Attempting to learn how to call q# from Python code and have it run for real on the IONQ QPU as it does (or appears to do) using VS and >dotnet run of the raw q# code. I followed the guides and workshop.
Python code:
import qsharp
import qsharp.azure
qsharp.projects.add("****path to *******/TestIONQ.csproj")
from TestIONQ import GetRandomResult
print(f"Simulated Result: {GetRandomResult.simulate()}")
print("------------------------------------------------")
qsharp.azure.connect(
subscription = "****************************",
resourceGroup = "**************",
workspace = "************",
location = "******* US")
qsharp.azure.target("ionq.qpu")
result = qsharp.azure.execute(GetRandomResult, jobName="Generate random bit")
print(f" Final result from IONQ - QPU: {result}")
q# code:
namespace TestIONQ {
open Microsoft.Quantum.Canon;
open Microsoft.Quantum.Intrinsic;
//@EntryPoint()
operation GetRandomResult() : Result {
use q = Qubit();
H(q);
return M(q);
}
}
and the .csproj file:
<Project Sdk="Microsoft.Quantum.Sdk/0.16.2104138035">
<PropertyGroup>
<OutputType>Exe</OutputType>
<TargetFramework>netcoreapp3.1</TargetFramework>
<ExecutionTarget>ionq.qpu</ExecutionTarget>
<IQSharpLoadAutomatically>true</IQSharpLoadAutomatically>
</PropertyGroup>
</Project>
The results of running the above Python code in Anaconda qsharp-env environment (Python 3.7.10) are as follows:
Simulated Result: 0
------------------------------------------------
Connected to Azure Quantum workspace ####### in location #####us.
Loading package Microsoft.Quantum.Providers.IonQ and dependencies...
Active target is now ionq.qpu
Submitting TestIONQ.GetRandomResult to target ionq.qpu...
Failed to submit Q# operation TestIONQ.GetRandomResult for execution.
Could not load file or assembly 'System.Text.Json, Version=5.0.0.0, Culture=neutral, PublicKeyToken=####token#####'. The system cannot find the file specified.
Obviously, no problem connecting to Azure and the Workspace. In fact I can run the container-ship optimization example no problem from Python. This also works fine for the first half of the Python code when .simulate() is invoked.
Next, when I try to bypass the IONQ QPU and use its own simulator by changing this one line:
qsharp.azure.target("ionq.simulator")
The resulting error is the same and the results are as follows:
Simulated Result: 1
------------------------------------------------
Connected to Azure Quantum workspace ######## in location #######.
Loading package Microsoft.Quantum.Providers.IonQ and dependencies...
Active target is now ionq.simulator
Submitting TestIONQ.GetRandomResult to target ionq.simulator...
Failed to submit Q# operation TestIONQ.GetRandomResult for execution.
Could not load file or assembly 'System.Text.Json, Version=5.0.0.0, Culture=neutral, PublicKeyToken='....token......'. The system cannot find the file specified.
Traceback (most recent call last):
File "ionq_sim_remote.py", line 18, in <module>
result = qsharp.azure.execute(GetRandomResult, jobName="Generate random bit")
File "F:\Python38\miniconda\envs\qsharp-env\lib\site-packages\qsharp\azure.py", line 137, in execute
if "error_code" in result: raise AzureError(result)
qsharp.azure.AzureError: {'error_code': 1010, 'error_name': 'JobSubmissionFailed', 'error_description': 'Failed to submit the job to the Azure Quantum workspace.'}
This runs very easily on Azure using the q# code snippet within Visual Studio at the command line using a variant of what was shown during the workshop
az quantum execute --target-id ionq.qpu --job-name IONQ_test --resource-group ***rg name*** --workspace-name ***ws name*** --location **** -o table
and indeed this appears to have run on the actual QPU hardware as compared to the simulator (which gives the exact 0.5/0.5 result).
Result Frequency
-------- ----------
0 0.49800000
1 0.50200000
But then calling that same q# code from Python - including the same .csproj file seems to throw this JSON file error - even with the qsharp-env loaded into Anaconda. I apologize if it is something silly that I have done- trying to learn here.
By the way, this works great as a way around the problem with no Anaconda environment required or anything special:
Python:
import os
os.system(f'powershell.exe az quantum execute --target-id ionq.qpu --job-name Pytest --resource-group **** --workspace-name **** --location **** -o table ')
And the result was definitely run on the actual hardware (took a good while):
Result Frequency
-------- ----------- -------------------------
0 0.53200000 ▐███████████ |
1 0.46800000 ▐█████████ |
@Joab.Ai, thank you for posting this issue! We've identified this to be specific to the latest version of qsharp (0.16.2104.138035).
While we are looking into a fix, a workaround will be to downgrade your :qsharp package version
Edit: we have fixed this issue in our latest release! Update to the latest version with this command:
conda install -c quantum-engineering qsharp=0.16.2105.140472
or simply run:
conda update -c quantum-engineering qsharp