Merge branch 'dev' of https://github.com/mit-han-lab/torchquantum into dev

Author: Zhuoyang Ye

.github/workflows/functional_tests.yaml

@@ -14,7 +14,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        python-version: ["3.7", "3.8", "3.9", "3.10", "3.11"]
+        python-version: ["3.8", "3.9", "3.10", "3.11", "3.12"]
 
     steps:
     - uses: actions/checkout@v3
@@ -25,8 +25,8 @@ jobs:
     - name: Install dependencies
       run: |
         python -m pip install --upgrade pip
-        python -m pip install flake8 pytest qiskit-aer qiskit-ibmq-provider
         if [ -f requirements.txt ]; then pip install -r requirements.txt; fi
+        python -m pip install flake8 pytest qiskit-aer qiskit_ibm_runtime
     - name: Lint with flake8
       run: |
         # stop the build if there are Python syntax errors or undefined names
@@ -36,3 +36,34 @@ jobs:
     - name: Test with pytest
       run: |
         python -m pytest -m "not skip"
+    - name: Install TorchQuantum
+      run: |
+        pip install --editable .
+    - name: Test Examples
+      run: |
+        python3 examples/qubit_rotation/qubit_rotation.py --epochs 1
+        python3 examples/vqe/vqe.py --epochs 1 --steps_per_epoch 1
+        python3 examples/train_unitary_prep/train_unitary_prep.py --epochs 1
+        python3 examples/train_state_prep/train_state_prep.py --epochs 1
+        python3 examples/superdense_coding/superdense_coding_torchquantum.py
+        python3 examples/regression/run_regression.py --epochs 1
+        python3 examples/param_shift_onchip_training/param_shift.py
+        python3 examples/mnist/mnist_2qubit_4class.py --epochs 1
+        python3 examples/hadamard_grad/circ.py
+        python3 examples/encoder_examples/encoder_8x2ry.py
+        python3 examples/converter_tq_qiskit/convert.py
+        python3 examples/amplitude_encoding_mnist/mnist_new.py --epochs 1
+        python3 examples/amplitude_encoding_mnist/mnist_example.py --epochs 1
+        python3 examples/PauliSumOp/pauli_sum_op.py
+        python3 examples/regression/new_run_regression.py --epochs 1
+        python3 examples/quanvolution/quanvolution_trainable_quantum_layer.py --epochs 1
+        python3 examples/grover/grover_example_sudoku.py
+        python3 examples/param_shift_onchip_training/param_shift.py
+        python3 examples/quanvolution/quanvolution.py --epochs 1
+        python3 examples/quantum_lstm/qlstm.py --epochs 1
+        python3 examples/qaoa/max_cut_backprop.py --steps 1
+        python3 examples/optimal_control/optimal_control.py --epochs 1
+        python3 examples/optimal_control/optimal_control_gaussian.py --epochs 1
+        python3 examples/optimal_control/optimal_control_multi_qubit.py --epochs 1
+        python3 examples/save_load_example/save_load.py
+        python3 examples/mnist/mnist.py --epochs 1

README.md

@@ -2,7 +2,7 @@
 <img src="torchquantum_logo.jpg" alt="torchquantum Logo" width="450">
 </p>
 
-<h2><p align="center">A PyTorch Library for Quantum Simulation and Quantum Machine Learning</p></h2>
+<h2><p align="center">Quantum Computing in PyTorch</p></h2>
 <h3><p align="center">Faster, Scalable, Easy Debugging, Easy Deployment on Real Machine</p></h3>
 
 
@@ -19,9 +19,9 @@
     <a href="https://discord.gg/VTHZAB5E">
         <img alt="Chat @ Discord" src="https://img.shields.io/badge/contact-me-blue?logo=discord&logoColor=white">
     </a>
-    <a href="https://qmlsys.hanruiwang.me">
+<!--     <a href="https://qmlsys.hanruiwang.me">
         <img alt="Forum" src="https://img.shields.io/discourse/status?server=https%3A%2F%2Fqmlsys.hanruiwang.me%2F">
-    </a>
+    </a> -->
     <a href="https://qmlsys.mit.edu">
         <img alt="Website" src="https://img.shields.io/website?up_message=qmlsys&url=https%3A%2F%2Fqmlsys.mit.edu">
     </a>
@@ -49,7 +49,7 @@
 
 #### What it is doing
 
-Quantum simulation framework based on PyTorch. It supports statevector simulation and pulse simulation (coming soon) on GPUs. It can scale up to the simulation of 30+ qubits with multiple GPUs.
+Simulate quantum computations on classical hardware using PyTorch. It supports statevector simulation and pulse simulation on GPUs. It can scale up to the simulation of 30+ qubits with multiple GPUs.
 #### Who will benefit
 
 Researchers on quantum algorithm design, parameterized quantum circuit training, quantum optimal control, quantum machine learning, quantum neural networks.
@@ -58,10 +58,10 @@ Researchers on quantum algorithm design, parameterized quantum circuit training,
 Dynamic computation graph, automatic gradient computation, fast GPU support, batch model tersorized processing.
 
 ## News
-
-- v0.1.7 Available!
+- v0.1.8 Available!
+- Check the [dev branch](https://github.com/mit-han-lab/torchquantum/tree/dev) for new latest features on quantum layers and quantum algorithms.
 - Join our [Slack](https://join.slack.com/t/torchquantum/shared_invite/zt-1ghuf283a-OtP4mCPJREd~367VX~TaQQ) for real time support!
-- Welcome to contribute! Please contact us or post in the [forum](https://qmlsys.hanruiwang.me) if you want to have new examples implemented by TorchQuantum or any other questions.
+- Welcome to contribute! Please contact us or post in the Github Issues if you want to have new examples implemented by TorchQuantum or any other questions.
 - Qmlsys website goes online: [qmlsys.mit.edu](https://qmlsys.mit.edu) and [torchquantum.org](https://torchquantum.org)
 
 ## Features
@@ -358,6 +358,7 @@ pre-commit install
 - [ICCAD'22] [Wang et al., "QuEst: Graph Transformer for Quantum Circuit Reliability Estimation"](https://arxiv.org/abs/2210.16724)
 - [ICML Workshop] [Yun et al., "Slimmable Quantum Federated Learning"](https://dynn-icml2022.github.io/spapers/paper_7.pdf)
 - [IEEE ICDCS] [Yun et al., "Quantum Multi-Agent Reinforcement Learning via Variational Quantum Circuit Design"](https://ieeexplore.ieee.org/document/9912289)
+- [QCE'23] [Zhan et al., "Quantum Sensor Network Algorithms for Transmitter Localization"](https://ieeexplore.ieee.org/abstract/document/10313806)
 <details>
   <summary>Manuscripts</summary>
 

docs/requirements.txt

@@ -0,0 +1,7 @@
+furo @ git+https://github.com/frogcjn/torchquantum-doc-furo-theme.git
+nbsphinx
+recommonmark
+
+torchquantum>=0.1
+opt_einsum
+qiskit_ibm_runtime

docs/source/conf.py

@@ -29,7 +29,6 @@
 import sys
 import os
 sys.path.insert(0, os.path.abspath('../..'))
-import furo
 
 #import pathlib
 #sys.path.insert(0, pathlib.Path(__file__).parents[2].resolve().as_posix())
@@ -160,4 +159,4 @@
 display_gitlab = False
 show_source = True
 # -- Options for EPUB output
-epub_show_urls = 'footnote'
+epub_show_urls = 'footnote'

examples/grover/grover_example_sudoku.py

@@ -28,7 +28,7 @@
 """
 
 import torchquantum as tq
-from torchquantum.algorithms import Grover
+from torchquantum.algorithm import Grover
 
 
 # To simplify the process, we can compile this set of comparisons into a list of clauses for convenience.
@@ -90,4 +90,4 @@ def XOR(input0, input1, output):
 	print("b = ", key[1])
 	print("c = ", key[2])
 	print("d = ", key[3])
-	print("")
+	print("")

examples/mnist/mnist.py

@@ -179,6 +179,7 @@ def main():
         "--static", action="store_true", help="compute with " "static mode"
     )
     parser.add_argument("--pdb", action="store_true", help="debug with pdb")
+    parser.add_argument("--qiskit-simulation", action="store_true", help="run on a real quantum computer")
     parser.add_argument(
         "--wires-per-block", type=int, default=2, help="wires per block int static mode"
     )
@@ -243,38 +244,39 @@ def main():
     # test
     valid_test(dataflow, "test", model, device, qiskit=False)
 
-    # run on Qiskit simulator and real Quantum Computers
-    try:
-        from qiskit import IBMQ
-        from torchquantum.plugin import QiskitProcessor
-
-        # firstly perform simulate
-        print(f"\nTest with Qiskit Simulator")
-        processor_simulation = QiskitProcessor(use_real_qc=False)
-        model.set_qiskit_processor(processor_simulation)
-        valid_test(dataflow, "test", model, device, qiskit=True)
-
-        # then try to run on REAL QC
-        backend_name = "ibmq_lima"
-        print(f"\nTest on Real Quantum Computer {backend_name}")
-        # Please specify your own hub group and project if you have the
-        # IBMQ premium plan to access more machines.
-        processor_real_qc = QiskitProcessor(
-            use_real_qc=True,
-            backend_name=backend_name,
-            hub="ibm-q",
-            group="open",
-            project="main",
-        )
-        model.set_qiskit_processor(processor_real_qc)
-        valid_test(dataflow, "test", model, device, qiskit=True)
-    except ImportError:
-        print(
-            "Please install qiskit, create an IBM Q Experience Account and "
-            "save the account token according to the instruction at "
-            "'https://github.com/Qiskit/qiskit-ibmq-provider', "
-            "then try again."
-        )
+    if args.qiskit_simulation:
+        # run on Qiskit simulator and real Quantum Computers
+        try:
+            from qiskit import IBMQ
+            from torchquantum.plugin import QiskitProcessor
+
+            # firstly perform simulate
+            print(f"\nTest with Qiskit Simulator")
+            processor_simulation = QiskitProcessor(use_real_qc=False)
+            model.set_qiskit_processor(processor_simulation)
+            valid_test(dataflow, "test", model, device, qiskit=True)
+
+            # then try to run on REAL QC
+            backend_name = "ibmq_lima"
+            print(f"\nTest on Real Quantum Computer {backend_name}")
+            # Please specify your own hub group and project if you have the
+            # IBMQ premium plan to access more machines.
+            processor_real_qc = QiskitProcessor(
+                use_real_qc=True,
+                backend_name=backend_name,
+                hub="ibm-q",
+                group="open",
+                project="main",
+            )
+            model.set_qiskit_processor(processor_real_qc)
+            valid_test(dataflow, "test", model, device, qiskit=True)
+        except ImportError:
+            print(
+                "Please install qiskit, create an IBM Q Experience Account and "
+                "save the account token according to the instruction at "
+                "'https://github.com/Qiskit/qiskit-ibmq-provider', "
+                "then try again."
+            )
 
 
 if __name__ == "__main__":

examples/optimal_control/optimal_control.py

@@ -26,11 +26,22 @@
 import torch.optim as optim
 
 import torchquantum as tq
-import pdb
+import argparse
 import numpy as np
 
 if __name__ == "__main__":
-    pdb.set_trace()
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--pdb", action="store_true", help="debug with pdb")
+    parser.add_argument(
+        "--epochs", type=int, default=1000, help="number of training epochs"
+    )
+
+    args = parser.parse_args()
+
+    if args.pdb:
+        import pdb
+        pdb.set_trace()
+
     # target_unitary = torch.tensor([[0, 1], [1, 0]], dtype=torch.complex64)
     theta = 0.6
     target_unitary = torch.tensor(
@@ -41,11 +52,11 @@
         dtype=torch.complex64,
     )
 
-    pulse = tq.QuantumPulseDirect(n_steps=4, hamil=[[0, 1], [1, 0]])
+    pulse = tq.pulse.QuantumPulseDirect(n_steps=4, hamil=[[0, 1], [1, 0]])
 
     optimizer = optim.Adam(params=pulse.parameters(), lr=5e-3)
 
-    for k in range(1000):
+    for k in range(args.epochs):
         # loss = (abs(pulse.get_unitary() - target_unitary)**2).sum()
         loss = (
             1

examples/optimal_control/optimal_control_gaussian.py

@@ -26,11 +26,22 @@
 import torch.optim as optim
 
 import torchquantum as tq
-import pdb
+import argparse
 import numpy as np
 
 if __name__ == "__main__":
-    pdb.set_trace()
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--pdb", action="store_true", help="debug with pdb")
+    parser.add_argument(
+        "--epochs", type=int, default=1000, help="number of training epochs"
+    )
+
+    args = parser.parse_args()
+
+    if args.pdb:
+        import pdb
+        pdb.set_trace()
+
     # target_unitary = torch.tensor([[0, 1], [1, 0]], dtype=torch.complex64)
     theta = 1.1
     target_unitary = torch.tensor(
@@ -41,11 +52,11 @@
         dtype=torch.complex64,
     )
 
-    pulse = tq.QuantumPulseGaussian(hamil=[[0, 1], [1, 0]])
+    pulse = tq.pulse.QuantumPulseGaussian(hamil=[[0, 1], [1, 0]])
 
     optimizer = optim.Adam(params=pulse.parameters(), lr=5e-3)
 
-    for k in range(1000):
+    for k in range(args.epochs):
         # loss = (abs(pulse.get_unitary() - target_unitary)**2).sum()
         loss = (
             1

examples/optimal_control/optimal_control_multi_qubit.py

@@ -26,11 +26,22 @@
 import torch.optim as optim
 
 import torchquantum as tq
-import pdb
+import argparse
 import numpy as np
 
 if __name__ == "__main__":
-    pdb.set_trace()
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--pdb", action="store_true", help="debug with pdb")
+    parser.add_argument(
+        "--epochs", type=int, default=1000, help="number of training epochs"
+    )
+
+    args = parser.parse_args()
+
+    if args.pdb:
+        import pdb
+        pdb.set_trace()
+
     # target_unitary = torch.tensor([[0, 1], [1, 0]], dtype=torch.complex64)
     theta = 0.6
     target_unitary = torch.tensor(
@@ -43,9 +54,9 @@
         dtype=torch.complex64,
     )
 
-    pulse_q0 = tq.QuantumPulseDirect(n_steps=10, hamil=[[0, 1], [1, 0]])
-    pulse_q1 = tq.QuantumPulseDirect(n_steps=10, hamil=[[0, 1], [1, 0]])
-    pulse_q01 = tq.QuantumPulseDirect(
+    pulse_q0 = tq.pulse.QuantumPulseDirect(n_steps=10, hamil=[[0, 1], [1, 0]])
+    pulse_q1 = tq.pulse.QuantumPulseDirect(n_steps=10, hamil=[[0, 1], [1, 0]])
+    pulse_q01 = tq.pulse.QuantumPulseDirect(
         n_steps=10,
         hamil=[
             [1, 0, 0, 0],
@@ -62,7 +73,7 @@
         lr=5e-3,
     )
 
-    for k in range(1000):
+    for k in range(args.epochs):
         u_0 = pulse_q0.get_unitary()
         u_1 = pulse_q1.get_unitary()
         u_01 = pulse_q01.get_unitary()

examples/qaoa/max_cut_backprop.py

@@ -27,6 +27,7 @@
 
 import random
 import numpy as np
+import argparse
 
 from torchquantum.functional import mat_dict
 
@@ -172,6 +173,12 @@ def backprop_optimize(model, n_steps=100, lr=0.1):
 
 
 def main():
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "--steps", type=int, default=300, help="number of steps"
+    )
+    args = parser.parse_args()
+
     # create a input_graph
     input_graph = [(0, 1), (0, 3), (1, 2), (2, 3)]
     n_wires = 4
@@ -184,7 +191,7 @@ def main():
     # print("The circuit is", circ.draw(output="mpl"))
     # circ.draw(output="mpl")
     # use backprop
-    backprop_optimize(model, n_steps=300, lr=0.01)
+    backprop_optimize(model, n_steps=args.steps, lr=0.01)
     # use parameter shift rule
     # param_shift_optimize(model, n_steps=500, step_size=100000)