianManyParticles and FickianSingleParticle are now no_distribution.FickianDiffusion and no_distribution.XAveragedFickianDiffusion (#1602)
• Changed sensitivity API. Removed ProcessedSymbolicVariable, all sensitivity now handled within the solvers and ProcessedVariable (#1552,#2276)
• The Yang2017 parameter set has been removed as the complete parameter set is not publicly available in the literature (#1577)
• Changed how options are specified for the "loss of active material" and "particle cracking" submodels. "loss of active material" can now be one of "none", "stress-driven", or "reaction-driven", or a 2-tuple for different options in negative and positive electrode. Similarly "particle cracking" (now called "particle mechanics") can now be "none", "swelling only", "swelling and cracking", or a 2-tuple (#1490)
• Changed the variable in the full diffusion model from "Electrolyte concentration" to "Porosity times concentration" (#1476)
• Renamed lithium-ion folder to lithium_ion and lead-acid folder to lead_acid in parameters (#1464)

v0.4.0 - 2021-03-28

This release introduces:

• several new models, including reversible and irreversible plating submodels, submodels for loss of active material, Yang et al.'s (2017) coupled SEI/plating/pore clogging model, and the Newman-Tobias model
• internal optimizations for solving models, particularly for simulating experiments, with more accurate event detection and more efficient numerical methods and post-processing
• parallel solutions of a model with different inputs
• a cleaner installation process for Mac when installing from PyPI, no longer requiring a Homebrew installation of Sundials
• improved plotting functionality, including adding a new 'voltage component' plot
• several other new features, optimizations, and bug fixes, summarized below

Features

• Added NewmanTobias li-ion battery model (#1423)
• Added plot_voltage_components to easily plot the component overpotentials that make up the voltage (#1419)
• Made QuickPlot more customizable and added an example (#1419)
• Solution objects can now be created by stepping different models (#1408)
• Added Yang et al 2017 model that couples irreversible lithium plating, SEI growth and change in porosity which produces a transition from linear to nonlinear degradation pattern of lithium-ion battery over extended cycles(#1398)
• Added support for Python 3.9 and dropped support for Python 3.6. Python 3.6 may still work but is now untested (#1370)
• Added the electrolyte overpotential and Ohmic losses for full conductivity, including surface form (#1350)
• Added functionality to Citations to print formatted citations (#1340)
• Updated the way events are handled in CasadiSolver for more accurate event location (#1328)
• Added error message if initial conditions are outside the bounds of a variable (#1326)
• Added temperature dependence to density, heat capacity and thermal conductivity (#1323)
• Added temperature dependence to the transference number (t_plus) (#1317)
• Added new functionality for Interpolant (#1312)
• Added option to express experiments (and extract solutions) in terms of cycles of operating condition (#1309)
• The event time and state are now returned as part of Solution.t and Solution.y so that the event is accurately captured in the returned solution (#1300)
• Added reversible and irreversible lithium plating models (#1287)
• Reformatted the BasicDFNHalfCell to be consistent with the other models (#1282)
• Added option to make the total interfacial current density a state (#1280)
• Added functionality to initialize a model using the solution from another model (#1278)
• Added submodels for active material (#1262)
• Updated solvers' method solve() so it can take a list of inputs dictionaries as the inputs keyword argument. In this case the model is solved for each input set in the list, and a list of solutions mapping the set of inputs to the solutions is returned. Note that solve() can still take a single dictionary as the inputs keyword argument. In this case the behaviour is unchanged compared to previous versions.(#1261)
• Added composite surface form electrolyte models: CompositeDifferential and CompositeAlgebraic (#1207)

Optimizations

• Improved the way an Experiment is simulated to reduce solve time (at the cost of slightly higher set-up time) (#1408)
• Add script and workflow to automatically update parameter_sets.py docstrings (#1371)
• Add URLs checker in workflows (#1347)
• The Solution class now only creates the concatenated y when the user asks for it. This is an optimization step as the concatenation can be slow, especially with larger experiments (#1331)
• If solver method solve() is passed a list of inputs as the inputs keyword argument, the resolution of the model for each input set is spread across several Python processes, usually running in parallel on different processors. The default number of processes is the number of processors available. solve() takes a new keyword argument nproc which can be used to set this number a manually.
• Variables are now post-processed using CasADi (#1316)
• Operations such as 1*x and 0+x now directly return x (#1252)

Bug fixes

• Fixed a bug on the boundary conditions of FickianSingleParticle and FickianManyParticles to ensure mass is conserved (#1421)
• Fixed a bug where the PolynomialSingleParticle submodel gave incorrect results with "dimensionality" equal to 2 (#1411)
• Fixed a bug where volume averaging in 0D gave the wrong result (#1411)
• Fixed a sign error in the positive electrode ohmic losses (#1407)
• Fixed the formulation of the EC reaction SEI model (#1397)
• Simulations now stop when an experiment becomes infeasible (#1395)
• Added a check for domains in Concatenation (#1368)
• Differentiation now works even when the differentiation variable is a constant (#1294)
• Fixed a bug where the event time and state were no longer returned as part of the solution (#1344)
• Fixed a bug in CasadiSolver safe mode which crashed when there were extrapolation events but no termination events (#1321)
• When an Interpolant is extrapolated an error is raised for CasadiSolver (and a warning is raised for the other solvers) (#1315)
• Fixed Simulation and model.new_copy to fix a bug where changes to the model were overwritten (#1278)

Breaking changes

• Removed Simplification class and .simplify() function (#1369)
• All example notebooks in PyBaMM's GitHub repository must now include the command pybamm.print_citations(), otherwise the tests will fail. This is to encourage people to use this command to cite the relevant papers (#1340)
• Notation has been homogenised to use positive and negative electrode (instead of cathode and anode). This applies to the parameter folders (now called 'positive_electrodes' and 'negative_electrodes') and the options of active_material and particle_cracking submodels (now called 'positive' and 'negative') (#1337)
• Interpolant now takes x and y instead of a single data entry (#1312)
• Boolean model options ('sei porosity change', 'convection') must now be given in string format ('true' or 'false' instead of True or False) (#1280)
• Operations such as 1*x and 0+x now directly return x. This can be bypassed by explicitly creating the binary operators, e.g. pybamm.Multiplication(1, x) (#1252)
• 'Cell capacity [A.h]' has been renamed to 'Nominal cell capacity [A.h]'. 'Cell capacity [A.h]' will be deprecated in the next release. (#1352)

v0.3.0 - 2020-12-01

This release introduces a new aging model for particle mechanics, a new reduced-order model (TSPMe), and a parameter set for A123 LFP cells. Additionally, there have been several backend optimizations to speed up model creation and solving, and other minor features and bug fixes.

Features

• Added a submodel for particle mechanics (#1232)
• Added a notebook on how to speed up the solver and handle instabilities (#1223)
• Improve string printing of BinaryOperator, Function, and Concatenation objects (#1223)
• Added Solution.integration_time, which is the time taken just by the integration subroutine, without extra setups (#1223)
• Added parameter set for an A123 LFP cell (#1209)
• Added variables related to equivalent circuit models (#1204)
• Added the Integrated electrolyte conductivity submodel (#1188)
• Added an example script to check conservation of lithium (#1186)
• Added erf and erfc functions (#1184)

Optimizations

• Add (optional) smooth approximations for the Minimum, Maximum, Heaviside, and AbsoluteValue operators (#1223)
• Avoid unnecessary repeated computations in the solvers (#1222)
• Rewrite Symbol.is_constant to be more efficient (#1222)
• Cache shape and size calculations (#1222)
• Only instantiate the geometric, electrical and thermal parameter classes once (#1222)

Bug fixes

• Quickplot now works when timescale or lengthscale is a function of an input parameter (#1234)
• Fix bug that was slowing down creation of the EC reaction SEI submodel (#1227)
• Add missing separator thermal parameters for the Ecker parameter set (#1226)
• Make sure simulation solves when evaluated timescale is a function of an input parameter (#1218)
• Raise error if saving to MATLAB with variable names that MATLAB can't read, and give option of providing alternative variable names (#1206)
• Raise error if the boundary condition at the origin in a spherical domain is other than no-flux (#1175)
• Fix boundary conditions at r = 0 for Creating Models notebooks (#1173)

Breaking changes

• The parameters "Positive/Negative particle distribution in x" and "Positive/Negative surface area to volume ratio distribution in x" have been deprecated. Instead, users can provide "Positive/Negative particle radius [m]" and "Positive/Negative surface area to volume ratio [m-1]" directly as functions of through-cell position (x [m]) (#1237)

v0.2.4 - 2020-09-07

This release adds new operators for more complex models, some basic sensitivity analysis, and a spectral volumes spatial method, as well as some small bug fixes.

Features

• Added variables which track the total amount of lithium in the system (#1136)
• Added Upwind and Downwind operators for convection (#1134)
• Added Getting Started notebook on solver options and changing the mesh. Also added a notebook detailing the different thermal options, and a notebook explaining the steps that occur behind the scenes in the Simulation class (#1131)
• Added particle submodel that use a polynomial approximation to the concentration within the electrode particles (#1130)
• Added Modulo, Floor and Ceiling operators (#1121)
• Added DFN model for a half cell (#1121)
• Automatically compute surface area to volume ratio based on particle shape for li-ion models (#1120)
• Added "R-averaged particle concentration" variables (#1118)
• Added support for sensitivity calculations to the casadi solver (#1109)
• Added support for index 1 semi-explicit dae equations and sensitivity calculations to JAX BDF solver (#1107)
• Allowed keyword arguments to be passed to Simulation.plot() (#1099)
• Added the Spectral Volumes spatial method and the submesh that it works with (#900)

Bug fixes

• Fixed bug where some parameters were not being set by the EcReactionLimited SEI model (#1136)
• Fixed bug on electrolyte potential for BasicDFNHalfCell (#1133)
• Fixed r_average to work with SecondaryBroadcast (#1118)
• Fixed finite volume discretisation of spherical integrals (#1118)
• t_eval now gets changed to a linspace if a list of length 2 is passed (#1113)
• Fixed bug when setting a function with an InputParameter (#1111)

Breaking changes

• The "fast diffusion" particle option has been renamed "uniform profile" (#1130)
• The modules containing standard parameters are now classes so they can take options (e.g. standard_parameters_lithium_ion is now LithiumIonParameters) (#1120)
• Renamed quick_plot_vars to output_variables in Simulation to be consistent with QuickPlot. Passing quick_plot_vars to Simulation.plot() has been deprecated and output_variables should be passed instead (#1099)

v0.2.3 - 2020-07-01

This release enables the use of Google Colab for running example notebooks, and adds some small new features and bug fixes.

Features

• Added JAX evaluator, and ODE solver (#1038)
• Reformatted Getting Started notebooks (#1083)
• Reformatted Landesfeind electrolytes (#1064)
• Adapted examples to be run in Google Colab (#1061)
• Added some new solvers for algebraic models (#1059)
• Added length_scales attribute to models (#1058)
• Added averaging in secondary dimensions (#1057)
• Added SEI reaction based on Yang et. al. 2017 and reduction in porosity (#1009)

Optimizations

• Reformatted CasADi "safe" mode to deal with events better (#1089)

Bug fixes

• Fixed a bug in InterstitialDiffusionLimited (#1097)
• Fixed Simulation to keep different copies of the model so that parameters can be changed between simulations (#1090)
• Fixed model.new_copy() to keep custom submodels (#1090)
• 2D processed variables can now be evaluated at the domain boundaries (#1088)
• Update the default variable points to better capture behaviour in the solid particles in li-ion models (#1081)
• Fix QuickPlot to display variables discretised by FEM (in y-z) properly (#1078)
• Add length scales to EffectiveResistance models (#1071)
• Allowed for pybamm functions exp, sin, cos, sqrt to be used in expression trees that are converted to casadi format (#1067)
• Fix a bug where variables that depend on y and z were transposed in QuickPlot (#1055)

Breaking changes

• Simulation.specs and Simulation.set_defaults have been deprecated. Users should create a new Simulation object for each different case instead (#1090)
• The solution times t_eval must now be provided to Simulation.solve() when not using an experiment or prescribing the current using drive cycle data (#1086)

v0.2.2 - 2020-06-01

New SEI models, simplification of submodel structure, as well as optimisations and general bug fixes.

Features

• Reformatted Geometry and Mesh classes (#1032)
• Added arbitrary geometry to the lumped thermal model (#718)
• Allowed ProcessedVariable to handle cases where len(solution.t)=1 (#1020)
• Added BackwardIndefiniteIntegral symbol (#1014)
• Added plot and plot2D to enable easy plotting of pybamm.Array objects (#1008)
• Updated effective current collector models and added example notebook (#1007)
• Added SEI film resistance as an option (#994)
• Added parameters attribute to pybamm.BaseModel and pybamm.Geometry that lists all of the required parameters (#993)
• Added tab, edge, and surface cooling (#965)
• Added functionality to solver to automatically discretise a 0D model (#947)
• Added sensitivity to CasadiAlgebraicSolver (#940)
• Added ProcessedSymbolicVariable class, which can handle symbolic variables (i.e. variables for which the inputs are symbolic) (#940)
• Made QuickPlot compatible with Google Colab (#935)
• Added BasicFull model for lead-acid (#932)
• Added 'arctan' function (#973)

Optimizations

• Implementing the use of GitHub Actions for CI (#855)
• Changed default solver for DAE models to CasadiSolver (#978)
• Added some extra simplifications to the expression tree (#971)
• Changed the behaviour of "safe" mode in CasadiSolver (#956)
• Sped up model building (#927)
• Changed default solver for lead-acid to CasadiSolver (#927)

Bug fixes

• Fix a bug where slider plots do not update properly in notebooks (#1041)
• Fix storing and plotting external variables in the solution (#1026)
• Fix running a simulation with a model that is already discretized (#1025)
• Fix CI not triggering for PR. (#1013)
• Fix schedule testing running too often. (#1010)
• Fix doctests failing due to mismatch in unsorted output.(#990)
• Added extra checks when creating a model, for clearer errors (#971)
• Fixed Interpolant ids to allow processing (#962)
• Fixed a bug in the initial conditions of the potential pair model (#954)
• Changed simulation attributes to assign copies rather than the objects themselves (#952)
• Added default values to base model so that it works with the Simulation class (#952)
• Fixed solver to recompute initial conditions when inputs are changed (#951)
• Reformatted thermal submodels (#938)
• Reformatted electrolyte submodels (#927)
• Reformatted convection submodels (#635)

Breaking changes

• Geometry should no longer be given keys 'primary' or 'secondary' (#1032)
• Calls to ProcessedVariable objects are now made using dimensional time and space (#1028)
• For variables discretised using finite elements the result returned by calling ProcessedVariable is now transposed (#1020)
• Renamed "surface area density" to "surface area to volume ratio" (#975)
• Replaced "reaction rate" with "exchange-current density" (#975)
• Changed the implementation of reactions in submodels (#948)
• Removed some inputs like T_inf, R_g and activation energies to some of the standard function parameters. This is because each of those inputs is specific to a particular function (e.g. the reference temperature at which the function was measured). To change a property such as the activation energy, users should create a new function, specifying the relevant property as a Parameter or InputParameter (#942)
• The thermal option 'xyz-lumped' has been removed. The option 'thermal current collector' has also been removed (#938)
• The 'C-rate' parameter has been deprecated. Use 'Current function [A]' instead. The cell capacity can be accessed as 'Cell capacity [A.h]', and used to calculate current from C-rate (#952)

v0.2.1 - 2020-03-31

New expression tree node types, models, parameter sets and solvers, as well as general bug fixes and new examples.

Features

• Store variable slices in model for inspection (#925)
• Added LiNiCoO2 parameter set from Ecker et. al. (#922)
• Made t_plus (optionally) a function of electrolyte concentration, and added (1 + dlnf/dlnc) to models (#921)
• Added DummySolver for empty models (#915)
• Added functionality to broadcast to edges (#891)
• Reformatted and cleaned up QuickPlot (#886)
• Added thermal effects to lead-acid models (#885)
• Added a helper function for info on function parameters (#881)
• Added additional notebooks showing how to create and compare models (#877)
• Added Minimum, Maximum and Sign operators (#876)
• Added a search feature to FuzzyDict (#875)
• Add ambient temperature as a function of time (#872)
• Added CasadiAlgebraicSolver for solving algebraic systems with CasADi (#868)
• Added electrolyte functions from Landesfeind (#860)
• Add new symbols VariableDot, representing the derivative of a variable wrt time, and StateVectorDot, representing the derivative of a state vector wrt time (#858)

Bug fixes

• Filter out discontinuities that occur after solve times (#941)
• Fixed tight layout for QuickPlot in jupyter notebooks (#930)
• Fixed bug raised if function returns a scalar (#919)
• Fixed event handling in ScipySolver (#905)
• Made input handling clearer in solvers (#905)
• Updated Getting started notebook 2 (#903)
• Reformatted external circuit submodels (#879)
• Some bug fixes to generalize specifying models that aren't battery models, see #846
• Reformatted interface submodels to be more readable (#866)
• Removed double-counted "number of electrodes connected in parallel" from simulation (#864)

Breaking changes

• Changed keyword argument u for inputs (when evaluating an object) to inputs (#905)
• Removed "set external temperature" and "set external potential" options. Use "external submodels" option instead (#862)

v0.2.0 - 2020-02-26

This release introduces many new features and optimizations. All models can now be solved using the pip installation - in particular, the DFN can be solved in around 0.1s. Other highlights include an improved user interface, simulations of experimental protocols (GITT, CCCV, etc), new parameter sets for NCA and LGM50, drive cycles, "input parameters" and "external variables" for quickly solving models with different parameter values and coupling with external software, and general bug fixes and optimizations.

Features

• Added LG M50 parameter set from Chen 2020 (#854)
• Changed rootfinding algorithm to CasADi, scipy.optimize.root still accessible as an option (#844)
• Added capacitance effects to lithium-ion models (#842)
• Added NCA parameter set (#824)
• Added functionality to Solution that automatically gets t_eval from the data when simulating drive cycles and performs checks to ensure the output has the required resolution to accurately capture the input current (#819)
• Added Citations object to print references when specific functionality is used (#818)
• Updated Solution to allow exporting to matlab and csv formats (#811)
• Allow porosity to vary in space (#809)
• Added functionality to solve DAE models with non-smooth current inputs (#808)
• Added functionality to simulate experiments and testing protocols (#807)
• Added fuzzy string matching for parameters and variables (#796)
• Changed ParameterValues to raise an error when a parameter that wasn't previously defined is updated (#796)
• Added some basic models (BasicSPM and BasicDFN) in order to clearly demonstrate the PyBaMM model structure for battery models (#795)
• Allow initial conditions in the particle to depend on x (#786)
• Added the harmonic mean to the Finite Volume method, which is now used when computing fluxes (#783)
• Refactored Solution to make it a dictionary that contains all of the solution variables. This automatically creates ProcessedVariable objects when required, so that the solution can be obtained much more easily. (#781)
• Added notebook to explain broadcasts (#776)
• Added a step to discretisation that automatically compute the inverse of the mass matrix of the differential part of the problem so that the underlying DAEs can be provided in semi-explicit form, as required by the CasADi solver (#769)
• Added the gradient operation for the Finite Element Method (#767)
• Added InputParameter node for quickly changing parameter values (#752)
• Added submodels for operating modes other than current-controlled (#751)
• Changed finite volume discretisation to use exact values provided by Neumann boundary conditions when computing the gradient instead of adding ghost nodes(#748)
• Added optional R(x) distribution in particle models (#745)
• Generalized importing of external variables (#728)
• Separated active and inactive material volume fractions (#726)
• Added submodels for tortuosity (#726)
• Simplified the interface for setting current functions (#723)
• Added Heaviside operator (#723)
• New extrapolation methods (#707)
• Added some "Getting Started" documentation (#703)
• Allow abs tolerance to be set by variable for IDA KLU solver (#700)
• Added Simulation class (#693) with load/save functionality (#732)
• Added interface to CasADi solver (#687, #691, #714). This makes the SUNDIALS DAE solvers (Scikits and KLU) truly optional (though IDA KLU is recommended for solving the DFN).
• Added option to use CasADi's Algorithmic Differentiation framework to calculate Jacobians (#687)
• Added method to evaluate parameters more easily (#669)
• Added Jacobian class to reuse known Jacobians of expressions (#665)
• Added Interpolant class to interpolate experimental data (e.g. OCP curves) (#661)
• Added interface (via pybind11) to sundials with the IDA KLU sparse linear solver (#657)
• Allowed parameters to be set by material or by specifying a particular paper (#647)
• Set relative and absolute tolerances independently in solvers (#645)
• Added basic method to allow (a part of) the State Vector to be updated with results obtained from another solution or package (#624)
• Added some non-uniform meshes in 1D and 2D (#617)

Optimizations

• Now simplifying objects that are constant as soon as they are created (#801)
• Simplified solver interface (#800)
• Added caching for shape evaluation, used during discretisation (#780)
• Added an option to skip model checks during discretisation, which could be slow for large models (#739)
• Use CasADi's automatic differentation algorithms by default when solving a model (#714)
• Avoid re-checking size when making a copy of an Index object (#656)
• Avoid recalculating _evaluation_array when making a copy of a StateVector object (#653)

Bug fixes

• Fixed a bug where current loaded from data was incorrectly scaled with the cell capacity (#852)
• Moved evaluation of initial conditions to solver (#839)
• Fixed a bug where the first line of the data wasn't loaded when parameters are loaded from data (#819)
• Made graphviz an optional dependency (#810)
• Fixed examples to run with basic pip installation (#800)
• Added events for CasADi solver when stepping (#800)
• Improved implementation of broadcasts (#776)
• Fixed a bug which meant that the Ohmic heating in the current collectors was incorrect if using the Finite Element Method (#767)
• Improved automatic broadcasting (#747)
• Fixed bug with wrong temperature in initial conditions (#737)
• Improved flexibility of parameter values so that parameters (such as diffusivity or current) can be set as functions or scalars (#723)
• Fixed a bug where boundary conditions were sometimes handled incorrectly in 1+1D models (#713)
• Corrected a sign error in Dirichlet boundary conditions in the Finite Element Method (#706)
• Passed the correct dimensional temperature to open circuit potential (#702)
• Added missing temperature dependence in electrolyte and interface submodels (#698)
• Fixed differentiation of functions that have more than one argument (#687)
• Added warning if ProcessedVariable is called outside its interpolation range (#681)
• Updated installation instructions for Mac OS (#680)
• Improved the way ProcessedVariable objects are created in higher dimensions (#581)

Breaking changes

• Time for solver should now be given in seconds (#832)
• Model events are now represented as a list of pybamm.Event (#759
• Removed ParameterValues.update_model, whose functionality is now replaced by InputParameter (#801)
• Removed Outer and Kron nodes as no longer used (#777)
• Moved results to separate repositories (#761)
• The parameters "Bruggeman coefficient" must now be specified separately as "Bruggeman coefficient (electrolyte)" and "Bruggeman coefficient (electrode)"
• The current classes (GetConstantCurrent, GetUserCurrent and GetUserData) have now been removed. Please refer to the change-input-current notebook for information on how to specify an input current
• Parameter functions must now use pybamm functions instead of numpy functions (e.g. pybamm.exp instead of numpy.exp), as these are then used to construct the expression tree directly. Generally, pybamm syntax follows numpy syntax; please get in touch if a function you need is missing.
• The current must now be updated by changing "Current function [A]" or "C-rate" instead of "Typical current [A]"

v0.1.0 - 2019-10-08

This is the first official version of PyBaMM. Please note that PyBaMM in still under active development, and so the API may change in the future.

Features

Models

Lithium-ion

• Single Particle Model (SPM)
• Single Particle Model with electrolyte (SPMe)
• Doyle-Fuller-Newman (DFN) model

with the following optional physics:

• Thermal effects
• Fast diffusion in particles
• 2+1D (pouch cell)

Lead-acid

• Leading-Order Quasi-Static model
• First-Order Quasi-Static model
• Composite model
• Full model

with the following optional physics:

• Hydrolysis side reaction
• Capacitance effects
• 2+1D

Spatial discretisations

• Finite Volume (1D only)
• Finite Element (scikit, 2D only)

Solvers

• Scipy
• Scikits ODE
• Scikits DAE
• IDA KLU sparse linear solver (Sundials)
• Algebraic (root-finding)