Merge genQC v0.2.4

MANIFEST.in

@@ -1,5 +1,5 @@
 include settings.ini
-include LICENSE
+include LICENSE.txt
 include CONTRIBUTING.md
 include README.md
 recursive-exclude * __pycache__

README.md

@@ -23,7 +23,9 @@ QFT.](https://github.com/FlorianFuerrutter/genQC/blob/main/src/webpage/assets/qf
 
 ## 📰 News
 
-- 🔥 \[2025-06-02\] Paper release: [Synthesis of discrete-continuous quantum circuits with multimodal diffusion models](https://www.arxiv.org/abs/2506.01666).
+- 🔥 \[2025-06-02\] Paper release: [Synthesis of discrete-continuous
+  quantum circuits with multimodal diffusion
+  models](https://www.arxiv.org/abs/2506.01666).
 - 🔥 \[2025-06-01\] *Discrete-continuous circuits with multimodal
   diffusion* - model released on [Hugging Face:
   huggingface.co/collections/Floki00](https://huggingface.co/collections/Floki00/discrete-continuous-circuits-with-multimodal-diffusion-6839c4e4553e56b957bbd5bf).
@@ -67,10 +69,9 @@ from genQC.benchmark.bench_compilation import SpecialUnitaries
 from genQC.platform.simulation import Simulator, CircuitBackendType
 
 device = infer_torch_device()
-set_seed(0)
 
 pipeline = MultimodalDiffusionPipeline_ParametrizedCompilation.from_pretrained(
-                                repo_id="Floki00/cirdit_multimodal_compile_3to5qubit", 
+                                repo_id="Floki00/cirdit_multimodal_compile_3to5qubit_v1.1", 
                                 device=device)
 
 pipeline.scheduler.set_timesteps(40) 
@@ -91,20 +92,44 @@ out_tensor, params = generate_compilation_tensors(pipeline,
 ```
 
 ``` python
-vocabulary = {g:i+1 for i, g in enumerate(pipeline.gate_pool)} 
+vocabulary = {g:i+1 for i, g in enumerate(pipeline.gate_pool)}
 tokenizer  = CircuitTokenizer(vocabulary)
-simulator  = Simulator(CircuitBackendType.QISKIT)
+simulator  = Simulator(CircuitBackendType.CUDAQ)
 
 qc_list, _ = decode_tensors_to_backend(simulator, tokenizer, out_tensor, params)
-qc_list[0].draw("mpl")
+
+simulator.backend.draw(qc_list[0], num_qubits=4)
 ```
 
-![](https://github.com/FlorianFuerrutter/genQC/blob/main/get_started_files/figure-commonmark/cell-3-output-1.png?raw=true)
+                                                                            »
+    q0 : ────────────────────────●────────────────────●───────────●───────╳─»
+                                 │         ╭───╮      │      ╭────┴─────╮ │ »
+    q1 : ────────────────────────┼───────╳─┤ h ├──────┼──────┤ r1(1.25) ├─┼─»
+                                 │       │ ╰───╯╭─────┴─────╮╰──────────╯ │ »
+    q2 : ───────────●────────────┼───────╳──────┤ r1(6.253) ├─────────────┼─»
+         ╭───╮╭─────┴─────╮╭─────┴─────╮        ╰───────────╯             │ »
+    q3 : ┤ h ├┤ r1(1.571) ├┤ r1(7.191) ├──────────────────────────────────╳─»
+         ╰───╯╰───────────╯╰───────────╯                                    »
+
+    ################################################################################
+
+
+    ─────────────────
+
+    ─────●───────────
+    ╭────┴─────╮╭───╮
+    ┤ r1(1.59) ├┤ h ├
+    ╰──┬───┬───╯╰───╯
+    ───┤ h ├─────────
+       ╰───╯         
 
 #### Further examples
 
-More detailed examples and tutorial notebooks are provided on the
-project page
+A detailed tutorial on the application with `CUDA-Q` is available on the
+[CUDA-Q
+documentation](https://nvidia.github.io/cuda-quantum/latest/applications/python/unitary_compilation_diffusion_models.html).
+
+More examples and tutorial notebooks are provided on the project page
 [\[tutorials\]](https://florianfuerrutter.github.io/genQC/examples/tutorials.html)
 or in the directory `src/examples/`.
 

genQC/__init__.py

@@ -1 +1 @@
-__version__ = "0.2.2"
+__version__ = "0.2.4"

genQC/pipeline/multimodal_diffusion_pipeline.py

@@ -400,7 +400,8 @@ def train_step(self, data, train, **kwargs):
         mse_loss_weight_h = (1.0 - alphas_cumprod_h) * F.sigmoid(SNR_h.log())
 
         SNR_w = alphas_cumprod_w / (1.0-alphas_cumprod_w+1e-8) + 1e-8
-
+        mse_loss_weight_w = (1.0 - alphas_cumprod_w) * F.sigmoid(SNR_w.log() + np.log(3.0 * np.pi**2))
+
         #comp mse
         mse_flat = lambda out, target: (out-target).square().mean(dim=list(range(1, len(out.shape))))
         loss_h = mse_flat(model_output[..., :self.embedder.clr_dim], pred_target_h.detach()) * mse_loss_weight_h.squeeze().detach()

genQC/pipeline/pipeline.py

@@ -75,9 +75,9 @@ def store_pipeline(self, config_path: str, save_path: str):
     def from_config_file(config_path, device: torch.device, save_path: str=None): return None 
 
     @classmethod
-    def from_pretrained(cls, repo_id: str, device: torch.device, use_auth_token: bool = False, **kwargs):  
+    def from_pretrained(cls, repo_id: str, device: torch.device, **kwargs):  
         """Load a model pipeline directly from Huggingface."""
-        model_path = snapshot_download(repo_id=repo_id, repo_type="model", allow_patterns=["*.pt", "*.yaml", "*.safetensors"], use_auth_token=use_auth_token, **kwargs) 
+        model_path = snapshot_download(repo_id=repo_id, repo_type="model", allow_patterns=["*.pt", "*.yaml", "*.safetensors"], **kwargs) 
         pipeline   = cls.from_config_file(model_path+"/", device)  
         return pipeline
 

genQC/platform/backends/circuits_cudaq.py

@@ -193,8 +193,15 @@ def get_unitary(self, parametrizedCudaqKernel: ParametrizedCudaqKernel, num_qubi
 
         return U
 
-    def draw(self, kernel: cudaq.kernel, num_qubits: int, **kwargs) -> None:
+    def draw(self, parametrizedCudaqKernel: ParametrizedCudaqKernel, num_qubits: int, return_str: bool = False, **kwargs) -> None:
         """Draw the given `cudaq.kernel` using cudaq.""" 
+
+        kernel, thetas = parametrizedCudaqKernel.kernel, parametrizedCudaqKernel.params
+
         c    = [0] * (2**num_qubits)
         c[0] = 1
-        print(cudaq.draw(kernel, c))
+
+        s = cudaq.draw(kernel, c, thetas)
+        if return_str:
+            return s
+        print(s)

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name="genQC"
-requires-python=">=3.12"
+requires-python=">=3.10"
 dynamic = [ "keywords", "description", "version", "dependencies", "optional-dependencies", "readme", "license", "authors", "classifiers", "entry-points", "scripts", "urls"]
 
 [tool.uv]

settings.ini

@@ -3,8 +3,8 @@
 ### Python library ###
 repo 			 = genQC
 lib_name 		 = %(repo)s
-version 		 = 0.2.2
-min_python       = 3.12
+version 		 = 0.2.4
+min_python       = 3.10
 license          = apache2   
 black_formatting = False
 

setup.py

@@ -2,9 +2,9 @@
 from configparser import ConfigParser
 
 import setuptools
-from pkg_resources import parse_version
 
-assert parse_version(setuptools.__version__)>=parse_version('36.2')
+#from pkg_resources import parse_version
+#assert parse_version(setuptools.__version__)>=parse_version('36.2')
 
 # note: all settings are in settings.ini; edit there, not here
 config = ConfigParser(delimiters=['='])
@@ -26,7 +26,7 @@
 }
 statuses = [ '0 - Pre-Planning', '1 - Planning', '2 - Pre-Alpha', '3 - Alpha',
     '4 - Beta', '5 - Production/Stable', '6 - Mature', '7 - Inactive' ]
-py_versions = '3.7 3.8 3.9 3.10 3.12 3.13'.split()
+py_versions = '3.10 3.11 3.12 3.13'.split()
 
 requirements = ['packaging']
 requirements += shlex.split(cfg.get('requirements', ''))

src/_quarto.yml

@@ -25,6 +25,7 @@ website:
   favicon: "webpage/assets/logo.png"
   open-graph: true
   repo-actions: [issue, source]
+  repo-subdir: "src"
   back-to-top-navigation: true
   page-navigation: true
 

src/examples/Quantum circuit synthesis with diffusion models/1_editing_and_masking.ipynb

@@ -52,9 +52,7 @@
     "from genQC.pipeline.diffusion_pipeline import DiffusionPipeline\n",
     "from genQC.inference.infer_srv import convert_tensors_to_srvs, schmidt_rank_vector\n",
     "import genQC.platform.qcircuit_dataset_construction as data_const\n",
-    "from genQC.platform.simulation.qcircuit_sim import instruction_name_to_qiskit_gate\n",
-    "# import genQC.util as util\n",
-    "# from qiskit.quantum_info import DensityMatrix"
+    "from genQC.platform.simulation.qcircuit_sim import instruction_name_to_qiskit_gate"
    ]
   },
   {
@@ -97,31 +95,7 @@
    "execution_count": null,
    "id": "e5d60c23-9514-4432-bc82-622c088fced6",
    "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "application/vnd.jupyter.widget-view+json": {
-       "model_id": "a1c55308f7c94327b24b9df417f05117",
-       "version_major": 2,
-       "version_minor": 0
-      },
-      "text/plain": [
-       "Fetching 2 files:   0%|          | 0/2 [00:00<?, ?it/s]"
-      ]
-     },
-     "metadata": {},
-     "output_type": "display_data"
-    },
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "[INFO]: `genQC.models.unet_qc.QC_Cond_UNet` instantiated from given config on cuda.\n",
-      "[INFO]: `genQC.models.frozen_open_clip.CachedFrozenOpenCLIPEmbedder` instantiated from given config on cuda.\n",
-      "[INFO]: `genQC.models.frozen_open_clip.CachedFrozenOpenCLIPEmbedder`. No save_path` provided. No state dict loaded.\n"
-     ]
-    }
-   ],
+   "outputs": [],
    "source": [
     "pipeline = DiffusionPipeline.from_pretrained(\"Floki00/qc_srv_3to8qubit\", device)"
    ]

src/examples/tutorials.qmd

@@ -16,7 +16,7 @@ listing:
 
 #### Welcome to the `genQC` Tutorials. 
 
-Here you can familiarize yourself with basics concepts and modules. Tutorials show how to reproduce some figures and results of the corresponding [research papers](../webpage/research.qmd).
+Here you can familiarize yourself with basics concepts and modules. The tutorials show how to reproduce some figures and results of the corresponding [research papers](../webpage/research.qmd).
 
 <br>
 

src/index.qmd

@@ -89,7 +89,7 @@ from genQC.platform.simulation import Simulator, CircuitBackendType
 device = infer_torch_device()
 
 pipeline = MultimodalDiffusionPipeline_ParametrizedCompilation.from_pretrained( # <1>
-                            repo_id="Floki00/cirdit_multimodal_compile_3to5qubit", # <1> 
+                            repo_id="Floki00/cirdit_multimodal_compile_3to5qubit_v1.1", # <1> 
                             device=device)  # <1>
 
 pipeline.scheduler.set_timesteps(40)   # <2>
@@ -112,17 +112,18 @@ out_tensor, params = generate_compilation_tensors(pipeline, # <4>
 
 vocabulary = {g:i+1 for i, g in enumerate(pipeline.gate_pool)} # <5>
 tokenizer  = CircuitTokenizer(vocabulary)                      # <5>
-simulator  = Simulator(CircuitBackendType.QISKIT)              # <5>
+simulator  = Simulator(CircuitBackendType.CUDAQ)               # <5>
 
 qc_list, _ = decode_tensors_to_backend(simulator, tokenizer, out_tensor, params) # <5>
 
-qc_list[0].draw("mpl")
+simulator.backend.draw(qc_list[0], num_qubits=4) # <6>
 ```
 1. Load a pre-trained Diffusion model. 
 2. Set inference parameters.
 3. Specify QFT unitary.
 4. Generate tokenized circuits.
 5. Decode tensors to circuits.
+6. Plot a generated circuit.
 :::
 
 :::

src/pipeline/multimodal_diffusion_pipeline.ipynb

@@ -438,7 +438,8 @@
     "        mse_loss_weight_h = (1.0 - alphas_cumprod_h) * F.sigmoid(SNR_h.log())\n",
     "\n",
     "        SNR_w = alphas_cumprod_w / (1.0-alphas_cumprod_w+1e-8) + 1e-8\n",
-    "\n",
+    "        mse_loss_weight_w = (1.0 - alphas_cumprod_w) * F.sigmoid(SNR_w.log() + np.log(3.0 * np.pi**2))\n",
+    "        \n",
     "        #comp mse\n",
     "        mse_flat = lambda out, target: (out-target).square().mean(dim=list(range(1, len(out.shape))))\n",
     "        loss_h = mse_flat(model_output[..., :self.embedder.clr_dim], pred_target_h.detach()) * mse_loss_weight_h.squeeze().detach()\n",

src/pipeline/pipeline.ipynb

@@ -137,9 +137,9 @@
     "    def from_config_file(config_path, device: torch.device, save_path: str=None): return None \n",
     "\n",
     "    @classmethod\n",
-    "    def from_pretrained(cls, repo_id: str, device: torch.device, use_auth_token: bool = False, **kwargs):  \n",
+    "    def from_pretrained(cls, repo_id: str, device: torch.device, **kwargs):  \n",
     "        \"\"\"Load a model pipeline directly from Huggingface.\"\"\"\n",
-    "        model_path = snapshot_download(repo_id=repo_id, repo_type=\"model\", allow_patterns=[\"*.pt\", \"*.yaml\", \"*.safetensors\"], use_auth_token=use_auth_token, **kwargs) \n",
+    "        model_path = snapshot_download(repo_id=repo_id, repo_type=\"model\", allow_patterns=[\"*.pt\", \"*.yaml\", \"*.safetensors\"], **kwargs) \n",
     "        pipeline   = cls.from_config_file(model_path+\"/\", device)  \n",
     "        return pipeline"
    ]

src/platform/backends/circuits_cudaq.ipynb

@@ -247,11 +247,18 @@
     "            \n",
     "        return U\n",
     "\n",
-    "    def draw(self, kernel: cudaq.kernel, num_qubits: int, **kwargs) -> None:\n",
+    "    def draw(self, parametrizedCudaqKernel: ParametrizedCudaqKernel, num_qubits: int, return_str: bool = False, **kwargs) -> None:\n",
     "        \"\"\"Draw the given `cudaq.kernel` using cudaq.\"\"\" \n",
+    "\n",
+    "        kernel, thetas = parametrizedCudaqKernel.kernel, parametrizedCudaqKernel.params\n",
+    "        \n",
     "        c    = [0] * (2**num_qubits)\n",
     "        c[0] = 1\n",
-    "        print(cudaq.draw(kernel, c))"
+    "\n",
+    "        s = cudaq.draw(kernel, c, thetas)\n",
+    "        if return_str:\n",
+    "            return s\n",
+    "        print(s)"
    ]
   },
   {
@@ -333,7 +340,7 @@
       "q2 : ┤ ry(0.1) ├──●───╳───────────────────────\n",
       "     ╰─────────╯                              \n",
       "\n",
-      "Measurement distribution:{ 000:85 010:401 100:85 110:429 }\n",
+      "Measurement distribution:{ 000:76 010:419 100:89 101:1 110:415 }\n",
       "\n"
      ]
     }
@@ -425,8 +432,8 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "Timeit target='qpp-cpu': 1.08 ms ± 58.9 μs per loop (mean ± std. dev. of 7 runs, 1,000 loops each)\n",
-      "Timeit target='nvidia': 13.5 ms ± 3.14 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)\n"
+      "Timeit target='qpp-cpu': 1.04 ms ± 102 μs per loop (mean ± std. dev. of 7 runs, 1,000 loops each)\n",
+      "Timeit target='nvidia': 13.6 ms ± 1.02 ms per loop (mean ± std. dev. of 7 runs, 100 loops each)\n"
      ]
     }
    ],