<br />

|---|
| [![](https://www.tensorflow.org/images/GitHub-Mark-32px.png) View source on GitHub](https://github.com/tensorflow/quantum/blob/v0.7.6/tensorflow_quantum/python/differentiators/adjoint.py#L22-L110) |

Differentiate a circuit with respect to its inputs by adjoint method.

Inherits From: [`Differentiator`](https://www.tensorflow.org/quantum/api_docs/python/tfq/differentiators/Differentiator)

> [!CAUTION]
> **Caution:** This differentiator is only compatible with analytic expectation calculations and the native C++ ops (`backend = None`). The methods used by this differentiation techniques can not be realized easily on a real device.

The Adjoint differentiator follows along with the methods described here:
[arXiv:1912.10877](https://arxiv.org/abs/1912.10877) and
[doi: 10.1111/j.1365-246X.2006.02978.x](https://academic.oup.com/gji/article-pdf/167/2/495/1492368/167-2-495.pdf).
The Adjoint method differentiates the input circuits in roughly one forward
and backward pass over the circuits, to calculate the gradient of
a symbol only a constant number of gate operations need to be applied to the
circuits state. When the number of parameters in a circuit is very large,
this differentiator performs much better than all the others found in TFQ.

    my_op = tfq.get_expectation_op()
    adjoint_differentiator = tfq.differentiators.Adjoint()
    # Get an expectation op, with this differentiator attached.
    op = adjoint_differentiator.generate_differentiable_op(
        analytic_op=my_op
    )
    qubit = cirq.GridQubit(0, 0)
    circuit = tfq.convert_to_tensor([
        cirq.Circuit(cirq.X(qubit) ** sympy.Symbol('alpha'))
    ])
    psums = tfq.convert_to_tensor([[cirq.Z(qubit)]])
    symbol_values = np.array([[0.123]], dtype=np.float32)
    # Calculate tfq gradient.
    symbol_values_t = tf.convert_to_tensor(symbol_values)
    symbol_names = tf.convert_to_tensor(['alpha'])
    with tf.GradientTape() as g:
        g.watch(symbol_values_t)
        expectations = op(circuit, symbol_names, symbol_values_t, psums
    )
    grads = g.gradient(expectations, symbol_values_t)
    grads
    tf.Tensor([[-1.1839]], shape=(1, 1), dtype=float32)

## Methods

### `differentiate_analytic`

[View source](https://github.com/tensorflow/quantum/blob/v0.7.6/tensorflow_quantum/python/differentiators/adjoint.py#L98-L103)

    @tf.function
    differentiate_analytic(
        programs, symbol_names, symbol_values, pauli_sums, forward_pass_vals, grad
    )

### `differentiate_sampled`

[View source](https://github.com/tensorflow/quantum/blob/v0.7.6/tensorflow_quantum/python/differentiators/adjoint.py#L105-L110)

    differentiate_sampled(
        programs,
        symbol_names,
        symbol_values,
        pauli_sums,
        num_samples,
        forward_pass_vals,
        grad
    )

Differentiate a circuit with sampled expectation.

This is called at graph runtime by TensorFlow. `differentiate_sampled`
calls he inheriting differentiator's `get_gradient_circuits` and uses
those components to construct the gradient.

> [!NOTE]
> **Note:** the default implementation does not use `forward_pass_vals`; the inheriting differentiator is free to override the default implementation and use this argument if desired.

<br />

<br />

<br />

| Args ||
|---|---|
| `programs` | [`tf.Tensor`](https://www.tensorflow.org/api_docs/python/tf/Tensor) of strings with shape \[batch_size\] containing the string representations of the circuits to be executed. |
| `symbol_names` | [`tf.Tensor`](https://www.tensorflow.org/api_docs/python/tf/Tensor) of strings with shape \[n_params\], which is used to specify the order in which the values in `symbol_values` should be placed inside of the circuits in `programs`. |
| `symbol_values` | [`tf.Tensor`](https://www.tensorflow.org/api_docs/python/tf/Tensor) of real numbers with shape \[batch_size, n_params\] specifying parameter values to resolve into the circuits specified by programs, following the ordering dictated by `symbol_names`. |
| `pauli_sums` | [`tf.Tensor`](https://www.tensorflow.org/api_docs/python/tf/Tensor) of strings with shape \[batch_size, n_ops\] containing the string representation of the operators that will be used on all of the circuits in the expectation calculations. |
| `num_samples` | [`tf.Tensor`](https://www.tensorflow.org/api_docs/python/tf/Tensor) of positive integers representing the number of samples per term in each term of pauli_sums used during the forward pass. |
| `forward_pass_vals` | [`tf.Tensor`](https://www.tensorflow.org/api_docs/python/tf/Tensor) of real numbers with shape \[batch_size, n_ops\] containing the output of the forward pass through the op you are differentiating. |
| `grad` | [`tf.Tensor`](https://www.tensorflow.org/api_docs/python/tf/Tensor) of real numbers with shape \[batch_size, n_ops\] representing the gradient backpropagated to the output of the op you are differentiating through. |

<br />

<br />

<br />

<br />

| Returns ||
|---|---|
| A [`tf.Tensor`](https://www.tensorflow.org/api_docs/python/tf/Tensor) with the same shape as `symbol_values` representing the gradient backpropageted to the `symbol_values` input of the op you are differentiating through. ||

<br />

### `generate_differentiable_op`

[View source](https://github.com/tensorflow/quantum/blob/v0.7.6/tensorflow_quantum/python/differentiators/adjoint.py#L65-L89)

    generate_differentiable_op(
        *, sampled_op=None, analytic_op=None
    )

Generate a differentiable op by attaching self to an op.

See [`tfq.differentiators.Differentiator`](https://www.tensorflow.org/quantum/api_docs/python/tfq/differentiators/Differentiator). This has been partially
re-implemented by the Adjoint differentiator to disallow the
`sampled_op` input.

<br />

<br />

<br />

| Args ||
|---|---|
| `sampled_op` | A `callable` op that you want to make differentiable using this differentiator's `differentiate_sampled` method. |
| `analytic_op` | A `callable` op that you want to make differentiable using this differentiators `differentiate_analytic` method. |

<br />

<br />

<br />

<br />

| Returns ||
|---|---|
| A `callable` op that who's gradients are now registered to be a call to this differentiators `differentiate_*` function. ||

<br />

### `get_gradient_circuits`

[View source](https://github.com/tensorflow/quantum/blob/v0.7.6/tensorflow_quantum/python/differentiators/adjoint.py#L91-L96)

    @tf.function
    get_gradient_circuits(
        programs, symbol_names, symbol_values
    )

See base class description.

### `refresh`

[View source](https://github.com/tensorflow/quantum/blob/v0.7.6/tensorflow_quantum/python/differentiators/differentiator.py#L203-L209)

    refresh()

Refresh this differentiator in order to use it with other ops.