@@ -27,7 +27,7 @@ of total circuit run-time. Circuits that exceed this limit will return a
27
27
28
28
### Moment structure
29
29
30
- The hardware will attempt to run your circuit as it exists in cirq to the
30
+ The hardware will attempt to run your circuit as it exists in Cirq to the
31
31
extent possible. The device will respect the moment structure of your circuit
32
32
and will execute successive moments in a serial fashion.
33
33
@@ -65,11 +65,40 @@ cirq.Circuit(
65
65
66
66
The duration of a moment is the time of its longest gate. For example,
67
67
if a moment has gates of duration 12ns, 25ns, and 32ns, the entire moment
68
- will take 32ns. Qubits executing the shorter gtes will idle during the rest
68
+ will take 32ns. Qubits executing the shorter gates will idle during the rest
69
69
of the time. To minimize the duration of the circuit, it is best to align
70
70
gates of the same duration together when possible. See the
71
71
[ best practices] ( ./best_practices.ipynb ) for more details.
72
72
73
+ ## Device Parameter Sweeps
74
+
75
+ Certain device parameters can be changed for the duration of
76
+ a circuit in order to support hardware parameter sweeps. For instance,
77
+ frequencies, amplitudes, and various other parameters can be modified
78
+ in order to find optimal values or explore the parameter space.
79
+
80
+ These parameter names are generally not public, so you will need to
81
+ work with a Google sponsor or resident in order to access the proper
82
+ key names. These parameters are specified as lists of strings representing
83
+ a path from the device config's folder (or the "sample folder").
84
+
85
+ These keys can be swept like any other symbol using the
86
+ ` cirq_google.study.DeviceParameter ` variable. For instance, the
87
+ following code will sweep qubit (4,8)'s pi amplitude from 0.0 to 1.0
88
+ in 0.02 increments.
89
+
90
+
91
+ ```
92
+ descriptor = cirq_google.study.DeviceParameter( ["q4_8", "piAmp"])
93
+ sweep = cirq.Linspace("q4_8.piAmp", 0, 1, 51, metadata=descriptor)
94
+ ```
95
+
96
+ Any ` DeviceParameter ` keys that are set to a single value using a ` cirq.Points `
97
+ object will change that value for all circuits run.
98
+
99
+ If units are required, they should be specified as a string (such as 'MHz')
100
+ using the ` units ` argument of the ` DeviceParameter ` .
101
+
73
102
## Gates supported
74
103
75
104
The following lists the gates supported by Google devices.
188
217
This gate has a duration of 32ns and can be used in
189
218
` cirq_google.SQRT_ISWAP_GATESET ` or in the ` cirq_google.FSIM_GATESET ` .
190
219
191
- This gate is implemented by using an entangling gate surrounding by
220
+ This gate is implemented by using an entangling gate surrounded by
192
221
Z gates. The preceding Z gates are physical Z gates and will absorb
193
222
any phases that have accumulated through the use of Virtual Z gates.
194
223
Following the entangler are virtual Z gates to match phases back. All
@@ -238,7 +267,8 @@ expressions, but only a subset of Sympy expression types are supported:
238
267
` sympy.Symbol ` , ` sympy.Add ` , ` sympy.Mul ` , and ` sympy.Pow ` .
239
268
240
269
## Specific Device Layouts
241
- The following devices are provided as part of cirq and can help you get your
270
+
271
+ The following devices are provided as part of Cirq and can help you get your
242
272
circuit ready for running on hardware by verifying that you are using
243
273
appropriate qubits.
244
274
@@ -272,7 +302,7 @@ It can be accessed using `cirq.GridQubit(row, col)` using grid coordinates speci
272
302
9 ----I-----
273
303
```
274
304
275
- It can be accessing by using ` cirq_google.Sycamore ` . This device has two possible
305
+ It can be accessed by using ` cirq_google.Sycamore ` . This device has two possible
276
306
two-qubits gates that can be used.
277
307
278
308
* Square root of ISWAP. The gate ` cirq.ISWAP ** 0.5 ` or ` cirq.ISWAP ** -0.5 ` can be
@@ -304,32 +334,3 @@ with and presents less hardware-related complications than using the full Sycamo
304
334
305
335
This grid can be accessed using ` cirq_google.Sycamore23 ` and uses the same gate sets and
306
336
compilation as the Sycamore device.
307
-
308
-
309
- ### Bristlecone
310
-
311
- The Bristlecone processor is a 72 qubit device
312
- [ announced by Google in 2018] ( https://ai.googleblog.com/2018/03/a-preview-of-bristlecone-googles-new.html ) .
313
-
314
- The device is arrayed on a grid in a diamond pattern like this.
315
-
316
- ```
317
- 11
318
- 012345678901
319
- 0 -----AB-----
320
- 1 ----ABCD----
321
- 2 ---ABCDEF---
322
- 3 --ABCDEFGH--
323
- 4 -ABCDEFGHIJ-
324
- 5 ABCDEFGHIJKL
325
- 6 -CDEFGHIJKL-
326
- 7 --EFGHIJKL--
327
- 8 ---GHIJKL---
328
- 9 ----IJKL----
329
- 10-----KL-----
330
- ```
331
-
332
- It can be accessing by using ` cirq_google.Bristlecone ` . Circuits can be compiled to it by using
333
- ` cirq_google.optimized_for_xmon ` or by using ` cirq_google.optimized_for_sycamore ` with
334
- optimizer_type ` xmon ` .
335
-
0 commit comments