# Change Log¶

## NetKet 3.4 (⚙️ In development)¶

### New features¶

• Support for Python 3.10 #952.

### Breaking Changes¶

• The gradient for models with real-parameter is now multiplied by 2. If your model had real parameters you might need to change the learning rate and halve it. #1069

### Bug Fixes¶

• Allow LazyOperator@densevector for operators such as lazy Adjoint, Transpose and Squared. #1068

• The gradient obtained with VarState.expect_and_grad for models with real-parameters was off by a factor of $$1/2$$ from the correct value. This has now been corrected. As a consequence, the correct gradient for real-parameter models is equal to the old times 2. If your model had real parameters you might need to change the learning rate and halve it. #1069

## NetKet 3.3.1 (🐛 Bug Fixes)¶

• Initialisation of all implementations of DenseSymm, DenseEquivariant, GCNN now defaults to truncated normals with Lecun variance scaling. For layers without masking, there should be no noticeable change in behaviour. For masked layers, the same variance scaling now works correctly. #1045

• Fix bug that prevented gradients of non-hermitian operators to be computed. The feature is still marked as experimental but will now run (we do not guarantee that results are correct). #1053

• Common lattice constructors such as Honeycomb now accepts the same keyword arguments as Lattice. #1046

• Multiplying a QGTOnTheFly representing the real part of the QGT (showing up when the ansatz has real parameters) with a complex vector now throws an error. Previously the result would be wrong, as the imaginary part was casted away. #885

## NetKet 3.3 (🎁 20 December 2021)¶

### New features¶

• The interface to define expectation and gradient function of arbitrary custom operators is now stable. If you want to define it for a standard operator that can be written as an average of local expectation terms, you can now define a dispatch rule for netket.vqs.get_local_kernel_arguments and netket.vqs.get_local_kernel. The old mechanism is still supported, but we encourage to use the new mechanism as it is more terse. #954

• nk.optimizer.Adam now supports complex parameters, and you can use nk.optimizer.split_complex to make optimizers process complex parameters as if they are pairs of real parameters. #1009

• Chunking of MCState.expect and MCState.expect_and_grad computations is now supported, which allows to bound the memory cost in exchange of a minor increase in computation time. #1006 (and discussions in #918 and #830)

• A new variational state that performs exact summation over the whole Hilbert space has been added. It can be constructed with nk.vqs.ExactState and supports the same Jax neural networks as nk.vqs.MCState. #953

• DenseSymm allows multiple input features. #1030

• [Experimental] A new time-evolution driver nk.experimental.TDVP using the time-dependent variational principle (TDVP) has been added. It works with time-independent and time-dependent Hamiltonians and Liouvillians. #1012

• [Experimental] A set of JAX-compatible Runge-Kutta ODE integrators has been added for use together with the new TDVP driver. #1012

### Breaking Changes¶

• The method sample_next in Sampler and exact samplers (ExactSampler and ARDirectSampler) is removed, and it is only defined in MetropolisSampler. The module function nk.sampler.sample_next also only works with MetropolisSampler. For exact samplers, please use the method sample instead. #1016

• The default value of n_chains_per_rank in Sampler and exact samplers is changed to 1, and specifying n_chains or n_chains_per_rank when constructing them is deprecated. Please change chain_length when calling sample. For MetropolisSampler, the default value is changed from n_chains = 16 (across all ranks) to n_chains_per_rank = 16. #1017

• GCNN_Parity allowed biasing both the parity-preserving and the parity-flip equivariant layers. These enter into the network output the same way, so having both is redundant and makes QGTs unstable. The biases of the parity-flip layers are now removed. The previous behaviour can be restored using the deprecated extra_bias switch; we only recommend this for loading previously saved parameters. Such parameters can be transformed to work with the new default using nk.models.update_GCNN_parity. #1030

• Kernels of DenseSymm are now three-dimensional, not two-dimensional. Parameters saved from earlier implementations can be transformed to the new convention using nk.nn.update_dense_symm. #1030

### Deprecations¶

• The method Sampler.samples is added to return a generator of samples. The module functions nk.sampler.sampler_state, reset, sample, samples, and sample_next are deprecated in favor of the corresponding class methods. #1025

• Kwarg in_features of DenseEquivariant is deprecated; the number of input features are inferred from the input. #1030

• Kwarg out_features of DenseEquivariant is deprecated in favour of features. #1030

### Internal Changes¶

• The definitions of MCState and MCMixedState have been moved to an internal module, nk.vqs.mc that is hidden by default. #954

• Custom deepcopy for LocalOperator to avoid building LocalOperator from scratch each time it is copied #964

### Bug Fixes¶

• The constructor of TensorHilbert (which is used by the product operator * for inhomogeneous spaces) no longer fails when one of the component spaces is non-indexable. #1004

• The nk.hilbert.random.flip_state method used by MetropolisLocal now throws an error when called on a nk.hilbert.ContinuousHilbert hilbert space instead of entering an endless loop. #1014

• Fixed bug in conversion to qutip for MCMixedState, where the resulting shape (hilbert space size) was wrong. #1020

• Setting MCState.sampler now recomputes MCState.chain_length according to MCState.n_samples and the new sampler.n_chains. #1028

• GCNN_Parity allowed biasing both the parity-preserving and the parity-flip equivariant layers. These enter into the network output the same way, so having both is redundant and makes QGTs unstable. The biases of the parity-flip layers are now removed. #1030

## NetKet 3.2 (26 November 2021)¶

### New features¶

• GraphOperator (and Heisenberg) now support passing a custom mapping of graph nodes to Hilbert space sites via the new acting_on_subspace argument. This makes it possible to create GraphOperators that act on a subset of sites, which is useful in composite Hilbert spaces. #924

• PauliString now supports any Hilbert space with local size 2. The Hilbert space is now the optional first argument of the constructor. #960

• PauliString now can be multiplied and summed together, performing some simple algebraic simplifications on the strings they contain. They also lazily initialize their internal data structures, making them faster to construct but slightly slower the first time that their matrix elements are accessed. #955

• PauliStrings can now be constructed starting from an OpenFermion operator. #956

• In addition to nearest-neighbor edges, Lattice can now generate edges between next-nearest and, more generally, k-nearest neighbors via the constructor argument max_neighbor_order. The edges can be distinguished by their color property (which is used, e.g., by GraphOperator to apply different bond operators). #970

• Two continuous-space operators (KineticEnergy and PotentialEnergy) have been implemented. #971

• Heisenberg Hamiltonians support different coupling strengths on Graph edges with different colors. #972.

• The little_group and space_group_irreps methods of SpaceGroupBuilder take the wave vector as either varargs or iterables. #975

• A new netket.experimental submodule has been created and all experimental features have been moved there. Note that in contrast to the other netket submodules, netket.experimental is not imported by default. #976

### Breaking Changes¶

• Moved nk.vqs.variables_from_*** to nk.experimental.vqs module. Also moved the experimental samplers to nk.sampler.MetropolisPt and nk.sampler.MetropolisPmap to nk.experimental.sampler. #976

• operator.size, has been deprecated. If you were using this function, please transition to operator.hilbert.size. #985

### Bug Fixes¶

• A bug where LocalOperator.get_conn_flattened would read out-of-bounds memory has been fixed. It is unlikely that the bug was causing problems, but it triggered warnings when running Numba with boundscheck activated. #966

• The dependency python-igraph has been updated to igraph following the rename of the upstream project in order to work on conda. #986

• n_samples_per_rank was returning wrong values and has now been fixed. #987

• The DenseSymm layer now also accepts objects of type HashableArray as symmetries argument. #989

• A bug where VMC.info() was erroring has been fixed. #984

## NetKet 3.1 (20 October 2021)¶

### New features¶

• Added Conversion methods to_qobj() to operators and variational states, that produce QuTiP’s qobjects.

• A function nk.nn.activation.reim has been added that transforms a nonlinearity to act seperately on the real and imaginary parts

• Nonlinearities reim_selu and reim_relu have been added

• Autoregressive Neural Networks (ARNN) now have a machine_pow field (defaults to 2) used to change the exponent used for the normalization of the wavefunction. #940.

### Breaking Changes¶

• The default initializer for netket.models.GCNN has been changed to from jax.nn.selu to netket.nn.reim_selu #892

• netket.nn.initializers has been deprecated in favor of jax.nn.initializers #935.

• Subclasses of AbstractARNN must define the field machine_pow #940

• nk.hilbert.HilbertIndex and nk.operator.spin.DType are now unexported (they where never intended to be visible). #904

• AbstractOperators have been renamed DiscreteOperators. AbstractOperators still exist, but have almost no functionality and they are intended as the base class for more arbitrary (eg. continuous space) operators. If you have defined a custom operator inheriting from AbstractOperator you should change it to derive from DiscreteOperator. #929

### Internal Changes¶

• PermutationGroup.product_table now consumes less memory and is more performant. This is helpfull when working with large symmetry groups. #884 #891

• Added size check to DiscreteOperator.get_conn and throw helpful error messages if those do not match. #927

• The internal numba4jax module has been factored out into a standalone library, named (how original) numba4jax. This library was never intended to be used by external users, but if for any reason you were using it, you should switch to the external library. #934

• netket.jax now includes several batching utilities like batched_vmap and batched_vjp. Those can be used to build memory efficient batched code, but are considered internal, experimental and might change without warning. #925.

### Bug Fixes¶

• Autoregressive networks now work with Qubit hilbert spaces. #937

## NetKet 3.0 (23 august 2021)¶

### Breaking Changes¶

• The default initializer for netket.nn.Dense layers now matches the same default as flax.linen, and it is lecun_normal instead of normal(0.01) #869

• The default initializer for netket.nn.DenseSymm layers is now chosen in order to give variance 1 to every output channel, therefore defaulting to lecun_normal #870

## NetKet 3.0b4 (17 august 2021)¶

### New features¶

• DenseSymm now accepts a mode argument to specify whever the symmetries should be computed with a full dense matrix or FFT. The latter method is much faster for sufficiently large systems. Other kwargs have been added to satisfy the interface. The api changes are also reflected in RBMSymm and GCNN. #792

### Breaking Changes¶

• The so-called legacy netket in netket.legacy has been removed. #773

### Internal Changes¶

• The methods expect and expect_and_grad of MCState now use dispatch to select the relevant implementation of the algorithm. They can therefore be expanded and overridden without editing NetKet’s source code. #804

• netket.utils.mpi_available has been moved to netket.utils.mpi.available to have a more consistent api interface (all mpi-related properties in the same submodule). #827

• netket.logging.TBLog has been renamed to netket.logging.TensorBoardLog for better readability. A deprecation warning is now issued if the older name is used #827

• When MCState initializes a model by calling model.init, the call is now jitted. This should speed it up for non-trivial models but might break non-jit invariant models. #832

• operator.get_conn_padded now supports arbitrarily-dimensioned bitstrings as input and reshapes the output accordingly. #834

• NetKet’s implementation of dataclasses now support pytree_node=True/False on cached properties. #835

• Plum version has been bumped to 1.5.1 to avoid broken versions (1.4, 1.5). #856.

• Numba version 0.54 is now allowed #857.

### Bug Fixes¶

• Fix Progress bar bug. #810

• Make the repr/printing of history objects nicer in the REPL. #819

• The field MCState.model is now read-only, to prevent user errors. #822

• The order of the operators in PauliString does no longer influences the estimate of the number of non-zero connected elements. #836

## NetKet 3.0b3 (published on 9 july 2021)¶

### New features¶

• The netket.utils.group submodule provides utilities for geometrical and permutation groups. Lattice (and its specialisations like Grid) use these to automatically construct the space groups of lattices, as well as their character tables for generating wave functions with broken symmetry. #724

• Autoregressive neural networks, sampler, and masked linear layers have been added to models, sampler and nn #705.

### Breaking Changes¶

• The netket.graph.Grid class has been removed. netket.graph.Grid will now return an instance of graph.Lattice supporting the same API but with new functionalities related to spatial symmetries. The color_edges optional keyword argument has been removed without deprecation. #724

• MCState.n_discard has been renamed MCState.n_discard_per_chain and the old binding has been deprecated #739.

• nk.optimizer.qgt.QGTOnTheFly option centered=True has been removed because we are now convinced the two options yielded equivalent results. QGTOnTheFly now always behaves as if centered=False #706.

### Internal Changes¶

• networkX has been replaced by igraph, yielding a considerable speedup for some graph-related operations #729.

• netket.hilbert.random module now uses plum-dispatch (through netket.utils.dispatch) to select the correct implementation of random_state and flip_state. This makes it easy to define new hilbert states and extend their functionality easily. #734.

• The AbstractHilbert interface is now much smaller in order to also support continuous Hilbert spaces. Any functionality specific to discrete hilbert spaces (what was previously supported) has been moved to a new abstract type netket.hilbert.DiscreteHilbert. Any Hilbert space previously subclassing netket.hilbert.AbstractHilbert should be modified to subclass netket.hilbert.DiscreteHilbert #800.

### Bug Fixes¶

• nn.to_array and MCState.to_array, if normalize=False, do not subtract the logarithm of the maximum value from the state #705.

• Autoregressive networks now work with Fock space and give correct errors if the hilbert space is not supported #806.

• Autoregressive networks are now much (x10-x100) faster #705.

• Do not throw errors when calling operator.get_conn_flattened(states) with a jax array #764.

• Fix bug with the driver progress bar when step_size != 1 #747.

## NetKet 3.0b2 (published on 31 May 2021)¶

### New features¶

• Group Equivariant Neural Networks have been added to models #620

• Permutation invariant RBM and Permutation invariant dense layer have been added to models and nn.linear #573

• Add the property acceptance to MetropolisSampler’s SamplerState, computing the MPI-enabled acceptance ratio. #592.

• Add StateLog, a new logger that stores the parameters of the model during the optimization in a folder or in a tar file. #645

• A warning is now issued if NetKet detects to be running under mpirun but MPI dependencies are not installed #631

• operator.LocalOperators now do not return a zero matrix element on the diagonal if the whole diagonal is zero. #623.

• logger.JSONLog now automatically flushes at every iteration if it does not consume significant CPU cycles. #599

• The interface of Stochastic Reconfiguration has been overhauled and made more modular. You can now specify the solver you wish to use, NetKet provides some dense solvers out of the box, and there are 3 different ways to compute the Quantum Geometric Tensor. Read the documentation to learn more about it. #674

• Unless you specify the QGT implementation you wish to use with SR, we use an automatic heuristic based on your model and the solver to pick one. This might affect SR performance. #674

### Breaking Changes¶

• For all samplers, n_chains now sets the total number of chains across all MPI ranks. This is a breaking change compared to the old API, where n_chains would set the number of chains on a single MPI rank. It is still possible to set the number of chains per MPI rank by specifying n_chains_per_rank instead of n_chains. This change, while breaking allows us to be consistent with the interface of variational.MCState, where n_samples is the total number of samples across MPI nodes.

• MetropolisSampler.reset_chain has been renamed to MetropolisSampler.reset_chains. Likewise in the constructor of all samplers.

• Briefly during development releases MetropolisSamplerState.acceptance_ratio returned the percentage (not ratio) of acceptance. acceptance_ratio is now deprecated in favour of the correct acceptance.

• models.Jastrow now internally symmetrizes the matrix before computing its value #644

• MCState.evaluate has been renamed to MCState.log_value #632

• nk.optimizer.SR no longer accepts keyword argument relative to the sparse solver. Those should be passed inside the closure or functools.partial passed as solver argument.

• nk.optimizer.sr.SRLazyCG and nk.optimizer.sr.SRLazyGMRES have been deprecated and will soon be removed.

• Parts of the Lattice API have been overhauled, with deprecations of several methods in favor of a consistent usage of Lattice.position for real-space location of sites and Lattice.basis_coords for location of sites in terms of basis vectors. Lattice.sites has been added, which provides a sequence of LatticeSite objects combining all site properties. Furthermore, Lattice now provides lookup of sites from their position via id_from_position using a hashing scheme that works across periodic boundaries. #703 #715

• nk.variational has been renamed to nk.vqs and will be removed in a future release.

### Bug Fixes¶

• Fix operator.BoseHubbard usage under jax Hamiltonian Sampling #662

• Fix SROnTheFly for R->C models with non homogeneous parameters #661

• Fix MPI Compilation deadlock when computing expectation values #655

• Fix bug preventing the creation of a hilbert.Spin Hilbert space with odd sites and even S. #641

• Fix bug #635 preventing the usage of NumpyMetropolisSampler with MCState.expect #635

• Fix bug #635 where the graph.Lattice was not correctly computing neighbours because of floating point issues. #633

• Fix bug the Y Pauli matrix, which was stored as its conjugate. #618 #617 #615

## NetKet 3.0b1 (published beta release)¶

### API Changes¶

• Hilbert space constructors do not store the lattice graph anymore. As a consequence, the constructor does not accept the graph anymore.

• Special Hamiltonians defined on a lattice, such as operator.BoseHubbard, operator.Ising and operator.Heisenberg, now require the graph to be passed explicitly through a graph keyword argument.

• operator.LocalOperator now default to real-valued matrix elements, except if you construct them with a complex-valued matrix. This is also valid for operators such as :func:operator.spin.sigmax and similars.

• When performing algebraic operations *, -, + on pairs of operator.LocalOperator, the dtype of the result iscomputed using standard numpy promotion logic.

• Doing an operation in-place +=, -=, *= on a real-valued operator will now fail if the other is complex. While this might seem annoying, it’s useful to ensure that smaller types such as float32 or complex64 are preserved if the user desires to do so.

• AbstractMachine has been removed. It’s functionality is now split among the model itself, which is defined by the user and variational.MCState for pure states or variational.MCMixedState for mixed states.

• The model, in general is composed by two functions, or an object with two functions: an init(rng, sample_val) function, accepting a jax.random.PRNGKey() object and an input, returning the parameters and the state of the model for that particular sample shape, and a apply(params, samples, **kwargs) function, evaluating the model for the given parameters and inputs.

• Some models (previously machines) such as the RBM (Restricted Boltzmann Machine) Machine, NDM (Neural Density Matrix) or MPS (Matrix Product State ansatz) are available in Pre-built models.

• Machines, now called models, should be written using Flax or another jax framework.

• Serialization and deserialization functionality has now been moved to netket.variational.MCState, which support the standard Flax interface through MsgPack. See Flax docs for more information

• AbstractMachine.init_random_parameters functionality has now been absorbed into netket.vqs.VariationalState.init_parameters(), which however has a different syntax.

• Samplers now require the Hilbert space upon which they sample to be passed in to the constructor. Also note that several keyword arguments of the samplers have changed, and new one are available.

• It’s now possible to change Samplers dtype, which controls the type of the output. By default they use double-precision samples (np.float64). Be wary of type promotion issues with your models.

• Samplers no longer take a machine as an argument.

• Samplers are now immutable (frozen) dataclasses (defined through flax.struct.dataclass) that only hold the sampling parameters. As a consequence it is no longer possible to change their settings such as n_chains or n_sweeps without creating a new sampler. If you wish to update only one parameter, it is possible to construct the new sampler with the updated value by using the sampler.replace(parameter=new_value) function.

• Samplers are no longer stateful objects. Instead, they can construct an immutable state object netket.sampler.init_state, which can be passed to sampling functions such as netket.sampler.sample, which now return also the updated state. However, unless you have particular use-cases we advise you use the variational state MCState instead.

• The netket.optimizer module has been overhauled, and now only re-exports flax optim module. We advise not to use netket’s optimizer but instead to use optax .

• The netket.optimizer.SR object now is only a set of options used to compute the SR matrix. The SR matrix, now called quantum_geometric_tensor can be obtained by calling variational.MCState.quantum_geometric_tensor(). Depending on the settings, this can be a lazy object.

• netket.Vmc has been renamed to netket.VMC

• netket.models.RBM replaces the old RBM machine, but has real parameters by default.

• As we rely on Jax, using dtype=float or dtype=complex, which are weak types, will sometimes lead to loss of precision because they might be converted to float32. Use np.float64 or np.complex128 instead if you want double precision when defining your models.