A JIT compiler for hybrid quantum programs in PennyLane
APACHE-2.0 License
Bot releases are hidden (Show)
Catalyst now provides a QJIT compatible catalyst.vmap
function, which makes it even easier to modify functions to map over inputs with additional batch dimensions. (#497) (#569)
When working with tensor/array frameworks in Python, it can be important to ensure that code is written to minimize usage of Python for loops (which can be slow and inefficient), and instead push as much of the computation through to the array manipulation library, by taking advantage of extra batch dimensions.
For example, consider the following QNode:
dev = qml.device("lightning.qubit", wires=1)
@qml.qnode(dev)
def circuit(x, y):
qml.RX(jnp.pi * x[0] + y, wires=0)
qml.RY(x[1] ** 2, wires=0)
qml.RX(x[1] * x[2], wires=0)
return qml.expval(qml.PauliZ(0))
>>> circuit(jnp.array([0.1, 0.2, 0.3]), jnp.pi)
Array(-0.93005586, dtype=float64)
We can use catalyst.vmap
to introduce additional batch dimensions to our input arguments, without needing to use a Python for loop:
>>> x = jnp.array([[0.1, 0.2, 0.3],
... [0.4, 0.5, 0.6],
... [0.7, 0.8, 0.9]])
>>> y = jnp.array([jnp.pi, jnp.pi / 2, jnp.pi / 4])
>>> qjit(vmap(cost))(x, y)
array([-0.93005586, -0.97165424, -0.6987465 ])
catalyst.vmap()
has been implemented to match the same behaviour of jax.vmap
, so should be a drop-in replacement in most cases. Under-the-hood, it is automatically inserting Catalyst-compatible for loops, which will be compiled and executed outside of Python for increased performance.
Catalyst now supports compiling and executing QJIT-compiled QNodes using the CUDA Quantum compiler toolchain. (#477) (#536) (#547)
Simply import the CUDA Quantum @cudaqjit
decorator to use this functionality:
from catalyst.cuda import cudaqjit
Or, if using Catalyst from PennyLane, simply specify @qml.qjit(compiler="cuda_quantum")
.
The following devices are available when compiling with CUDA Quantum:
softwareq.qpp
: a modern C++ statevector simulatornvidia.custatevec
: The NVIDIA CuStateVec GPU simulator (with support for multi-gpu)nvidia.cutensornet
: The NVIDIA CuTensorNet GPU simulator (with support for matrix product state)For example:
dev = qml.device("softwareq.qpp", wires=2)
@cudaqjit
@qml.qnode(dev)
def circuit(x):
qml.RX(x[0], wires=0)
qml.RY(x[1], wires=1)
qml.CNOT(wires=[0, 1])
return qml.expval(qml.PauliY(0))
>>> circuit(jnp.array([0.5, 1.4]))
-0.47244976756708373
Note that CUDA Quantum compilation currently does not have feature parity with Catalyst compilation; in particular, AutoGraph, control flow, differentiation, and various measurement statistics (such as probabilities and variance) are not yet supported. Classical code support is also limited.
Catalyst now supports just-in-time compilation of static (compile-time constant) arguments. (#476) (#550)
The @qjit
decorator takes a new argument static_argnums
, which specifies positional arguments of the decorated function should be treated as compile-time static arguments.
This allows any hashable Python object to be passed to the function during compilation; the function will only be re-compiled if the hash value of the static arguments change. Otherwise, re-using previous static argument values will result in no re-compilation.
@qjit(static_argnums=(1,))
def f(x, y):
print(f"Compiling with y={y}")
return x + y
>>> f(0.5, 0.3)
Compiling with y=0.3
array(0.8)
>>> f(0.1, 0.3) # no re-compilation occurs
array(0.4)
>>> f(0.1, 0.4) # y changes, re-compilation
Compiling with y=0.4
array(0.5)
This functionality can be used to support passing arbitrary Python objects to QJIT-compiled functions, as long as they are hashable:
from dataclasses import dataclass
@dataclass
class MyClass:
val: int
def __hash__(self):
return hash(str(self))
@qjit(static_argnums=(1,))
def f(x: int, y: MyClass):
return x + y.val
>>> f(1, MyClass(5))
array(6)
>>> f(1, MyClass(6)) # re-compilation
array(7)
>>> f(2, MyClass(5)) # no re-compilation
array(7)
Mid-circuit measurements now support post-selection and qubit reset when used with the Lightning simulators. (#491) (#507)
To specify post-selection, simply pass the postselect
argument to the catalyst.measure
function:
dev = qml.device("lightning.qubit", wires=1)
@qjit
@qml.qnode(dev)
def f():
qml.Hadamard(0)
m = measure(0, postselect=1)
return qml.expval(qml.PauliZ(0))
Likewise, to reset a wire after mid-circuit measurement, simply specify reset=True
:
dev = qml.device("lightning.qubit", wires=1)
@qjit
@qml.qnode(dev)
def f():
qml.Hadamard(0)
m = measure(0, reset=True)
return qml.expval(qml.PauliZ(0))
Catalyst now supports Python 3.12 (#532)
The JAX version used by Catalyst has been updated to v0.4.23
. (#428)
Catalyst now supports the qml.GlobalPhase
operation. (#563)
Native support for qml.PSWAP
and qml.ISWAP
gates on Amazon Braket devices has been added. (#458)
Specifically, a circuit like
dev = qml.device("braket.local.qubit", wires=2, shots=100)
@qjit
@qml.qnode(dev)
def f(x: float):
qml.Hadamard(0)
qml.PSWAP(x, wires=[0, 1])
qml.ISWAP(wires=[1, 0])
return qml.probs()
would no longer decompose the PSWAP
and ISWAP
gates.
The qml.BlockEncode
operator is now supported with Catalyst. (#483)
Catalyst no longer relies on a TensorFlow installation for its AutoGraph functionality. Instead, the standalone diastatic-malt
package is used and automatically installed as a dependency. (#401)
The @qjit
decorator will remember previously compiled functions when the PyTree metadata of arguments changes, in addition to also remembering compiled functions when static arguments change. (#522)
The following example will no longer trigger a third compilation:
@qjit
def func(x):
print("compiling")
return x
>>> func([1,]); # list
compiling
>>> func((2,)); # tuple
compiling
>>> func([3,]); # list
Note however that in order to keep overheads low, changing the argument type or shape (in a promotion incompatible way) may override a previously stored function (with identical PyTree metadata and static argument values):
@qjit
def func(x):
print("compiling")
return x
>>> func(jnp.array(1)); # scalar
compiling
>>> func(jnp.array([2.])); # 1-D array
compiling
>>> func(jnp.array(3)); # scalar
compiling
Catalyst gradient functions (grad
, jacobian
, vjp
, and jvp
) now support being applied to functions that use (nested) container types as inputs and outputs. This includes lists and dictionaries, as well as any data structure implementing the PyTree protocol. (#500) (#501) (#508) (#549)
dev = qml.device("lightning.qubit", wires=1)
@qml.qnode(dev)
def circuit(phi, psi):
qml.RY(phi, wires=0)
qml.RX(psi, wires=0)
return [{"expval0": qml.expval(qml.PauliZ(0))}, qml.expval(qml.PauliZ(0))]
psi = 0.1
phi = 0.2
>>> qjit(jacobian(circuit, argnum=[0, 1]))(psi, phi)
[{'expval0': (array(-0.0978434), array(-0.19767681))}, (array(-0.0978434), array(-0.19767681))]
Support has been added for linear algebra functions which depend on computing the eigenvalues of symmetric matrices, such as np.sqrt_matrix()
. (#488)
For example, you can compile qml.math.sqrt_matrix
:
@qml.qjit
def workflow(A):
B = qml.math.sqrt_matrix(A)
return B @ A
Internally, this involves support for lowering the eigenvectors/values computation lapack method lapack_dsyevd
via stablehlo.custom_call
.
Additional debugging functions are now available in the catalyst.debug
directory. (#529) (#522)
This includes:
filter_static_args(args, static_argnums)
to remove static values from arguments using the
provided index list.
get_cmain(fn, *args)
to return a C program that calls a jitted function with the provided
arguments.
print_compilation_stage(fn, stage)
to print one of the recorded compilation stages for a
JIT-compiled function.
For more details, please see the catalyst.debug
documentation.
Remove redundant copies of TOML files for lightning.kokkos
and lightning.qubit
. (#472)
lightning.kokkos
and lightning.qubit
now ship with their own TOML file. As such, we use the TOML file provided by them.
Capturing quantum circuits with many gates prior to compilation is now quadratically faster (up to a factor), by removing qextract_p
and qinst_p
from forced-order primitives. (#469)
Update AllocateQubit
and AllocateQubits
in LightningKokkosSimulator
to preserve the current state-vector before qubit re-allocations in the runtime dynamic qubits management. (#479)
The PennyLane custom compiler entry point name convention has changed, necessitating a change to the Catalyst entry points. (#493)
Catalyst gradient functions now match the Jax convention for the returned axes of gradients, Jacobians, VJPs, and JVPs. As a result, the returned tensor shape from various Catalyst gradient functions may differ compared to previous versions of Catalyst. (#500) (#501) (#508)
The Catalyst Python frontend has been partially refactored. The impact on user-facing functionality is minimal, but the location of certain classes and methods used by the package may have changed. (#529) (#522)
The following changes have been made:
Some debug methods and features on the QJIT class have been turned into free functions and moved to the catalyst.debug
module, which will now appear in the public documention. This includes compiling a program from IR, obtaining a C program to invoke a compiled function from, and printing fine-grained MLIR compilation stages.
The compilation_pipelines.py
module has been renamed to jit.py
, and certain functionality has been moved out (see following items).
A new module compiled_functions.py
now manages low-level access to compiled functions.
A new module tracing/type_signatures.py
handles functionality related managing arguments and type signatures during the tracing process.
The contexts.py
module has been moved from utils
to the new tracing
sub-module.
Changes to the runtime QIR API and dependencies, to avoid symbol conflicts with other libraries that utilize QIR. (#464) (#470)
The existing Catalyst runtime implements QIR as a library that can be linked against a QIR module. This works great when Catalyst is the only implementor of QIR, however it may generate symbol conflicts when used alongside other QIR implementations.
To avoid this, two changes were necessary:
The Catalyst runtime now has a different API from QIR instructions.
The runtime has been modified such that QIR instructions are lowered to functions where the __quantum__
part of the function name is replaced with __catalyst__
. This prevents the possibility of symbol conflicts with other libraries that implement QIR as a library.
The Catalyst runtime no longer depends on QIR runner's stdlib.
We no longer depend nor link against QIR runner's stdlib. By linking against QIR runner's stdlib, some definitions persisted that may be different than ones used by third party implementors. To prevent symbol conflicts QIR runner's stdlib was removed and is no longer linked against. As a result, the following functions are now defined and implemented in Catalyst's runtime:
int64_t __catalyst__rt__array_get_size_1d(QirArray *)
int8_t *__catalyst__rt__array_get_element_ptr_1d(QirArray *, int64_t)
and the following functions were removed since the frontend does not generate them
QirString *__catalyst__rt__qubit_to_string(QUBIT *)
QirString *__catalyst__rt__result_to_string(RESULT *)
Fix an issue when no qubit number was specified for the qinst
primitive. The primitive now correctly deduces the number of qubits when no gate parameters are present. This change is not user facing. (#496)
Fixed a bug where differentiation of sliced arrays would result in an error. (#552)
def f(x):
return jax.numpy.sum(x[::2])
x = jax.numpy.array([0.1, 0.2, 0.3, 0.4])
>>> catalyst.qjit(catalyst.grad(f))(x)
[1. 0. 1. 0.]
Fixed a bug where quantum control applied to a subcircuit was not correctly mapping wires, and the wires in the nested region remained unchanged. (#555)
Catalyst will no longer print a warning that recompilation is triggered when a @qjit
decorated function with no arguments is invoke without having been compiled first, for example via the use of target="mlir"
. (#522)
Fixes a bug in the configuration of dynamic shaped arrays that would cause certain program to error with TypeError: cannot unpack non-iterable ShapedArray object
. (#526)
This is fixed by replacing the code which updates the JAX_DYNAMIC_SHAPES
option with a transient_jax_config()
context manager which temporarily sets the value of JAX_DYNAMIC_SHAPES
to True and then restores the original configuration value following the yield. The context manager is used by trace_to_jaxpr()
and lower_jaxpr_to_mlir()
.
Exceptions encountered in the runtime when using the @qjit
option async_qnodes=Tue
will now be properly propagated to the frontend. (#447) (#510)
This is done by:
llvm.call
to llvm.invoke
Fixes a bug when computing gradients with the indexing/slicing, by fixing the scatter operation lowering when updatedWindowsDim
is empty. (#475)
Fix the issue in LightningKokkos::AllocateQubits
with allocating too many qubit IDs on qubit re-allocation. (#473)
Fixed an issue where wires was incorrectly set as <Wires = [<WiresEnum.AnyWires: -1>]>
when using catalyst.adjoint
and catalyst.ctrl
, by adding a wires
property to these operations. (#480)
Fix the issue with multiple lapack symbol definitions in the compiled program by updating the stablehlo.custom_call
conversion pass. (#488)
This release contains contributions from (in alphabetical order):
Mikhail Andrenkov,
Ali Asadi,
David Ittah,
Tzung-Han Juang,
Erick Ochoa Lopez,
Romain Moyard,
Raul Torres,
Haochen Paul Wang.
Published by josh146 9 months ago
Catalyst wheels are now packaged with OpenMP and ZStd, which avoids installing additional requirements separately in order to use pre-packaged Catalyst binaries. (#457) (#478)
Note that OpenMP support for the lightning.kokkos
backend has been disabled on macOS x86_64, due to memory issues in the computation of Lightning's adjoint-jacobian in the presence of multiple OMP threads.
Resolve an infinite recursion in the decomposition of the Controlled
operator whenever computing a Unitary matrix for the operator fails. (#468)
Resolve a failure to generate gradient code for specific input circuits. (#439)
In this case, jnp.mod
was used to compute wire values in a for loop, which prevented the gradient architecture from fully separating quantum and classical code. The following program is now supported:
@qjit
@grad
@qml.qnode(dev)
def f(x):
def cnot_loop(j):
qml.CNOT(wires=[j, jnp.mod((j + 1), 4)])
for_loop(0, 4, 1)(cnot_loop)()
return qml.expval(qml.PauliZ(0))
Resolve unpredictable behaviour when importing libraries that share Catalyst's LLVM dependency (e.g. TensorFlow). In some cases, both packages exporting the same symbols from their shared libraries can lead to process crashes and other unpredictable behaviour, since the wrong functions can be called if both libraries are loaded in the current process. The fix involves building shared libraries with hidden (macOS) or protected (linux) symbol visibility by default, exporting only what is necessary. (#465)
Resolve a failure to find the SciPy OpenBLAS library when running Catalyst, due to a different SciPy version being used to build Catalyst than to run it. (#471)
Resolve a memory leak in the runtime stemming from missing calls to device destructors at the end of programs. (#446)
This release contains contributions from (in alphabetical order):
Ali Asadi, David Ittah.
Published by dime10 9 months ago
Catalyst is now accessible directly within the PennyLane user interface, once Catalyst is installed, allowing easy access to Catalyst just-in-time functionality.
Through the use of the qml.qjit
decorator, entire workflows can be JIT compiled down to a machine binary on first-function execution, including both quantum and classical processing. Subsequent calls to the compiled function will execute the previously-compiled binary, resulting in significant performance improvements.
import pennylane as qml
dev = qml.device("lightning.qubit", wires=2)
@qml.qjit
@qml.qnode(dev)
def circuit(theta):
qml.Hadamard(wires=0)
qml.RX(theta, wires=1)
qml.CNOT(wires=[0, 1])
return qml.expval(qml.PauliZ(wires=1))
>>> circuit(0.5) # the first call, compilation occurs here
array(0.)
>>> circuit(0.5) # the precompiled quantum function is called
array(0.)
Currently, PennyLane supports the Catalyst hybrid compiler with the qml.qjit
decorator, which directly aliases Catalyst's catalyst.qjit
.
In addition to the above qml.qjit
integration, the following native PennyLane functions can now be used with the qjit
decorator: qml.adjoint
, qml.ctrl
, qml.grad
, qml.jacobian
, qml.vjp
, qml.jvp
, and qml.adjoint
, qml.while_loop
, qml.for_loop
, qml.cond
. These will alias to the corresponding Catalyst functions when used within a qjit
context.
For more details on these functions, please refer to the PennyLane compiler documentation and compiler module documentation.
Just-in-time compiled functions now support asynchronous execution of QNodes. (#374) (#381) (#420) (#424) (#433)
Simply specify async_qnodes=True
when using the @qjit
decorator to enable the async execution of QNodes. Currently, asynchronous execution is only supported by lightning.qubit
and lightning.kokkos
.
Asynchronous execution will be most beneficial for just-in-time compiled functions that contain --- or generate --- multiple QNodes.
For example,
dev = qml.device("lightning.qubit", wires=2)
@qml.qnode(device=dev)
def circuit(params):
qml.RX(params[0], wires=0)
qml.RY(params[1], wires=1)
qml.CNOT(wires=[0, 1])
return qml.expval(qml.PauliZ(wires=0))
@qjit(async_qnodes=True)
def multiple_qnodes(params):
x = jnp.sin(params)
y = jnp.cos(params)
z = jnp.array([circuit(x), circuit(y)]) # will be executed in parallel
return circuit(z)
>>> func(jnp.array([1.0, 2.0]))
1.0
Here, the first two circuit executions will occur in parallel across multiple threads, as their execution can occur independently.
Preliminary support for PennyLane transforms has been added. (#280)
@qjit
@qml.transforms.split_non_commuting
@qml.qnode(dev)
def circuit(x):
qml.RX(x,wires=0)
return [qml.expval(qml.PauliY(0)), qml.expval(qml.PauliZ(0))]
>>> circuit(0.4)
[array(-0.51413599), array(0.85770868)]
Currently, most PennyLane transforms will work with Catalyst as long as:
The circuit does not include any Catalyst-specific features, such
as Catalyst control flow or measurement,
The QNode returns only lists of measurement processes,
AutoGraph is disabled, and
The transformation does not require or depend on the numeric value of
dynamic variables.
Catalyst now supports just-in-time compilation of dynamically-shaped arrays. (#366) (#386) (#390) (#411)
The @qjit
decorator can now be used to compile functions that accepts or contain tensors whose dimensions are not known at compile time; runtime execution with different shapes is supported without recompilation.
In addition, standard tensor initialization functions jax.numpy.ones
, jnp.zeros
, and jnp.empty
now accept dynamic variables (where the value is only known at runtime).
@qjit
def func(size: int):
return jax.numpy.ones([size, size], dtype=float)
>>> func(3)
[[1. 1. 1.]
[1. 1. 1.]
[1. 1. 1.]]
When passing tensors as arguments to compiled functions, the abstracted_axes
keyword argument to the @qjit
decorator can be used to specify which axes of the input arguments should be treated as abstract (and thus avoid recompilation).
For example, without specifying abstracted_axes
, the following sum
function would recompile each time an array of different size is passed as an argument:
>>> @qjit
>>> def sum_fn(x):
>>> return jnp.sum(x)
>>> sum_fn(jnp.array([1])) # Compilation happens here.
>>> sum_fn(jnp.array([1, 1])) # And here!
By passing abstracted_axes
, we can specify that the first axes of the first argument is to be treated as dynamic during initial compilation:
>>> @qjit(abstracted_axes={0: "n"})
>>> def sum_fn(x):
>>> return jnp.sum(x)
>>> sum_fn(jnp.array([1])) # Compilation happens here.
>>> sum_fn(jnp.array([1, 1])) # No need to recompile.
Note that support for dynamic arrays in control-flow primitives (such as loops), is not yet supported.
Error mitigation using the zero-noise extrapolation method is now available through the catalyst.mitigate_with_zne
transform. (#324) (#414)
For example, given a noisy device (such as noisy hardware available through Amazon Braket):
dev = qml.device("noisy.device", wires=2)
@qml.qnode(device=dev)
def circuit(x, n):
@for_loop(0, n, 1)
def loop_rx(i):
qml.RX(x, wires=0)
loop_rx()
qml.Hadamard(wires=0)
qml.RZ(x, wires=0)
loop_rx()
qml.RZ(x, wires=0)
qml.CNOT(wires=[1, 0])
qml.Hadamard(wires=1)
return qml.expval(qml.PauliY(wires=0))
@qjit
def mitigated_circuit(args, n):
s = jax.numpy.array([1, 2, 3])
return mitigate_with_zne(circuit, scale_factors=s)(args, n)
>>> mitigated_circuit(0.2, 5)
0.5655341100116512
In addition, a mitigation dialect has been added to the MLIR layer of Catalyst. It contains a Zero Noise Extrapolation (ZNE) operation, with a lowering to a global folded circuit.
The three backend devices provided with Catalyst, lightning.qubit
, lightning.kokkos
, and braket.aws
, are now dynamically loaded at runtime. (#343) (#400)
This takes advantage of the new backend plugin system provided in Catalyst v0.3.2, and allows the devices to be packaged separately from the runtime CAPI. Provided backend devices are now loaded at runtime, instead of being linked at compile time.
For more details on the backend plugin system, see the custom devices documentation.
Finite-shot measurement statistics (expval
, var
, and probs
) are now supported for the lightning.qubit
and lightning.kokkos
devices. Previously, exact statistics were returned even when finite shots were specified. (#392) (#410)
>>> dev = qml.device("lightning.qubit", wires=2, shots=100)
>>> @qjit
>>> @qml.qnode(dev)
>>> def circuit(x):
>>> qml.RX(x, wires=0)
>>> return qml.probs(wires=0)
>>> circuit(0.54)
array([0.94, 0.06])
>>> circuit(0.54)
array([0.93, 0.07])
Catalyst gradient functions grad
, jacobian
, jvp
, and vjp
can now be invoked from outside a @qjit
context. (#375)
This simplifies the process of writing functions where compilation can be turned on and off easily by adding or removing the decorator. The functions dispatch to their JAX equivalents when the compilation is turned off.
dev = qml.device("lightning.qubit", wires=2)
@qml.qnode(dev)
def circuit(x):
qml.RX(x, wires=0)
return qml.expval(qml.PauliZ(0))
>>> grad(circuit)(0.54) # dispatches to jax.grad
Array(-0.51413599, dtype=float64, weak_type=True)
>>> qjit(grad(circuit))(0.54). # differentiates using Catalyst
array(-0.51413599)
New lightning.qubit
configuration options are now supported via the qml.device
loader, including Markov Chain Monte Carlo sampling support. (#369)
dev = qml.device("lightning.qubit", wires=2, shots=1000, mcmc=True)
@qml.qnode(dev)
def circuit(x):
qml.RX(x, wires=0)
return qml.expval(qml.PauliZ(0))
>>> circuit(0.54)
array(0.856)
Improvements have been made to the runtime and quantum MLIR dialect in order to support asynchronous execution.
The runtime now supports multiple active devices managed via a device pool. The new RTDevice
data-class and RTDeviceStatus
along with the thread_local
device instance pointer enable the runtime to better scope the lifetime of device instances concurrently. With these changes, one can create multiple active devices and execute multiple programs in a multithreaded environment. (#381)
The ability to dynamically release devices has been added via DeviceReleaseOp
in the Quantum MLIR dialect. This is lowered to the __quantum__rt__device_release()
runtime instruction, which updates the status of the device instance from Active
to Inactive
. The runtime will reuse this deactivated instance instead of creating a new one automatically at runtime in a multi-QNode workflow when another device with identical specifications is requested. (#381)
The DeviceOp
definition in the Quantum MLIR dialect has been updated to lower a tuple of device information ('lib', 'name', 'kwargs')
to a single device initialization call __quantum__rt__device_init(int8_t *, int8_t *, int8_t *)
. This allows the runtime to initialize device instances without keeping partial information of the device. (#396)
The quantum adjoint compiler routine has been extended to support function calls that affect the quantum state within an adjoint region. Note that the function may only provide a single result consisting of the quantum register. By itself this provides no user-facing changes, but compiler pass developers may now generate quantum adjoint operations around a block of code containing function calls as well as quantum operations and control flow operations. (#353)
The allocation and deallocation operations in MLIR (AllocOp
, DeallocOp
) now follow simple value semantics for qubit register values, instead of modelling memory in the MLIR trait system. Similarly, the frontend generates proper value semantics by deallocating the final register value.
The change enables functions at the MLIR level to accept and return quantum register values, which would otherwise not be correctly identified as aliases of existing register values by the bufferization system. (#360)
Resolves a bug in the compiler's differentiation engine that results in a segmentation fault when attempting to differentiate non-differentiable quantum operations. The fix ensures that all existing quantum operation types are removed during gradient passes that extract classical code from a QNode function. It also adds a verification step that will raise an error if a gradient pass cannot successfully eliminate all quantum operations for such functions. (#397)
Resolves a bug that caused unpredictable behaviour when printing string values with the debug.print
function. The issue was caused by non-null-terminated strings. (#418)
This release contains contributions from (in alphabetical order):
Ali Asadi,
David Ittah,
Romain Moyard,
Sergei Mironov,
Erick Ochoa Lopez,
Shuli Shu.
Published by dime10 11 months ago
This post-release updates the docs with up-to-date information & additional sections for the installation guide.
Published by dime10 11 months ago
The experimental AutoGraph feature now supports Python while
loops, allowing native Python loops to be captured and compiled with Catalyst. (#318)
dev = qml.device("lightning.qubit", wires=4)
@qjit(autograph=True)
@qml.qnode(dev)
def circuit(n: int, x: float):
i = 0
while i < n:
qml.RX(x, wires=i)
i += 1
return qml.expval(qml.PauliZ(0))
>>> circuit(4, 0.32)
array(0.94923542)
This feature extends the existing AutoGraph support for Python for
loops and if
statements introduced in v0.3. Note that TensorFlow must be installed for AutoGraph support.
For more details, please see the AutoGraph guide.
In addition to loops and conditional branches, AutoGraph now supports native Python and
, or
and not
operators in Boolean expressions. (#325)
dev = qml.device("lightning.qubit", wires=1)
@qjit(autograph=True)
@qml.qnode(dev)
def circuit(x: float):
if x >= 0 and x < jnp.pi:
qml.RX(x, wires=0)
return qml.probs()
>>> circuit(0.43)
array([0.95448287, 0.04551713])
>>> circuit(4.54)
array([1., 0.])
Note that logical Boolean operators will only be captured by AutoGraph if all operands are dynamic variables (that is, a value known only at runtime, such as a measurement result or function argument). For other use cases, it is recommended to use the jax.numpy.logical_*
set of functions where appropriate.
Debug compiled programs and print dynamic values at runtime with debug.print
(#279) (#356)
You can now print arbitrary values from your running program, whether they are arrays, constants, strings, or abitrary Python objects. Note that while non-array Python objects will be printed at runtime, their string representation is captured at compile time, and thus will always be the same regardless of program inputs. The output for arrays optionally includes a descriptor for how the data is stored in memory ("memref").
@qjit
def func(x: float):
debug.print(x, memref=True)
debug.print("exit")
>>> func(jnp.array(0.43))
MemRef: base@ = 0x5629ff2b6680 rank = 0 offset = 0 sizes = [] strides = [] data =
0.43
exit
Catalyst now officially supports macOS X86_64 devices, with macOS binary wheels available for both AARCH64 and X86_64. (#347) (#313)
It is now possible to dynamically load third-party Catalyst compatible devices directly into a pre-installed Catalyst runtime on Linux. (#327)
To take advantage of this, third-party devices must implement the Catalyst::Runtime::QuantumDevice
interface, in addition to defining the following method:
extern "C" Catalyst::Runtime::QuantumDevice*
getCustomDevice() { return new CustomDevice(); }
This support can also be integrated into existing PennyLane Python devices that inherit from the QuantumDevice
class, by defining the get_c_interface
static method.
For more details, see the custom devices documentation.
Return values of conditional functions no longer need to be of exactly the same type. Type promotion is automatically applied to branch return values if their types don't match. (#333)
@qjit
def func(i: int, f: float):
@cond(i < 3)
def cond_fn():
return i
@cond_fn.otherwise
def otherwise():
return f
return cond_fn()
>>> func(1, 4.0)
array(1.0)
Automatic type promotion across conditional branches also works with AutoGraph:
@qjit(autograph=True)
def func(i: int, f: float):
if i < 3:
i = i
else:
i = f
return i
>>> func(1, 4.0)
array(1.0)
AutoGraph now supports converting functions even when they are invoked through functional wrappers such as adjoint
, ctrl
, grad
, jacobian
, etc. (#336)
For example, the following should now succeed:
def inner(n):
for i in range(n):
qml.T(i)
@qjit(autograph=True)
@qml.qnode(dev)
def f(n: int):
adjoint(inner)(n)
return qml.state()
To prepare for Catalyst's frontend being integrated with PennyLane, the appropriate plugin entry point interface has been added to Catalyst. (#331)
For any compiler packages seeking to be registered in PennyLane, the entry_points
metadata under the the group name pennylane.compilers
must be added, with the following try points:
context
: Path to the compilation evaluation context manager. This context manager should have the method context.is_tracing()
, which returns True if called within a program that is being traced or captured.
ops
: Path to the compiler operations module. This operations module may contain compiler specific versions of PennyLane operations. Within a JIT context, PennyLane operations may dispatch to these.
qjit
: Path to the JIT compiler decorator provided by the compiler. This decorator should have the signature qjit(fn, *args, **kwargs)
, where fn
is the function to be compiled.
The compiler driver diagnostic output has been improved, and now includes failing IR as well as the names of failing passes. (#349)
The scatter operation in the Catalyst dialect now uses an SCF for loop to avoid ballooning the compiled code. (#307)
The CopyGlobalMemRefPass
pass of our MLIR processing pipeline now supports dynamically shaped arrays. (#348)
The Catalyst utility dialect is now included in the Catalyst MLIR C-API. (#345)
Fix an issue with the AutoGraph conversion system that would prevent the fallback to Python from working correctly in certain instances. (#352)
The following type of code is now supported:
@qjit(autograph=True)
def f():
l = jnp.array([1, 2])
for _ in range(2):
l = jnp.kron(l, l)
return l
catalyst.jacobian
is updated to match jax.jacobian
. Assuming we have parameters of shape [a,b]
and results of shape [c,d]
, the returned Jacobian will now have shape [c, d, a, b]
instead of [a, b, c, d]
. (#283)
An upstream change in the PennyLane-Lightning project was addressed to prevent compilation issues in the StateVectorLQubitDynamic
class in the runtime. The issue was introduced in #499. (#322)
The requirements.txt
file to build Catalyst from source has been updated with a minimum pip version, >=22.3
. Previous versions of pip are unable to perform editable installs when the system-wide site-packages are read-only, even when the --user
flag is provided. (#311)
The frontend has been updated to make it compatible with PennyLane MeasurementProcess
objects now being PyTrees in PennyLane version 0.33. (#315)
This release contains contributions from (in alphabetical order):
Ali Asadi,
David Ittah,
Sergei Mironov,
Romain Moyard,
Erick Ochoa Lopez.
Published by josh146 12 months ago
This post-release updates the docs to include the AutoGraph guide.
Published by dime10 about 1 year ago
The experimental AutoGraph feature, now supports Python for
loops, allowing native Python loops to be captured and compiled with Catalyst. (#258)
dev = qml.device("lightning.qubit", wires=n)
@qjit(autograph=True)
@qml.qnode(dev)
def f(n):
for i in range(n):
qml.Hadamard(wires=i)
return qml.expval(qml.PauliZ(0))
This feature extends the existing AutoGraph support for Python if
statements introduced in v0.3. Note that TensorFlow must be installed for AutoGraph support.
The quantum control operation can now be used in conjunction with Catalyst control flow, such as loops and conditionals, via the new catalyst.ctrl
function. (#282)
Similar in behaviour to the qml.ctrl
control modifier from PennyLane, catalyst.ctrl
can additionally wrap around quantum functions which contain control flow, such as the Catalyst cond
, for_loop
, and while_loop
primitives.
@qjit
@qml.qnode(qml.device("lightning.qubit", wires=4))
def circuit(x):
@for_loop(0, 3, 1)
def repeat_rx(i):
qml.RX(x / 2, wires=i)
catalyst.ctrl(repeat_rx, control=3)()
return qml.expval(qml.PauliZ(0))
>>> circuit(0.2)
array(1.)
Catalyst now supports JAX's array.at[index]
notation for array element assignment and updating. (#273)
@qjit
def add_multiply(l: jax.core.ShapedArray((3,), dtype=float), idx: int):
res = l.at[idx].multiply(3)
res2 = l.at[idx].add(2)
return res + res2
res = add_multiply(jnp.array([0, 1, 2]), 2)
>>> res
[0, 2, 10]
For more details on available methods, see the JAX documentation.
A new compiler driver has been implemented in C++. This improves compile-time performance by avoiding round-tripping, which is when the entire program being compiled is dumped to a textual form and re-parsed by another tool.
This is also a requirement for providing custom metadata at the LLVM level, which is necessary for better integration with tools like Enzyme. Finally, this makes it more natural to improve error messages originating from C++ when compared to the prior subprocess-based approach. (#216)
Support the braket.devices.Devices
enum class and s3_destination_folder
device options for AWS Braket remote devices. (#278)
Improvements have been made to the build process, including avoiding unnecessary processes such as removing opt
and downloading the wheel. (#298)
Remove a linker warning about duplicate rpath
s when Catalyst wheels are installed on macOS. (#314)
Fix incompatibilities with GCC on Linux introduced in v0.3.0 when compiling user programs. Due to these, Catalyst v0.3.0 only works when clang is installed in the user environment.
Remove undocumented package dependency on the zlib/zstd compression library. (#308)
Fix filesystem issue when compiling multiple functions with the same name and keep_intermediate=True
. (#306)
Add support for applying the adjoint
operation to QubitUnitary
gates. QubitUnitary
was not able to be adjoint
ed when the variable holding the unitary matrix might change. This can happen, for instance, inside of a for loop. To solve this issue, the unitary matrix gets stored in the array list via push and pops. The unitary matrix is later reconstructed from the array list and QubitUnitary
can be executed in the adjoint
ed context. (#304) (#310)
This release contains contributions from (in alphabetical order):
Ali Asadi,
David Ittah,
Erick Ochoa Lopez,
Jacob Mai Peng,
Sergei Mironov,
Romain Moyard.
Published by dime10 about 1 year ago
Catalyst now officially supports macOS ARM devices, such as Apple M1/M2 machines, with macOS binary wheels available on PyPI. For more details on the changes involved to support macOS, please see the improvements section. (#229) (#232) (#233) (#234)
Write Catalyst-compatible programs with native Python conditional statements. (#235)
AutoGraph is a new, experimental, feature that automatically converts Python conditional statements like if
, else
, and elif
, into their equivalent functional forms provided by Catalyst (such as catalyst.cond
).
This feature is currently opt-in, and requires setting the autograph=True
flag in the qjit
decorator:
dev = qml.device("lightning.qubit", wires=1)
@qjit(autograph=True)
@qml.qnode(dev)
def f(x):
if x < 0.5:
qml.RY(jnp.sin(x), wires=0)
else:
qml.RX(jnp.cos(x), wires=0)
return qml.expval(qml.PauliZ(0))
The implementation is based on the AutoGraph module from TensorFlow, and requires a working TensorFlow installation be available. In addition, Python loops (for
and while
) are not yet supported, and do not work in AutoGraph mode.
Note that there are some caveats when using this feature especially around the ues of global variables or object mutation inside of methods. A functional style is always recommended when using qjit
or AutoGraph.
The quantum adjoint operation can now be used in conjunction with Catalyst control flow, such as loops and conditionals. For this purpose a new instruction, catalyst.adjoint
, has been added. (#220)
catalyst.adjoint
can wrap around quantum functions which contain the Catalyst cond
, for_loop
, and while_loop
primitives. Previously, the usage of qml.adjoint
on functions with these primitives would result in decomposition errors. Note that a future release of Catalyst will
merge the behaviour of catalyst.adjoint
into qml.adjoint
for convenience.
dev = qml.device("lightning.qubit", wires=3)
@qjit
@qml.qnode(dev)
def circuit(x):
@for_loop(0, 3, 1)
def repeat_rx(i):
qml.RX(x / 2, wires=i)
adjoint(repeat_rx)()
return qml.expval(qml.PauliZ(0))
>>> circuit(0.2)
array(0.99500417)
Additionally, the ability to natively represent the adjoint construct in Catalyst's program representation (IR) was added.
QJIT-compiled programs now support (nested) container types as inputs and outputs of compiled functions. This includes lists and dictionaries, as well as any data structure implementing the PyTree protocol. (#215) (#221)
For example, a program that accepts and returns a mix of dictionaries, lists, and tuples:
@qjit
def workflow(params1, params2):
res1 = params1["a"][0][0] + params2[1]
return {"y1": jnp.sin(res1), "y2": jnp.cos(res1)}
>>> params1 = {"a": [[0.1], 0.2]}
>>> params2 = (0.6, 0.8)
>>> workflow(params1, params2)
array(0.78332691)
Compile-time backpropagation of arbitrary hybrid programs is now supported, via integration with Enzyme AD. (#158) (#193) (#224) (#225) (#239) (#244)
This allows catalyst.grad
to differentiate hybrid functions that contain both classical pre-processing (inside & outside of QNodes), QNodes, as well as classical post-processing (outside of QNodes) via a combination of backpropagation and quantum gradient methods.
The new default for the differentiation method
attribute in catalyst.grad
has been changed to "auto"
, which performs Enzyme-based reverse mode AD on classical code, in conjunction with the quantum diff_method
specified on each QNode:
dev = qml.device("lightning.qubit", wires=1)
@qml.qnode(dev, diff_method="parameter-shift")
def circuit(theta):
qml.RX(jnp.exp(theta ** 2) / jnp.cos(theta / 4), wires=0)
return qml.expval(qml.PauliZ(wires=0))
>>> grad = qjit(catalyst.grad(circuit, method="auto"))
>>> grad(jnp.pi)
array(0.05938718)
The reworked differentiation pipeline means you can now compute exact derivatives of programs with both classical pre- and post-processing, as shown below:
@qml.qnode(qml.device("lightning.qubit", wires=1), diff_method="adjoint")
def circuit(theta):
qml.RX(jnp.exp(theta ** 2) / jnp.cos(theta / 4), wires=0)
return qml.expval(qml.PauliZ(wires=0))
def loss(theta):
return jnp.pi / jnp.tanh(circuit(theta))
@qjit
def grad_loss(theta):
return catalyst.grad(loss)(theta)
>>> grad_loss(1.0)
array(-1.90958669)
You can also use multiple QNodes with different differentiation methods:
@qml.qnode(qml.device("lightning.qubit", wires=1), diff_method="parameter-shift")
def circuit_A(params):
qml.RX(jnp.exp(params[0] ** 2) / jnp.cos(params[1] / 4), wires=0)
return qml.probs()
@qml.qnode(qml.device("lightning.qubit", wires=1), diff_method="adjoint")
def circuit_B(params):
qml.RX(jnp.exp(params[1] ** 2) / jnp.cos(params[0] / 4), wires=0)
return qml.expval(qml.PauliZ(wires=0))
def loss(params):
return jnp.prod(circuit_A(params)) + circuit_B(params)
@qjit
def grad_loss(theta):
return catalyst.grad(loss)(theta)
>>> grad_loss(jnp.array([1.0, 2.0]))
array([ 0.57367285, 44.4911605 ])
And you can differentiate purely classical functions as well:
def square(x: float):
return x ** 2
@qjit
def dsquare(x: float):
return catalyst.grad(square)(x)
>>> dsquare(2.3)
array(4.6)
Note that the current implementation of reverse mode AD is restricted to 1st order derivatives, but you can still use catalyst.grad(method="fd")
is still available to perform a finite differences approximation of any differentiable function.
Add support for the new PennyLane arithmetic operators. (#250)
PennyLane is in the process of replacing Hamiltonian
and Tensor
observables with a set of general arithmetic operators. These consist of Prod, Sum and SProd.
By default, using dunder methods (eg. +
, -
, @
, *
) to combine operators with scalars or other operators will create Hamiltonian
and Tensor
objects. However, these two methods will be deprecated in coming releases of PennyLane.
To enable the new arithmetic operators, one can use Prod
, Sum
, and Sprod
directly or activate them by calling enable_new_opmath at the beginning of your PennyLane program.
dev = qml.device("lightning.qubit", wires=2)
@qjit
@qml.qnode(dev)
def circuit(x: float, y: float):
qml.RX(x, wires=0)
qml.RX(y, wires=1)
qml.CNOT(wires=[0, 1])
return qml.expval(0.2 * qml.PauliX(wires=0) - 0.4 * qml.PauliY(wires=1))
>>> qml.operation.enable_new_opmath()
>>> qml.operation.active_new_opmath()
True
>>> circuit(np.pi / 4, np.pi / 2)
array(0.28284271)
Better support for Hamiltonian observables:
Allow Hamiltonian observables with integer coefficients. (#248)
For example, compiling the following circuit wasn't previously allowed, but is now supported in Catalyst:
dev = qml.device("lightning.qubit", wires=2)
@qjit
@qml.qnode(dev)
def circuit(x: float, y: float):
qml.RX(x, wires=0)
qml.RY(y, wires=1)
coeffs = [1, 2]
obs = [qml.PauliZ(0), qml.PauliZ(1)]
return qml.expval(qml.Hamiltonian(coeffs, obs))
Allow nested Hamiltonian observables. (#255)
@qjit
@qml.qnode(qml.device("lightning.qubit", wires=3))
def circuit(x, y, coeffs1, coeffs2):
qml.RX(x, wires=0)
qml.RX(y, wires=1)
qml.RY(x + y, wires=2)
obs = [
qml.PauliX(0) @ qml.PauliZ(1),
qml.Hamiltonian(coeffs1, [qml.PauliZ(0) @ qml.Hadamard(2)]),
]
return qml.var(qml.Hamiltonian(coeffs2, obs))
Various performance improvements:
The execution and compile time of programs has been reduced, by generating more efficient code and avoiding unnecessary optimizations. Specifically, a scalarization procedure was added to the MLIR pass pipeline, and LLVM IR compilation is now invoked with optimization level 0. (#217)
The execution time of compiled functions has been improved in the frontend. (#213)
Specifically, the following changes have been made, which leads to a small but measurable improvement when using larger matrices as inputs, or functions with many inputs:
Peak memory utilization of a JIT compiled program has been reduced, by allowing tensors to be scheduled for deallocation. Previously, the tensors were not deallocated until the end of the call to the JIT compiled function. (#201)
Various improvements have been made to enable Catalyst to compile on macOS:
Remove unnecessary reinterpret_cast
from ObsManager
. Removal of these reinterpret_cast
allows compilation of the runtime to succeed in macOS. macOS uses an ILP32 mode for Aarch64 where they use the full 64 bit mode but with 32 bit Integer, Long, and Pointers. This patch also changes a test file to prevent a mismatch in machines which compile using ILP32 mode. (#229)
Allow runtime to be compiled on macOS. Substitute nproc
with a call to os.cpu_count()
and use correct flags for ld.64
. (#232)
Improve portability on the frontend to be available on macOS. Use .dylib
, remove unnecessary flags, and address behaviour difference in flags. (#233)
Small compatibility changes in order for all integration tests to succeed on macOS. (#234)
Dialects can compile with older versions of clang by avoiding type mismatches. (#228)
The runtime is now built against qir-stdlib
pre-build artifacts. (#236)
Small improvements have been made to the CI/CD, including fixing the Enzyme cache, generalize caches to other operating systems, fix build wheel recipe, and remove references to QIR in runtime's Makefile. (#243) (#247)
Support for Python 3.8 has been removed. (#231)
The default differentiation method on grad
and jacobian
is reverse-mode automatic differentiation instead of finite differences. When a QNode does not have a diff_method
specified, it will default to using the parameter shift method instead of finite-differences. (#244) (#271)
The JAX version used by Catalyst has been updated to v0.4.14
, the minimum PennyLane version required is now v0.32
. (#264)
Due to the change allowing Python container objects as inputs to QJIT-compiled functions, Python lists are no longer automatically converted to JAX arrays. (#231)
This means that indexing on lists when the index is not static will cause a TracerIntegerConversionError
, consistent with JAX's behaviour.
That is, the following example is no longer support:
@qjit
def f(x: list, index: int):
return x[index]
However, if the parameter x
above is a JAX or NumPy array, the compilation will continue to succeed.
The catalyst.grad
function has been renamed to catalyst.jacobian
and supports differentiation of functions that return multiple or non-scalar outputs. A new catalyst.grad
function has been added that enforces that it is differentiating a function with a single scalar return value. (#254)
Fixed an issue preventing the differentiation of qml.probs
with the parameter-shift method. (#211)
Fixed the incorrect return value data-type with functions returning qml.counts
. (#221)
Fix segmentation fault when differentiating a function where a quantum measurement is used multiple times by the same operation. (#242)
This release contains contributions from (in alphabetical order):
Ali Asadi, David Ittah, Erick Ochoa Lopez, Jacob Mai Peng, Romain Moyard, Sergei Mironov.
Published by josh146 over 1 year ago
When using OpenQASM-based devices the string representation of the circuit is printed on exception. #199
Use pybind11::module
interface library instead of pybind11::embed
in the runtime for OpenQasm backend to avoid linking to the python library at compile time. #200
This release contains contributions from (in alphabetical order):
Ali Asadi, David Ittah.
Published by josh146 over 1 year ago
Catalyst programs can now be used inside of a larger JAX workflow which uses JIT compilation, automatic differentiation, and other JAX transforms. #96 #123 #167 #192
For example, call a Catalyst qjit-compiled function from within a JAX jit-compiled function:
dev = qml.device("lightning.qubit", wires=1)
@qjit
@qml.qnode(dev)
def circuit(x):
qml.RX(jnp.pi * x[0], wires=0)
qml.RY(x[1] ** 2, wires=0)
qml.RX(x[1] * x[2], wires=0)
return qml.probs(wires=0)
@jax.jit
def cost_fn(weights):
x = jnp.sin(weights)
return jnp.sum(jnp.cos(circuit(x)) ** 2)
>>> cost_fn(jnp.array([0.1, 0.2, 0.3]))
Array(1.32269195, dtype=float64)
Catalyst-compiled functions can now also be automatically differentiated via JAX, both in forward and reverse mode to first-order,
>>> jax.grad(cost_fn)(jnp.array([0.1, 0.2, 0.3]))
Array([0.49249037, 0.05197949, 0.02991883], dtype=float64)
as well as vectorized using jax.vmap
:
>>> jax.vmap(cost_fn)(jnp.array([[0.1, 0.2, 0.3], [0.4, 0.5, 0.6]]))
Array([1.32269195, 1.53905377], dtype=float64)
In particular, this allows for a reduction in boilerplate when using JAX-compatible optimizers such as jaxopt
:
>>> opt = jaxopt.GradientDescent(cost_fn)
>>> params = jnp.array([0.1, 0.2, 0.3])
>>> (final_params, _) = jax.jit(opt.run)(params)
>>> final_params
Array([-0.00320799, 0.03475223, 0.29362844], dtype=float64)
Note that, in general, best performance will be seen when the Catalyst @qjit
decorator is used to JIT the entire hybrid workflow. However, there may be cases where you may want to delegate only the quantum part of your workflow to Catalyst, and let JAX handle classical components (for example, due to missing a feature or compatibility issue in Catalyst).
Support for Amazon Braket devices provided via the PennyLane-Braket plugin. #118 #139 #179 #180
This enables quantum subprograms within a JIT-compiled Catalyst workflow to execute on Braket simulator and hardware devices, including remote cloud-based simulators such as SV1.
def circuit(x, y):
qml.RX(y * x, wires=0)
qml.RX(x * 2, wires=1)
return qml.expval(qml.PauliY(0) @ qml.PauliZ(1))
@qjit
def workflow(x: float, y: float):
device = qml.device("braket.local.qubit", backend="braket_sv", wires=2)
g = qml.qnode(device)(circuit)
h = catalyst.grad(g)
return h(x, y)
workflow(1.0, 2.0)
For a list of available devices, please see the PennyLane-Braket documentation.
Internally, the quantum instructions are generating OpenQASM3 kernels at runtime; these are then executed on both local (braket.local.qubit
) and remote (braket.aws.qubit
) devices backed by Amazon Braket Python SDK, with measurement results then propagated back to the frontend.
Note that at initial release, not all Catalyst features are supported with Braket. In particular, dynamic circuit features, such as mid-circuit measurements, will not work with Braket devices.
Catalyst conditional functions defined via @catalyst.cond
now support an arbitrary number of 'else if' chains. #104
dev = qml.device("lightning.qubit", wires=1)
@qjit
@qml.qnode(dev)
def circuit(x):
@catalyst.cond(x > 2.7)
def cond_fn():
qml.RX(x, wires=0)
@cond_fn.else_if(x > 1.4)
def cond_elif():
qml.RY(x, wires=0)
@cond_fn.otherwise
def cond_else():
qml.RX(x ** 2, wires=0)
cond_fn()
return qml.probs(wires=0)
Iterating in reverse is now supported with constant negative step sizes via catalyst.for_loop
. #129
dev = qml.device("lightning.qubit", wires=1)
@qjit
@qml.qnode(dev)
def circuit(n):
@catalyst.for_loop(n, 0, -1)
def loop_fn(_):
qml.PauliX(0)
loop_fn()
return measure(0)
Additional gradient transforms for computing the vector-Jacobian product (VJP) and Jacobian-vector product (JVP) are now available in Catalyst. #98
Use catalyst.vjp
to compute the forward-pass value and VJP:
@qjit
def vjp(params, cotangent):
def f(x):
y = [jnp.sin(x[0]), x[1] ** 2, x[0] * x[1]]
return jnp.stack(y)
return catalyst.vjp(f, [params], [cotangent])
>>> x = jnp.array([0.1, 0.2])
>>> dy = jnp.array([-0.5, 0.1, 0.3])
>>> vjp(x, dy)
[array([0.09983342, 0.04 , 0.02 ]),
array([-0.43750208, 0.07000001])]
Use catalyst.jvp
to compute the forward-pass value and JVP:
@qjit
def jvp(params, tangent):
def f(x):
y = [jnp.sin(x[0]), x[1] ** 2, x[0] * x[1]]
return jnp.stack(y)
return catalyst.jvp(f, [params], [tangent])
>>> x = jnp.array([0.1, 0.2])
>>> tangent = jnp.array([0.3, 0.6])
>>> jvp(x, tangent)
[array([0.09983342, 0.04 , 0.02 ]),
array([0.29850125, 0.24000006, 0.12 ])]
Support for multiple backend devices within a single qjit-compiled function is now available. #89
For example, if you compile the Catalyst runtime with lightning.kokkos
support (via the compilation flag ENABLE_LIGHTNING_KOKKOS=ON
), you can use lightning.qubit
and lightning.kokkos
within a singular workflow:
dev1 = qml.device("lightning.qubit", wires=1)
dev2 = qml.device("lightning.kokkos", wires=1)
@qml.qnode(dev1)
def circuit1(x):
qml.RX(jnp.pi * x[0], wires=0)
qml.RY(x[1] ** 2, wires=0)
qml.RX(x[1] * x[2], wires=0)
return qml.var(qml.PauliZ(0))
@qml.qnode(dev2)
def circuit2(x):
@catalyst.cond(x > 2.7)
def cond_fn():
qml.RX(x, wires=0)
@cond_fn.otherwise
def cond_else():
qml.RX(x ** 2, wires=0)
cond_fn()
return qml.probs(wires=0)
@qjit
def cost(x):
return circuit2(circuit1(x))
>>> x = jnp.array([0.54, 0.31])
>>> cost(x)
array([0.80842369, 0.19157631])
Support for returning the variance of Hamiltonians, Hermitian matrices, and Tensors via qml.var
has been added. #124
dev = qml.device("lightning.qubit", wires=2)
@qjit
@qml.qnode(dev)
def circuit(x):
qml.RX(jnp.pi * x[0], wires=0)
qml.RY(x[1] ** 2, wires=1)
qml.CNOT(wires=[0, 1])
qml.RX(x[1] * x[2], wires=0)
return qml.var(qml.PauliZ(0) @ qml.PauliX(1))
>>> x = jnp.array([0.54, 0.31])
>>> circuit(x)
array(0.98851544)
The catalyst.grad
function now supports using the differentiation method defined on the QNode (via the diff_method
argument) rather than applying a global differentiation method. #163
As part of this change, the method
argument now accepts the following options:
method="defer"
: Quantum components of the hybrid function are differentiated according to the corresponding QNode diff_method
, while the classical computation is differentiated using traditional auto-diff.
With this strategy, Catalyst only currently supports QNodes with diff_method="param-shift" and
diff_method="adjoint"`.
method="fd"
: First-order finite-differences for the entire hybrid function. The diff_method
argument for each QNode is ignored.
This is an intermediate step towards differentiating functions that internally call multiple QNodes, and towards supporting differentiation of classical postprocessing.
Catalyst has been upgraded to work with JAX v0.4.13. #143 #185
Add a Backprop operation for using autodifferentiation (AD) at the LLVM level with Enzyme AD. The Backprop operations has a bufferization pattern and a lowering to LLVM. #107 #116
Error handling has been improved. The runtime now throws more descriptive and unified expressions for runtime errors and assertions. #92
In preparation for easier debugging, the compiler has been refactored to allow easy prototyping of new compilation pipelines. #38
In the future, this will allow the ability to generate MLIR or LLVM-IR by loading input from a string or file, rather than generating it from Python.
As part of this refactor, the following changes were made:
Passes are now classes. This allow developers/users looking to change flags to inherit from these passes and change the flags.
Passes are now passed as arguments to the compiler. Custom passes can just be passed to the compiler as an argument, as long as they implement a run method which takes an input and the output of this method can be fed to the next pass.
Improved Python compatibility by providing a stable signature for user generated functions. #106
Handle C++ exceptions without unwinding the whole stack. #99
Reduce the number of classical invocations by counting the number of gate parameters in the argmap
function. #136
Prior to this, the computation of hybrid gradients executed all of the classical code being differentiated in a pcount
function that solely counted the number of gate parameters in the quantum circuit. This was so argmap
and other downstream functions could allocate memrefs large enough to store all gate parameters.
Now, instead of counting the number of parameters separately, a dynamically-resizable array is used in the argmap
function directly to store the gate parameters. This removes one invocation of all of the classical code being differentiated.
Use Tablegen to define MLIR passes instead of C++ to reduce overhead of adding new passes. #157
Perform constant folding on wire indices for quantum.insert
and quantum.extract
ops, used when writing (resp. reading) qubits to (resp. from) quantum registers. #161
Represent known named observables as members of an MLIR Enum rather than a raw integer. This improves IR readability. #165
Fix a bug in the mapping from logical to concrete qubits for mid-circuit measurements. #80
Fix a bug in the way gradient result type is inferred. #84
Fix a memory regression and reduce memory footprint by removing unnecessary temporary buffers. #100
Provide a new abstraction to the QuantumDevice
interface in the runtime called MemRefView
. C++ implementations of the interface can iterate through and directly store results into the MemRefView
independant of the underlying memory layout. This can eliminate redundant buffer copies at the interface boundaries, which has been applied to existing devices. #109
Reduce memory utilization by transferring ownership of buffers from the runtime to Python instead of copying them. This includes adding a compiler pass that copies global buffers into the heap as global buffers cannot be transferred to Python. #112
Temporary fix of use-after-free and dependency of uninitialized memory. #121
Fixes file renaming within pass pipelines. #126
Fixes the issue with the do_queue
deprecation warnings in PennyLane. #146
Fixes the issue with gradients failing to work with hybrid functions that contain constant jnp.array
objects. This will enable PennyLane operators that have data in the form of a jnp.array
, such as a Hamiltonian, to be included in a qjit-compiled function. #152
An example of a newly supported workflow:
coeffs = jnp.array([0.1, 0.2])
terms = [qml.PauliX(0) @ qml.PauliZ(1), qml.PauliZ(0)]
H = qml.Hamiltonian(coeffs, terms)
@qjit
@qml.qnode(qml.device("lightning.qubit", wires=2))
def circuit(x):
qml.RX(x[0], wires=0)
qml.RY(x[1], wires=0)
qml.CNOT(wires=[0, 1])
return qml.expval(H)
params = jnp.array([0.3, 0.4])
jax.grad(circuit)(params)
This release contains contributions from (in alphabetical order):
Ali Asadi, David Ittah, Erick Ochoa Lopez, Jacob Mai Peng, Romain Moyard, Sergei Mironov.
Published by dime10 over 1 year ago
Add an option to print verbose messages explaining the compilation process
#68
Allow catalyst.grad
to be used on any traceable function (within a qjit context).
This means the operation is no longer resticted to acting on qml.qnode
s only.
#75
Work in progress on a Lightning-Kokkos backend:
Bring feature parity to the Lightning-Kokkos backend simulator.
#55
Add support for variance measurements for all observables.
#70
Build the runtime against qir-stdlib v0.1.0.
#58
Replace input-checking assertions with exceptions.
#67
Perform function inlining to improve optimizations and memory management within the compiler.
#72
Several fixes to address memory leaks in the compiled program:
Fix memory leaks from data that flows back into the Python environment.
#54
Fix memory leaks resulting from partial bufferization at the MLIR level. This fix makes the
necessary changes to reintroduce the -buffer-deallocation
pass into the MLIR pass pipeline.
The pass guarantees that all allocations contained within a function (that is allocations that are
not returned from a function) are also deallocated.
#61
Lift heap allocations for quantum op results from the runtime into the MLIR compiler core. This
allows all memref buffers to be memory managed in MLIR using the
MLIR bufferization infrastructure.
#63
Eliminate all memory leaks by tracking memory allocations at runtime. The memory allocations
which are still alive when the compiled function terminates, will be freed in the
finalization / teardown function.
#78
Fix returning complex scalars from the compiled function.
#77
This release contains contributions from (in alphabetical order):
Ali Asadi,
David Ittah,
Erick Ochoa Lopez,
Sergei Mironov.
Published by dime10 over 1 year ago
Fixes to codecov, build-lightning, and GH actions.
#34
Fixes the Catalyst image path in the readme to properly render on PyPI.
This release contains contributions from (in alphabetical order):
Ali Asadi,
Erick Ochoa Lopez.
Published by dime10 over 1 year ago
Initial public release.
This release contains contributions from (in alphabetical order):
Ali Asadi,
Sam Banning,
David Ittah,
Josh Izaac,
Erick Ochoa Lopez,
Sergei Mironov,
Isidor Schoch.