# ValueError: need at least one array to stack **URL:** \ **Category:** PennyLane Help\ **Created:** [June 21, 2024, 12:01pm UTC](https://discuss.pennylane.ai/t/valueerror-need-at-least-one-array-to-stack/4781 "2024-06-21T12:01:26Z")\ **Posts on this page:** 2\ **Page:** 1
**Author:** ![RX1](https://yyz2.discourse-cdn.com/flex012/user_avatar/discuss.pennylane.ai/rx1/32/698_2.png) [@RX1](https://discuss.pennylane.ai/u/RX1)\ **Post date:** [June 21, 2024, 12:01pm UTC](https://discuss.pennylane.ai/t/valueerror-need-at-least-one-array-to-stack/4781/1 "2024-06-21T12:01:26Z")
I used penny==0.36.0 to train my neural network, and found this error. My code for training is like: ```auto def TrainQKSAN(): opt = qml.NesterovMomentumOptimizer(0.2) batch_size = 15 weights = paras #bias = np.array(0.0, requires_grad=True) his_acc_val = [] his_acc_train = [] his_cost = [] for it in range(30): # Update the weights by one optimizer step batch_index = np.random.randint(0, 200, (batch_size,)) feats_train_batch = x_train[batch_index] Y_train_batch = y_train[batch_index] #cost(params, bias, features, labels): weights, _, _ = opt.step(cost1, weights, feats_train_batch, Y_train_batch) # Compute predictions on train and validation set predictions_train = [np.sign(variational_classifier(weights, f)) for f in x_train] predictions_val = [np.sign(variational_classifier(weights, f)) for f in x_val] # Compute accuracy on train and validation set acc_train = accuracy(y_train, predictions_train) acc_val = accuracy(y_val, predictions_val) his_acc_val.append(acc_val) his_acc_train.append(acc_train) his_cost.append(cost1(weights, features, Y)) print("Iter: {:5d} | Cost: {:0.7f} | Acc train: {:0.7f} | Acc validation: {:0.7f} " "".format(it + 1, cost(weights, features, Y), acc_train, acc_val)) return his_acc_val, his_acc_train, np.array(his_cost) his_acc_val1, his_acc_train1, his_cost1 = TrainQKSAN() print(his_acc_val1, his_acc_train1, his_cost1) ``` ```auto --------------------------------------------------------------------------- ValueError Traceback (most recent call last) Cell In[5], line 81 76 print("Iter: {:5d} | Cost: {:0.7f} | Acc train: {:0.7f} | Acc validation: {:0.7f} " 77 "".format(it + 1, cost(weights, features, Y), acc_train, acc_val)) 79 return his_acc_val, his_acc_train, np.array(his_cost) ---> 81 his_acc_val1, his_acc_train1, his_cost1 = TrainQKSAN() 83 print(his_acc_val1, his_acc_train1, his_cost1) Cell In[5], line 62, in TrainQKSAN() 60 Y_train_batch = y_train[batch_index] 61 #cost(params, bias, features, labels): ---> 62 weights, _, _ = opt.step(cost1, weights, feats_train_batch, Y_train_batch) 64 # Compute predictions on train and validation set 65 predictions_train = [np.sign(variational_classifier(weights, f)) for f in x_train] File /opt/conda/envs/tf1/lib/python3.9/site-packages/pennylane/optimize/gradient_descent.py:93, in GradientDescentOptimizer.step(self, objective_fn, grad_fn, *args, **kwargs) 75 def step(self, objective_fn, *args, grad_fn=None, **kwargs): 76 """Update trainable arguments with one step of the optimizer. 77 78 Args: (...) 90 If single arg is provided, list [array] is replaced by array. 91 """ ---> 93 g, _ = self.compute_grad(objective_fn, args, kwargs, grad_fn=grad_fn) 94 new_args = self.apply_grad(g, args) 96 # unwrap from list if one argument, cleaner return File /opt/conda/envs/tf1/lib/python3.9/site-packages/pennylane/optimize/nesterov_momentum.py:76, in NesterovMomentumOptimizer.compute_grad(self, objective_fn, args, kwargs, grad_fn) 73 shifted_args[index] = args[index] - self.momentum * self.accumulation[index] 75 g = get_gradient(objective_fn) if grad_fn is None else grad_fn ---> 76 grad = g(*shifted_args, **kwargs) 77 forward = getattr(g, "forward", None) 79 grad = (grad,) if len(trainable_indices) == 1 else grad File /opt/conda/envs/tf1/lib/python3.9/site-packages/pennylane/_grad.py:165, in grad. __call__ (self, *args, **kwargs) 162 self._forward = self._fun(*args, **kwargs) 163 return () --> 165 grad_value, ans = grad_fn(*args, **kwargs) # pylint: disable=not-callable 166 self._forward = ans 168 return grad_value File /opt/conda/envs/tf1/lib/python3.9/site-packages/autograd/wrap_util.py:20, in unary_to_nary..nary_operator..nary_f(*args, **kwargs) 18 else: 19 x = tuple(args[i] for i in argnum) ---> 20 return unary_operator(unary_f, x, *nary_op_args, **nary_op_kwargs) File /opt/conda/envs/tf1/lib/python3.9/site-packages/pennylane/_grad.py:191, in grad._grad_with_forward(fun, x) 185 if vspace(ans).size != 1: 186 raise TypeError( 187 "Grad only applies to real scalar-output functions. " 188 "Try jacobian, elementwise_grad or holomorphic_grad." 189 ) --> 191 grad_value = vjp(vspace(ans).ones()) 192 return grad_value, ans File /opt/conda/envs/tf1/lib/python3.9/site-packages/autograd/core.py:14, in make_vjp..vjp(g) ---> 14 def vjp(g): return backward_pass(g, end_node) File /opt/conda/envs/tf1/lib/python3.9/site-packages/autograd/core.py:21, in backward_pass(g, end_node) 19 for node in toposort(end_node): 20 outgrad = outgrads.pop(node) ---> 21 ingrads = node.vjp(outgrad[0]) 22 for parent, ingrad in zip(node.parents, ingrads): 23 outgrads[parent] = add_outgrads(outgrads.get(parent), ingrad) File /opt/conda/envs/tf1/lib/python3.9/site-packages/autograd/core.py:67, in defvjp..vjp_argnums..(g) 64 raise NotImplementedError( 65 "VJP of {} wrt argnum 0 not defined".format(fun. __name__ )) 66 vjp = vjpfun(ans, *args, **kwargs) ---> 67 return lambda g: (vjp(g),) 68 elif L == 2: 69 argnum_0, argnum_1 = argnums File /opt/conda/envs/tf1/lib/python3.9/site-packages/pennylane/workflow/interfaces/autograd.py:199, in vjp..grad_fn(dy) 196 def grad_fn(dy): 197 """Returns the vector-Jacobian product with given 198 parameter values and output gradient dy""" --> 199 vjps = jpc.compute_vjp(tapes, dy) 200 return tuple( 201 qml.math.to_numpy(v, max_depth=1) if isinstance(v, ArrayBox) else v for v in vjps 202 ) File /opt/conda/envs/tf1/lib/python3.9/site-packages/pennylane/workflow/jacobian_products.py:296, in TransformJacobianProducts.compute_vjp(self, tapes, dy) 293 logger.debug("compute_vjp called with (%s, %s)", tapes, dy) 295 if self._cache_full_jacobian: --> 296 jacs = self.compute_jacobian(tapes) 297 return _compute_vjps(jacs, dy, tapes) 299 vjp_tapes, processing_fn = qml.gradients.batch_vjp( 300 tapes, dy, self._gradient_transform, gradient_kwargs=self._gradient_kwargs 301 ) File /opt/conda/envs/tf1/lib/python3.9/site-packages/pennylane/workflow/jacobian_products.py:326, in TransformJacobianProducts.compute_jacobian(self, tapes) 324 if tapes in self._cache: 325 return self._cache[tapes] --> 326 jac_tapes, batch_post_processing = self._gradient_transform(tapes, **self._gradient_kwargs) 327 results = self._inner_execute(jac_tapes) 328 jacs = tuple(batch_post_processing(results)) File /opt/conda/envs/tf1/lib/python3.9/site-packages/pennylane/transforms/core/transform_dispatcher.py:130, in TransformDispatcher. __call__ (self, *targs, **tkwargs) 128 return self._qfunc_transform(obj, targs, tkwargs) 129 if isinstance(obj, Sequence) and all(isinstance(q, qml.tape.QuantumScript) for q in obj): --> 130 return self._batch_transform(obj, targs, tkwargs) 132 # Input is not a QNode nor a quantum tape nor a device. 133 # Assume Python decorator syntax: 134 # 135 # result = some_transform(*transform_args)(qnode)(*qnode_args) 137 raise TransformError( 138 "Decorating a QNode with @transform_fn(**transform_kwargs) has been " 139 "removed. Please decorate with @functools.partial(transform_fn, **transform_kwargs) " (...) 142 "https://docs.pennylane.ai/en/stable/development/deprecations.html#completed-deprecation-cycles", 143 ) File /opt/conda/envs/tf1/lib/python3.9/site-packages/pennylane/transforms/core/transform_dispatcher.py:337, in TransformDispatcher._batch_transform(self, original_batch, targs, tkwargs) 331 tape_counts = [] 333 for t in original_batch: 334 # Preprocess the tapes by applying batch transforms 335 # to each tape, and storing corresponding tapes 336 # for execution, processing functions, and list of tape lengths. --> 337 new_tapes, fn = self(t, *targs, **tkwargs) 338 execution_tapes.extend(new_tapes) 339 batch_fns.append(fn) File /opt/conda/envs/tf1/lib/python3.9/site-packages/pennylane/transforms/core/transform_dispatcher.py:101, in TransformDispatcher. __call__ (self, *targs, **tkwargs) 99 start = 0 100 for tape in expanded_tapes: --> 101 intermediate_tapes, post_processing_fn = self._transform( 102 tape, *targs, **tkwargs 103 ) 104 transformed_tapes.extend(intermediate_tapes) 105 end = start + len(intermediate_tapes) File /opt/conda/envs/tf1/lib/python3.9/site-packages/pennylane/gradients/parameter_shift.py:1117, in param_shift(tape, argnum, shifts, gradient_recipes, fallback_fn, f0, broadcast) 1115 g_tapes, fn = var_param_shift(tape, argnum, shifts, gradient_recipes, f0, broadcast) 1116 else: -> 1117 g_tapes, fn = expval_param_shift(tape, argnum, shifts, gradient_recipes, f0, broadcast) 1119 gradient_tapes.extend(g_tapes) 1121 if unsupported_params: File /opt/conda/envs/tf1/lib/python3.9/site-packages/pennylane/gradients/parameter_shift.py:396, in expval_param_shift(tape, argnum, shifts, gradient_recipes, f0, broadcast) 393 gradient_data.append((1, np.array([1.0]), h_fn, None, g_tapes[0].batch_size)) 394 continue --> 396 recipe = _choose_recipe(argnum, idx, gradient_recipes, shifts, tape) 397 recipe, at_least_one_unshifted, unshifted_coeff = _extract_unshifted( 398 recipe, at_least_one_unshifted, f0, gradient_tapes, tape 399 ) 400 coeffs, multipliers, op_shifts = recipe.T File /opt/conda/envs/tf1/lib/python3.9/site-packages/pennylane/gradients/parameter_shift.py:116, in _choose_recipe(argnum, idx, gradient_recipes, shifts, tape) 114 else: 115 op_shifts = None if shifts is None else shifts[arg_idx] --> 116 recipe = _get_operation_recipe(tape, idx, shifts=op_shifts) 117 return recipe File /opt/conda/envs/tf1/lib/python3.9/site-packages/pennylane/gradients/parameter_shift.py:286, in _get_operation_recipe(tape, t_idx, shifts, order) 280 raise qml.operation.OperatorPropertyUndefined( 281 f"The operation {op.name} does not have a grad_recipe, parameter_frequencies or " 282 "a generator defined. No parameter shift rule can be applied." 283 ) from e 285 # Create shift rule from frequencies with given shifts --> 286 coeffs, shifts = qml.gradients.generate_shift_rule(frequencies, shifts=shifts, order=order).T 287 # The generated shift rules do not include a rescaling of the parameter, only shifts. 288 mults = np.ones_like(coeffs) File /opt/conda/envs/tf1/lib/python3.9/site-packages/pennylane/gradients/general_shift_rules.py:316, in generate_shift_rule(frequencies, shifts, order) 244 r"""Computes the parameter shift rule for a unitary based on its generator's eigenvalue 245 frequency spectrum. 246 (...) 313 :math:`\frac{\partial^2 f}{\partial \phi^2} = \frac{1}{2} \left[f(\phi) - f(\phi-\pi)\right]`. 314 """ 315 frequencies = tuple(f for f in frequencies if f > 0) --> 316 rule = _get_shift_rule(frequencies, shifts=shifts) 318 if order > 1: 319 T = frequencies_to_period(frequencies) File /opt/conda/envs/tf1/lib/python3.9/site-packages/pennylane/gradients/general_shift_rules.py:142, in _get_shift_rule(frequencies, shifts) 139 @functools.lru_cache(maxsize=None) 140 def _get_shift_rule(frequencies, shifts=None): 141 n_freqs = len(frequencies) --> 142 frequencies = qml.math.sort(qml.math.stack(frequencies)) 143 freq_min = frequencies[0] 145 if len(set(frequencies)) != n_freqs or freq_min <= 0: File /opt/conda/envs/tf1/lib/python3.9/site-packages/pennylane/math/multi_dispatch.py:151, in multi_dispatch..decorator..wrapper(*args, **kwargs) 148 interface = interface or get_interface(*dispatch_args) 149 kwargs["like"] = interface --> 151 return fn(*args, **kwargs) File /opt/conda/envs/tf1/lib/python3.9/site-packages/pennylane/math/multi_dispatch.py:497, in stack(values, axis, like) 468 """Stack a sequence of tensors along the specified axis. 469 470 .. warning:: (...) 494 [5.00e+00, 8.00e+00, 1.01e+02]], dtype=float32)> 495 """ 496 values = np.coerce(values, like=like) --> 497 return np.stack(values, axis=axis, like=like) File /opt/conda/envs/tf1/lib/python3.9/site-packages/autoray/autoray.py:80, in do(fn, like, *args, **kwargs) 31 """Do function named ``fn`` on ``(*args, **kwargs)``, peforming single 32 dispatch to retrieve ``fn`` based on whichever library defines the class of 33 the ``args[0]``, or the ``like`` keyword argument if specified. (...) 77 78 """ 79 backend = choose_backend(fn, *args, like=like, **kwargs) ---> 80 return get_lib_fn(backend, fn)(*args, **kwargs) File < __array_function__ internals>:180, in stack(*args, **kwargs) File /opt/conda/envs/tf1/lib/python3.9/site-packages/numpy/core/shape_base.py:422, in stack(arrays, axis, out) 420 arrays = [asanyarray(arr) for arr in arrays] 421 if not arrays: --> 422 raise ValueError('need at least one array to stack') 424 shapes = {arr.shape for arr in arrays} 425 if len(shapes) != 1: ValueError: need at least one array to stack ``` ​ ---
**Author:** ![CatalinaAlbornoz](https://yyz2.discourse-cdn.com/flex012/user_avatar/discuss.pennylane.ai/catalinaalbornoz/32/1196_2.png) [@CatalinaAlbornoz](https://discuss.pennylane.ai/u/CatalinaAlbornoz)\ **Post date:** [June 24, 2024, 6:44pm UTC](https://discuss.pennylane.ai/t/valueerror-need-at-least-one-array-to-stack/4781/2 "2024-06-24T18:44:00Z")
Hi @RX1 , Unfortunately the code you shared is not self-contained so I can’t run it to try to replicate your issue. It looks to me like you’re looking to adapt the [variational classifier demo](https://pennylane.ai/qml/demos/tutorial_variational_classifier/). However if you change the dataset there are many things you need to change. In this case, with your new data (or the way you’re handling it) `Y_train_batch = y_train[batch_index]` no longer works. I don’t know your dataset but maybe something like this can work: `Y_train_batch = [Y[i] for i in batch_index]`. I hope this helps.