mypy in vqnhe codebase

Author: refraction-ray

docs/source/tutorials/tfim_vqe_cn.ipynb

@@ -164,9 +164,7 @@
    "source": [
     "def vqe_tfim(param, n, nlayers):\n",
     "    c = tc.Circuit(n)\n",
-    "    paramc = tc.backend.cast(\n",
-    "        param, tc.dtypestr\n",
-    "    )  # 我们假设输入参数的 dtype 为 float64\n",
+    "    paramc = tc.backend.cast(param, tc.dtypestr)  # 我们假设输入参数的 dtype 为 float64\n",
     "    for i in range(n):\n",
     "        c.H(i)\n",
     "    for j in range(nlayers):\n",

docs/source/tutorials/tfim_vqe_diffreph_cn.ipynb

@@ -64,7 +64,7 @@
     "tc.set_dtype(\"complex128\")\n",
     "dtype = np.complex128\n",
     "\n",
-    "xx = tc.gates._xx_matrix  #要使用的 xx 门矩阵"
+    "xx = tc.gates._xx_matrix  # 要使用的 xx 门矩阵"
    ]
   },
   {
@@ -84,7 +84,7 @@
    "source": [
     "n = 4  # 量子比特数\n",
     "nlayers = 2  # 电路层数\n",
-    "ntrials = 2  #随机电路实例数"
+    "ntrials = 2  # 随机电路实例数"
    ]
   },
   {
@@ -169,7 +169,7 @@
    "source": [
     "vqe_tfim_paulistring_vvag = tc.backend.jit(\n",
     "    tc.backend.vectorized_value_and_grad(vqe_tfim_paulistring)\n",
-    ")  # 使用 vvag 获取不同随机电路实例的损失函数和梯度\n"
+    ")  # 使用 vvag 获取不同随机电路实例的损失函数和梯度"
    ]
   },
   {
@@ -222,9 +222,7 @@
     "    )  # 初始参数\n",
     "    opt = tf.keras.optimizers.Adam(1e-2)\n",
     "    for i in range(maxiter):\n",
-    "        e, grad = vqe_tfim_paulistring_vvag(\n",
-    "            param.value(), n, nlayers\n",
-    "        )  # 能量和梯度\n",
+    "        e, grad = vqe_tfim_paulistring_vvag(param.value(), n, nlayers)  # 能量和梯度\n",
     "        opt.apply_gradients([(grad, param)])\n",
     "        if i % 200 == 0:\n",
     "            print(e)\n",
@@ -383,9 +381,7 @@
     "\n",
     "    opt = tf.keras.optimizers.Adam(1e-2)\n",
     "    for i in range(maxiter):\n",
-    "        e, grad = vqe_tfim_vvag(\n",
-    "            param.value(), n, nlayers, hamiltonian\n",
-    "        )  # 能量和梯度\n",
+    "        e, grad = vqe_tfim_vvag(param.value(), n, nlayers, hamiltonian)  # 能量和梯度\n",
     "        opt.apply_gradients([(grad, param)])\n",
     "        if i % 200 == 0:\n",
     "            print(e)\n",
@@ -424,9 +420,7 @@
     "\n",
     "Jx = np.array([1.0 for _ in range(n - 1)])  # xx 相互作用的强度 (OBC)\n",
     "Bz = np.array([1.0 for _ in range(n)])  # 横向场强\n",
-    "hamiltonian_mpo = tn.matrixproductstates.mpo.FiniteTFI(\n",
-    "    Jx, Bz, dtype=dtype\n",
-    ")  # 矩阵乘积算子\n",
+    "hamiltonian_mpo = tn.matrixproductstates.mpo.FiniteTFI(Jx, Bz, dtype=dtype)  # 矩阵乘积算子\n",
     "hamiltonian_mpo = quoperator_mpo(hamiltonian_mpo)  # 从 mpo 生成 QuOperator"
    ]
   },

examples/vqnhe_h6.py

@@ -5,7 +5,6 @@
 
 sys.path.insert(0, "../")
 import numpy as np
-import tensorflow as tf
 import tensorcircuit as tc
 from tensorcircuit.applications.vqes import VQNHE, JointSchedule, construct_matrix_v3
 

tensorcircuit/applications/vqes.py

@@ -239,7 +239,7 @@ def __init__(
         self.base = tf.constant(list(product(*[[0, 1] for _ in range(n)])))
         self.hamiltonian = hamiltonian
         self.shortcut = shortcut
-        self.history = []
+        self.history: List[float] = []
 
     def assign(
         self, c: Optional[List[Tensor]] = None, q: Optional[Tensor] = None
@@ -503,7 +503,7 @@ def training(
         debug: int = 100,
         onlyq: int = 0,
         checkpoints: Optional[List[Tuple[int, float]]] = None,
-    ) -> Tuple[Array, Array, Array, int]:
+    ) -> Tuple[Array, Array, Array, int, List[float]]:
         loss_prev = 0
         loss_opt = 1e8
         j_opt = 0
@@ -513,25 +513,25 @@ def training(
         self.ccache = self.model.get_weights()
 
         if not optc:
-            optc = 0.001
+            optc = 0.001  # type: ignore
         if isinstance(optc, float) or isinstance(
             optc, tf.keras.optimizers.schedules.LearningRateSchedule
         ):
             optc = tf.keras.optimizers.Adam(optc)
         if isinstance(optc, tf.keras.optimizers.Optimizer):
             optcf = lambda _: optc
         else:
-            optcf = optc
+            optcf = optc  # type: ignore
         if not optq:
-            optq = 0.01
+            optq = 0.01  # type: ignore
         if isinstance(optq, float) or isinstance(
             optq, tf.keras.optimizers.schedules.LearningRateSchedule
         ):
             optq = tf.keras.optimizers.Adam(optq)
         if isinstance(optq, tf.keras.optimizers.Optimizer):
             optqf = lambda _: optq
         else:
-            optqf = optq
+            optqf = optq  # type: ignore
 
         nm = tf.constant(1.0)
         times = []
@@ -592,7 +592,7 @@ def training(
         quantume, _ = self.plain_evaluation(self.circuit_variable)
         print(
             "vqnhe prediction: ",
-            loss_prev.numpy(),
+            loss_prev.numpy(),  # type: ignore
             "quantum part energy: ",
             quantume.numpy(),
             "classical model scale: ",
@@ -601,8 +601,8 @@ def training(
         print("----------END TRAINING------------")
         self.history += history[:j_opt]
         return (
-            loss_opt.numpy(),
-            np.real(nm_opt.numpy()),
+            loss_opt.numpy(),  # type: ignore
+            np.real(nm_opt.numpy()),  # type: ignore
             quantume.numpy(),
             j_opt,
             history[:j_opt],
@@ -666,6 +666,6 @@ def multi_training(
         if rs:
             es = [r["energy"] for r in rs]
             ind = np.argmin(es)
-            self.assign(rs[ind]["model_weights"], rs[ind]["circuit_weights"])
-            self.history = rs[ind]["history"]
+            self.assign(rs[ind]["model_weights"], rs[ind]["circuit_weights"])  # type: ignore
+            self.history = rs[ind]["history"]  # type: ignore
         return rs