Here I share a code snippet for training Encoder_Decoder Model for machine translation. While Using the Embedding layer (trained previously) during inference mode( on test_data) . It threw the following error --->
# Encoder
encoder_inputs = Input(shape=(None ,))
enc_emb = Embedding(eng_vocab_size, latent_dim, mask_zero = True)(encoder_inputs)
encoder_lstm = LSTM(latent_dim, return_state=True)
encoder_outputs, state_h, state_c = encoder_lstm(enc_emb)
# We discard `encoder_outputs` and only keep the states.
encoder_states = [state_h, state_c]
# Set up the decoder, using `encoder_states` as initial state.
decoder_inputs = Input(shape=(None,))
dec_emb = Embedding(deu_vocab_size, latent_dim, mask_zero = True)(decoder_inputs)
# decoder return full output sequences, and internal states as well.
# We don't use the return states in the training model,
# but we will use them in inference.
decoder_lstm = LSTM(latent_dim, return_sequences=True, return_state=True)
decoder_outputs, _, _ = decoder_lstm(dec_emb,
initial_state=encoder_states)
decoder_dense = Dense(deu_vocab_size, activation='softmax')
decoder_outputs = decoder_dense(decoder_outputs)
# Define the model that will turn
# `encoder_input_data` & `decoder_input_data` into `decoder_target_data`
model = Model([encoder_inputs, decoder_inputs], decoder_outputs)
# Compile the model
model.compile(optimizer='rmsprop', loss='categorical_crossentropy', metrics=['acc'])
# Encode the input sequence to get the "thought vectors"
encoder_model = Model(encoder_inputs, encoder_states)
# Decoder setup
# Below tensors will hold the states of the previous time step
decoder_state_input_h = Input(shape=(latent_dim,))
decoder_state_input_c = Input(shape=(latent_dim,))
decoder_states_inputs = [decoder_state_input_h, decoder_state_input_c]
dec_emb2= dec_emb(decoder_inputs) # reusing embedding layer
decoder_outputs2, state_h2, state_c2 = decoder_lstm(dec_emb2, initial_state=decoder_states_inputs) # reusing lstm layer
decoder_outputs2 = decoder_dense(decoder_outputs2) # softmax_layer to generate prob_dist. over target vocab
# Final decoder model
decoder_model = Model(
[decoder_inputs] + decoder_states_inputs,
[decoder_outputs2] )
ERROR
8 decoder_states_inputs = [decoder_state_input_h, decoder_state_input_c]
---> 10 dec_emb2= dec_emb(decoder_inputs) # reusing embedding layer
11
12 decoder_outputs2, state_h2, state_c2 = decoder_lstm(dec_emb2, initial_state=decoder_states_inputs) # reusing lstm layer
TypeError: 'KerasTensor' object is not callableenter image description here
I read through various solutions available for this issue , but couldn't understand what 2 modes of model they were talking about and what their soltion was effectively doing .
Pls explain in detail. Thanks in advance
Related
I have 100s of csvs in a directory, with headers. I am trying to create a feedforward NN using tensorflow for regression.
What's the best way to import these csvs and train using tf & train it?
Also help to look at my preprocessing if I am doing it right?
Note: My features has mixed datatypes (int,float,string), My target is float
I can not concat the csv and use pandas to import, my data size is >50 GB so can not load it in-memory, have to read it iteratively from disc
Directory Path:
./data/train/ -> 100s of csvs
./data/test -> 100s of csvs
./data/valid -> 100s of csvs
Code:
Methodology:
Create Generator
Use Dataset API to load the data
Preprocess the Data (embedding, one-hot,etc)
Train fit
But, in generator I was able to give only output formats where the inputs/ outputs are homogeneous ddtypes.
Code:
def data_generator(file_list, batch_size = 2):
i = 0
while True:
if i*batch_size >= len(file_list): # This loop is used to run the generator indefinitely.
i = 0
np.random.shuffle(file_list)
else:
file_chunk = file_list[i*batch_size:(i+1)*batch_size]
data = []
labels = []
for file in file_chunk:
temp = pd.read_csv(open(file,'r')) # Change this line to read any other type of file
labels = temp.pop('ACTUAL_BOXES')
data.append(temp.values) # Convert column data to matrix like data with one channel
labels.append(labels)
data = np.asarray(data)
labels = np.asarray(labels)
yield data, labels # Here data will be mixed datatype arrays & lables will be a float dtype array
i = i + 1
#getting list of files inside the directory
train_file_list = np.sort(glob.glob('././data/train/*.csv'))
test_file_list = np.sort(glob.glob('././data/test/*.csv'))
val_file_list = np.sort(glob.glob('././data/val/*.csv'))
train_dataset = tf.data.Dataset.from_generator(data_generator,args= [train_file_list , batch_size = 2],
output_types = (tf.float32, tf.float32), #This is where I am struck
#my sample data and lables will be like this
data = ['a','b',1,2,3.14,2] #Mixed dtypes
lables = [1.0] #float
)
val_dataset = tf.data.Dataset.from_generator(data_generator,args= [val_file_list , batch_size = 2],
output_types = (tf.float32, tf.float32), #This is where I am struck
)
# Pre processing Part:
def encode_inputs(EMBEDDING_FEATURES,INDICATOR_FEATURES):
''' Function for encoding the deatures'''
encoded_features = []
for feature_name in EMBEDDING_FEATURES:
#Getting unique vocab list
vocabulary = np.array(list(flatten(vocab_list[feature_name])))
# categorical columns using the lists created above:
cat_col = tf.feature_column.categorical_column_with_vocabulary_list(
feature_name, vocabulary)
embedding_dims = int(math.sqrt(len(vocabulary)))
# create an embedding from the categorical column:
cat_emb = tf.feature_column.embedding_column(cat_col,8) #,dimension=embedding_dims
# add the embeddings to the list of feature columns
encoded_features.append(cat_emb)
for feature_name in INDICATOR_FEATURES:
#Getting unique vocab list
vocabulary = list(flatten(vocab_list[feature_name]))
# indicator columns using the lists created above:
ind_col = tf.feature_column.categorical_column_with_vocabulary_list(
feature_name, vocabulary)
# create an embedding from the categorical column:
cat_one_hot = tf.feature_column.indicator_column(ind_col)
# add the embeddings to the list of feature columns
encoded_features.append(cat_one_hot)
# create the input layer for the model
feature_layer = tf.keras.layers.DenseFeatures(encoded_features)
return feature_layer
# Opening JSON file that contains vocab list for str cols
f = open('./vocab_list.json') # File that contains the unique values of each feature
vocab_list = json.load(f)
features_layer = encode_inputs(EMBEDDING_FEATURES,INDICATOR_FEATURES)
# Model Part
model = tf.keras.models.Sequential([
features_layer,
tf.keras.layers.Dense(30, activation = 'relu'),
tf.keras.layers.Dense(1)
])
m_loss = tf.keras.losses.mean_squared_error
m_optimizer = tf.keras.optimizers.SGD(lr = 1e-3)
batch_size = 32
model.compile(loss = m_loss, optimizer = m_optimizer, metrics = ['accuracy'])
model.fit(train_dataset ,epochs = 10, validation_data = val_dataset )
I am using the https://github.com/mdhmz1/Auto-Annotate repo. I have tried to custom train my own dataset which has it own COCO JSON format file.
When I try to run
python3 customTrain.py train --dataset=path/to/dir --weights=coco
I get the following error:
Traceback (most recent call last):
File "customTrain.py", line 279, in
train(model)
File "customTrain.py", line 179, in train
dataset_train.load_custom(args.dataset, "train")
File "customTrain.py", line 87, in load_custom
annotations = [a for a in annotations if a['regions']]
File "customTrain.py", line 87, in
annotations = [a for a in annotations if a['regions']]
TypeError: list indices must be integers or slices, not str
My customtrain.py looks like the following:
import os
import sys
import json
import datetime
import numpy as np
import skimage.draw
Root directory of the project
ROOT_DIR = "/home/hiwi/Auto-Annotate"
Import Mask RCNN
sys.path.append(ROOT_DIR) # To find local version of the library
from mrcnn.config import Config
from mrcnn import model as modellib, utils
Path to trained weights file
COCO_WEIGHTS_PATH = os.path.join(ROOT_DIR, "mask_rcnn_coco.h5")
Directory to save logs and model checkpoints, if not provided
through the command line argument --logs
DEFAULT_LOGS_DIR = os.path.join(ROOT_DIR, "logs")
############################################################
Configurations
############################################################
class CustomConfig(Config):
"""Configuration for training on the toy dataset.
Derives from the base Config class and overrides some values.
"""
# Give the configuration a recognizable name
NAME = "custom"
IMAGES_PER_GPU = 1
# Number of classes (including background)
NUM_CLASSES = 1 + 2 # Background + 2 classes
# Number of training steps per epoch
STEPS_PER_EPOCH = 100
# Skip detections with < 90% confidence
DETECTION_MIN_CONFIDENCE = 0.9
############################################################
Dataset
############################################################
class CustomDataset(utils.Dataset):
def load_custom(self, dataset_dir, subset):
"""Load a subset of the Custom dataset.
dataset_dir: Root directory of the dataset.
subset: Subset to load: train or val
"""
# Add classes. We have only one class to add.
self.add_class("custom", 0, "Primary_Track")
self.add_class("custom", 1, "Secondary_Track")
# Train or validation dataset?
assert subset in ["train", "val"]
dataset_dir = os.path.join(dataset_dir, subset)
# Load annotations
# VGG Image Annotator (up to version 1.6) saves each image in the form:
# { 'filename': '28503151_5b5b7ec140_b.jpg',
# 'regions': {
# '0': {
# 'region_attributes': {},
# 'shape_attributes': {
# 'all_points_x': [...],
# 'all_points_y': [...],
# 'name': 'polygon'}},
# ... more regions ...
# },
# 'size': 100202
# }
# We mostly care about the x and y coordinates of each region
# Note: In VIA 2.0, regions was changed from a dict to a list.
annotations1 = json.load(open(os.path.join(dataset_dir, "train.json")))
annotations = list(annotations1.values()) # don't need the dict keys
# The VIA tool saves images in the JSON even if they don't have any
# annotations. Skip unannotated images.
annotations = [a for a in annotations if a['regions']]
# Add images
for a in annotations:
# Get the x, y coordinaets of points of the polygons that make up
# the outline of each object instance. These are stores in the
# shape_attributes (see json format above)
# The if condition is needed to support VIA versions 1.x and 2.x.
if type(a['regions']) is dict:
polygons = [r['shape_attributes'] for r in a['regions'].values()]
else:
polygons = [r['shape_attributes'] for r in a['regions']]
#labelling each class in the given image to a number
custom = [s['region_attributes'] for s in a['regions']]
num_ids=[]
#Add the classes according to the requirement
for n in custom:
try:
if n['name']=="Primary_Track":
num_ids.append(0)
elif n['name']=='Secondary_Track':
num_ids.append(1)
except:
pass
# load_mask() needs the image size to convert polygons to masks.
# Unfortunately, VIA doesn't include it in JSON, so we must read
# the image. This is only managable since the dataset is tiny.
image_path = os.path.join(dataset_dir, a['filename'])
image = skimage.io.imread(image_path)
height, width = image.shape[:2]
self.add_image(
"custom",
image_id=a['filename'], # use file name as a unique image id
path=image_path,
width=width, height=height,
polygons=polygons,
num_ids=num_ids)
def load_mask(self, image_id):
"""Generate instance masks for an image.
Returns:
masks: A bool array of shape [height, width, instance count] with
one mask per instance.
class_ids: a 1D array of class IDs of the instance masks.
"""
# If not a custom dataset image, delegate to parent class.
image_info = self.image_info[image_id]
if image_info["source"] != "custom":
return super(self.__class__, self).load_mask(image_id)
num_ids = image_info['num_ids']
#print("Here is the numID",num_ids)
# Convert polygons to a bitmap mask of shape
# [height, width, instance_count]
info = self.image_info[image_id]
mask = np.zeros([info["height"], info["width"], len(info["polygons"])],
dtype=np.uint8)
for i, p in enumerate(info["polygons"]):
if p['name'] == 'polygon':
# Get indexes of pixels inside the polygon and set them to 1
rr, cc = skimage.draw.polygon(p['all_points_y'], p['all_points_x'])
else:
rr, cc = skimage.draw.rectangle((p['y'], p['x']), extent=(p['height'], p['width']))
rr[rr > mask.shape[0]-1] = mask.shape[0]-1
cc[cc > mask.shape[1]-1] = mask.shape[1]-1
mask[rr, cc, i] = 1
# Return mask, and array of class IDs of each instance. Since we have
# one class ID only, we return an array of 1s
num_ids = np.array(num_ids, dtype=np.int32)
return mask.astype(np.bool), num_ids.astype(np.bool), np.ones([mask.shape[-1]], dtype=np.int32)
#return mask.astype(np.bool), np.ones([mask.shape[-1]], dtype=np.int32)
def image_reference(self, image_id):
"""Return the path of the image."""
info = self.image_info[image_id]
if info["source"] == "Railtrack":
return info["path"]
else:
super(self.__class__, self).image_reference(image_id)
def train(model):
"""Train the model."""
# Training dataset.
dataset_train = CustomDataset()
dataset_train.load_custom(args.dataset, "train")
dataset_train.prepare()
# Validation dataset
dataset_val = CustomDataset()
dataset_val.load_custom(args.dataset, "val")
dataset_val.prepare()
# *** This training schedule is an example. Update to your needs ***
# Since we're using a very small dataset, and starting from
# COCO trained weights, we don't need to train too long. Also,
# no need to train all layers, just the heads should do it.
print("Training network heads")
model.train(dataset_train, dataset_val,
learning_rate=config.LEARNING_RATE,
epochs=30,
layers='heads')
############################################################
Training
############################################################
if name == 'main':
import argparse
# Parse command line arguments
parser = argparse.ArgumentParser(
description='Train Mask R-CNN to detect custom objects.')
parser.add_argument("command",
metavar="<command>",
help="'train' or 'splash'")
parser.add_argument('--dataset', required=False,
metavar="/path/to/custom/dataset/",
help='Directory of the Custom dataset')
parser.add_argument('--weights', required=True,
metavar="/path/to/weights.h5",
help="Path to weights .h5 file or 'coco'")
parser.add_argument('--logs', required=False,
default=DEFAULT_LOGS_DIR,
metavar="/path/to/logs/",
help='Logs and checkpoints directory (default=logs/)')
parser.add_argument('--image', required=False,
metavar="path or URL to image",
help='Image to apply the color splash effect on')
parser.add_argument('--video', required=False,
metavar="path or URL to video",
help='Video to apply the color splash effect on')
args = parser.parse_args()
# Validate arguments
if args.command == "train":
assert args.dataset, "Argument --dataset is required for training"
elif args.command == "splash":
assert args.image or args.video,\
"Provide --image or --video to apply color splash"
print("Weights: ", args.weights)
print("Dataset: ", args.dataset)
print("Logs: ", args.logs)
# Configurations
if args.command == "train":
config = CustomConfig()
# Create model
if args.command == "train":
model = modellib.MaskRCNN(mode="training", config=config,
model_dir=args.logs)
# Select weights file to load
if args.weights.lower() == "coco":
weights_path = COCO_WEIGHTS_PATH
# Download weights file
if not os.path.exists(weights_path):
utils.download_trained_weights(weights_path)
elif args.weights.lower() == "last":
# Find last trained weights
weights_path = model.find_last()
elif args.weights.lower() == "imagenet":
# Start from ImageNet trained weights
weights_path = model.get_imagenet_weights()
else:
weights_path = args.weights
# Load weights
print("Loading weights ", weights_path)
if args.weights.lower() == "coco":
# Exclude the last layers because they require a matching
# number of classes
model.load_weights(weights_path, by_name=True, exclude=[
"mrcnn_class_logits", "mrcnn_bbox_fc",
"mrcnn_bbox", "mrcnn_mask"])
else:
model.load_weights(weights_path, by_name=True)
# Train or evaluate
if args.command == "train":
train(model)
else:
print("'{}' is not recognized. "
"Use 'train' or 'splash'".format(args.command))
I am trying to convert pytorch model with multiple networks to ONNX, and encounter some problem.
The git repo: https://github.com/InterDigitalInc/HRFAE
The Trainer Class:
class Trainer(nn.Module):
def __init__(self, config):
super(Trainer, self).__init__()
# Load Hyperparameters
self.config = config
# Networks
self.enc = Encoder()
self.dec = Decoder()
self.mlp_style = Mod_Net()
self.dis = Dis_PatchGAN()
...
Here is how the trained model process image:
def gen_encode(self, x_a, age_a, age_b=0, training=False, target_age=0):
if target_age:
self.target_age = target_age
age_modif = self.target_age*torch.ones(age_a.size()).type_as(age_a)
else:
age_modif = self.random_age(age_a, diff_val=25)
# Generate modified image
self.content_code_a, skip_1, skip_2 = self.enc(x_a)
style_params_a = self.mlp_style(age_a)
style_params_b = self.mlp_style(age_modif)
x_a_recon = self.dec(self.content_code_a, style_params_a, skip_1, skip_2)
x_a_modif = self.dec(self.content_code_a, style_params_b, skip_1, skip_2)
return x_a_recon, x_a_modif, age_modif
And as following is how I did to convert to onnx:
enc = Encoder()
dec = Decoder()
mlp = Mod_Net()
layers = [enc, mlp, dec]
model = torch.nn.Sequential(*layers)
# here is my confusion: how do I specify the inputs of each layer??
# E.g. one of the outputs of 'enc' layer should be input of 'mlp' layer,
# or the outputs of 'enc' layer should be part of inputs of 'dec' layer...
params = torch.load('./logs/001/checkpoint')
model[0].load_state_dict(params['enc_state_dict'])
model[1].load_state_dict(params['mlp_style_state_dict'])
model[2].load_state_dict(params['dec_state_dict'])
torch.onnx.export(model, torch.randn([1, 3, 1024, 1024]), 'trained_hrfae.onnx', do_constant_folding=True)
Maybe the convert-part code is in wrong way??
Could anyone help, many thanks!
#20210629-11:52GMT Edit:
I found there's constraint of using torch.nn.Sequential. The output of former layer in Sequential should be consistent with latter input.
So my code shouldn't work at all because the output of 'enc' layer is not consistent with input of 'mlp' layer.
Could anyone help how to convert this type of pytorch model to onnx? Many thanks, again :)
After research and try, I found a method which maybe in correct way:
Convert each net(Encoder, Mod_Net, Decoder) to onnx model, and handle their input/output in latter logic-process or any further procedure (e.g convert to tflite model).
I'm trying to port onto Android using this method.
#Edit 20210705-03:52GMT#
Another approach may be better: write a new net combines the three nets. I've prove the output is same as origin pytorch model.
class HRFAE(nn.Module):
def __init__(self):
super(HRFAE, self).__init__()
self.enc = Encoder()
self.mlp_style = Mod_Net()
self.dec = Decoder()
def forward(self, x, age_modif):
content_code_a, skip_1, skip_2 = self.enc(x)
style_params_b = self.mlp_style(age_modif)
x_a_modif = self.dec(content_code_a, style_params_b, skip_1, skip_2)
return x_a_modif
and then convert use following:
net = HRFAE()
params = torch.load('./logs/002/checkpoint')
net.enc.load_state_dict(params['enc_state_dict'])
net.mlp_style.load_state_dict(params['mlp_style_state_dict'])
net.dec.load_state_dict(params['dec_state_dict'])
net.eval()
torch.onnx.export(net, (torch.randn([1, 3, 512, 512]), torch.randn([1]).type(torch.long)), 'test_hrfae.onnx')
This should be the answer.
I have trained a network using MXnet, but am not sure how I can save and load the parameters for later use. First I define and train the network:
dataIn = mx.sym.var('data')
fc1 = mx.symbol.FullyConnected(data=dataIn, num_hidden=100)
act1 = mx.sym.Activation(data=fc1, act_type="relu")
fc2 = mx.symbol.FullyConnected(data=act1, num_hidden=50)
act2 = mx.sym.Activation(data=fc2, act_type="relu")
fc3 = mx.symbol.FullyConnected(data=act2, num_hidden=25)
act3 = mx.sym.Activation(data=fc3, act_type="relu")
fc4 = mx.symbol.FullyConnected(data=act3, num_hidden=10)
act4 = mx.sym.Activation(data=fc4, act_type="relu")
fc5 = mx.symbol.FullyConnected(data=act4, num_hidden=2)
lenet = mx.sym.SoftmaxOutput(data=fc5, name='softmax',normalization = 'batch')
# create iterator around training and validation data
train_iter = mx.io.NDArrayIter(data=data[:ntrain], label = phen[:ntrain],batch_size=batch_size, shuffle=True)
val_iter = mx.io.NDArrayIter(data=data[ntrain:], label=phen[ntrain:], batch_size=batch_size)
# create a trainable module on GPU 0
lenet_model = mx.mod.Module(symbol=lenet, context=mx.gpu())
# train with the same
lenet_model.fit(train_iter,
eval_data=val_iter,
optimizer='adam',
optimizer_params={'learning_rate':0.00001},
eval_metric='f1',
batch_end_callback = mx.callback.Speedometer(batch_size, 10),
num_epoch=1000)
This model performs well on the test set, so I want to keep it. Next, I save the network layout and the parameterization:
lenet.save('./testNet_symbol.mxnet')
lenet_model.save_params('./testNet_module.mxnet')
All the documentation I can find on loading the network seem to have implemented the save function within the training routine, to save the network parameters at the end of each epoch. I haven't set these checkpoints during the training process Other methods use the mx.model.FeedForward class, which doesn't seem appropriate. Still other methods load the network from a .json file, which I don't have as a result of my save functions. How can I save/load a network after it's already finished training?
You just have to do this instead to save:
lenet_model.save_checkpoint('lenet', num_epoch, save_optimizer_states=True)
This would create 3 files if the states flag is set to True else 2 files:
.params (weights),
.json (symbol),
.states
And this to load:
lenet_model = mx.mod.Module.load(prefix,epoch)
lenet_model.bind(for_training=False, data_shapes=[('data', (1,3,224,224))])
We are interested in using a surrogate model in an aircraft design process implemented in OpenMDAO. Basically we want to use an aerodynamic code (such as VSPaero in our aim) to produce a database (using a DOE ) and then built a surrogate that will be used in the design process. It looks like your proposal 2) in use of MOE in openMDAO and we also want to access to the "gradient" information of the surrogate to be used in the full design problem .
We started from the code you have provided in nested problem question and try to built a mock up case with simplified component for aerodynamic . The example code is below (using kriging) and we have two concerns to finish it:
we need to implement a "linearize" function in our component if we want to use surrogate gradient information: I guess we should use the "calc_gradient" function of problem to do this . Is it right ?
in our example code, the training will be done each time we call the component what is not very efficient : is there a way to call it only once or to do the surrogate training only after the setup() of the bigger problem (aircraft design in our case )?
Here is the code (sorry it is a bit long):
from openmdao.api import IndepVarComp, Group, Problem, ScipyOptimizer, ExecComp, DumpRecorder, Component, NLGaussSeidel,ScipyGMRES, Newton,SqliteRecorder,MetaModel, \
KrigingSurrogate, FloatKrigingSurrogate
from openmdao.drivers.latinhypercube_driver import LatinHypercubeDriver, OptimizedLatinHypercubeDriver
from openmdao.solvers.solver_base import NonLinearSolver
import numpy as np
import sys
alpha_test = np.array([0.56, 0.24, 0.30, 0.32, 0.20])
eta_test = np.array([-0.30, -0.14, -0.19, -0.18, -0.12])
num_elem = len(alpha_test)
class SysAeroSurrogate(Component):
""" Simulates the presence of an aero surrogate mode using linear aerodynamic model """
""" coming from pymission code """
""" https://github.com/OpenMDAO-Plugins/pyMission/blob/master/src/pyMission/aerodynamics.py """
def __init__(self, num_elem=1):
super(SysAeroSurrogate, self).__init__()
self.add_param('alpha', 0.5)
self.add_param('eta', -0.33)
self.add_param('AR', 0.0)
self.add_param('oswald', 0.0)
self.add_output('CL', val=0.0)
self.add_output('CD', val=0.0) ## Drag Coefficient
def solve_nonlinear(self, params, unknowns, resids):
""" Compute lift and drag coefficient using angle of attack and tail
rotation angles. Linear aerodynamics is assumed."""
alpha = params['alpha']
eta = params['eta']
aspect_ratio = params['AR']
oswald = params['oswald']
lift_c0 = 0.30
lift_ca = 6.00
lift_ce = 0.27
drag_c0 = 0.015
unknowns['CL'] = lift_c0 + lift_ca*alpha*1e-1 + lift_ce*eta*1e-1
unknowns['CD'] = (drag_c0 + (unknowns['CL'])**2 /(np.pi * aspect_ratio * oswald))/1e-1
class SuroMM(Group):
def __init__(self):
super(SuroMM, self).__init__()
#kriging
AeroMM = self.add("AeroMM", MetaModel())
AeroMM.add_param('alpha', val=0.)
AeroMM.add_param('eta', val=0.)
AeroMM.add_output('CL_MM', val=0., surrogate=FloatKrigingSurrogate())
AeroMM.add_output('CD_MM', val=0., surrogate=FloatKrigingSurrogate())
class SurrogateAero(Component):
def __init__(self):
super(SurrogateAero, self).__init__()
## Inputs to this subprob
self.add_param('alpha', val=0.5*np.ones(num_elem)) ## Angle of attack
self.add_param('eta', val=0.5*np.ones(num_elem)) ## Tail rotation angle
self.add_param('AR', 0.0)
self.add_param('oswald', 0.0)
## Unknowns for this sub prob
self.add_output('CD', val=np.zeros(num_elem))
self.add_output('CL', val=np.zeros(num_elem))
#####
self.problem = prob = Problem()
prob.root = Group()
prob.root.add('d1', SuroMM(), promotes=['*'])
prob.setup()
#### training of metamodel
prob['AeroMM.train:alpha'] = DOEX1
prob['AeroMM.train:eta'] = DOEX2
prob['AeroMM.train:CL_MM'] = DOEY1
prob['AeroMM.train:CD_MM'] =DOEY2
def solve_nonlinear(self, params, unknowns, resids):
CL_temp=np.zeros(num_elem)
CD_temp=np.zeros(num_elem)
prob = self.problem
# Pass values into our problem
for i in range(len(params['alpha'])):
prob['AeroMM.alpha'] = params['alpha'][i]
prob['AeroMM.eta'] = params['eta'][i]
# Run problem
prob.run()
CL_temp[i] = prob['AeroMM.CL_MM']
CD_temp[i] = prob['AeroMM.CD_MM']
# Pull values from problem
unknowns['CL'] = CL_temp
unknowns['CD'] = CD_temp
if __name__ == "__main__":
###### creation of database with DOE #####
top = Problem()
root = top.root = Group()
root.add('comp', SysAeroSurrogate(), promotes=['*'])
root.add('p1', IndepVarComp('alpha', val=0.50), promotes=['*'])
root.add('p2', IndepVarComp('eta',val=0.50), promotes=['*'])
root.add('p3', IndepVarComp('AR', 10.), promotes=['*'])
root.add('p4', IndepVarComp('oswald', 0.92), promotes=['*'])
top.driver = OptimizedLatinHypercubeDriver(num_samples=16, seed=0, population=20, generations=4, norm_method=2)
top.driver.add_desvar('alpha', lower=-5.0*(np.pi/180.0)*1e-1, upper=15.0*(np.pi/180.0)*1e-1)
top.driver.add_desvar('eta', lower=-5.0*(np.pi/180.0)*1e-1, upper=15.0*(np.pi/180.0)*1e-1)
top.driver.add_objective('CD')
recorder = SqliteRecorder('Aero')
recorder.options['record_params'] = True
recorder.options['record_unknowns'] = True
recorder.options['record_resids'] = False
recorder.options['record_metadata'] = False
top.driver.add_recorder(recorder)
top.setup()
top.run()
import sqlitedict
db = sqlitedict.SqliteDict( 'Aero', 'openmdao' )
print( list( db.keys() ) )
DOEX1 = []
DOEX2 = []
DOEY1 = []
DOEY2 = []
for i in list(db.keys()):
data = db[i]
p = data['Parameters']
DOEX1.append(p['comp.alpha'])
DOEX2.append(p['comp.eta'])
p = data['Unknowns']
DOEY1.append(p['CL'])
DOEY2.append(p['CD'])
################ use of surrogate model ######
prob2 = Problem(root=Group())
prob2.root.add('SurrAero', SurrogateAero(), promotes=['*'])
prob2.root.add('v1', IndepVarComp('alpha', val=alpha_test), promotes=['*'])
prob2.root.add('v2', IndepVarComp('eta',val=eta_test), promotes=['*'])
prob2.setup()
prob2.run()
print'CL predicted:', prob2['CL']
print'CD predicted:', prob2['CD']
The way you have your model set up seems correct. The MetaModel component will only train its data one time (the first pass through the model), as you can see in this part of the source code. Every subsequent iteration, it just uses the trained surrogate thats already there.
The meta-model is also already setup to provide analytic derivatives of the predicted output with respect to the input independent variables. Derivatives of the prediction with respect to the training point values are not available in the base implementation. That requires a more complex setup that, at least for the moment, will require some custom setup that is not in the standard library.