Skip to main content

Table 2 CAE-space architecture for Burger’s and Stoker’s test cases

From: Reduced-order modeling for stochastic large-scale and time-dependent flow problems using deep spatial and temporal convolutional autoencoders

Layer

Nb of filters

Kernel size

Activation function

Output shape

Encoder-space

    

Input

–

–

–

\(1000\times 1\)

Conv-pool

32

3–2

PReLU

\(500\times 32\)

Conv-pool

64

3–2

PReLU

\(250\times 64\)

Conv-pool

128

3–5

PReLU

\(50\times 128\)

Flatten

–

–

–

6400

Dense

–

–

PReLU

60

Dense (output)

-

-

PReLU

\(L_{x}=50\)

Decoder-space

    

Input

–

–

–

\(L_{x}=50\)

Dense

–

–

PReLU

60

Dense

–

–

PReLU

6400

Reshape

–

–

–

\(50\times 128\)

Conv-Upsamp

128

3–5

PReLU

\(250\times 128\)

Conv-Upsamp

64

3–2

PReLU

\(500\times 64\)

Conv-Upsamp

32

3–2

PReLU

\(1000\times 32\)

Conv (output)

1

3

PReLU

\(1000\times 1\)