Table 1 Data comparison of PINN predicted values for cavity with Ref.11 at \(Ra=10^{3}\), \(10^{4}\), \(10^{5}\), and \(10^{6}\). The order of magnitude of the MSE loss and the runtime are also tabulated.

From: Physics-informed neural network simulation of thermal cavity flow

Parameter

\(Ra=10^3\)

\(Ra=10^4\)

\(Ra=10^5\)

\(Ra=10^6\)

Nu

a

b

c

d

\(|\psi |_{ref}\) location (xz)

(0.5, 0.5)

(0.5, 0.5)

(0.5, 0.5)

(0.5, 0.5)

\(|\psi |_{ref}\)

1.174

5.071

9.111

16.32

\(|\psi |_{pred}\)

1.173

5.065

9.116

16.38

\(u_{ref}\) location (xz)

(0.5, 0.813)

(0.5, 0.823)

(0.5, 0.855)

(0.5, 0.850)

\(u_{ref}\)

3.649

16.178

34.73

64.63

\(u_{pred}\)

3.643

16.135

34.74

64.81

\(w_{ref}\) location (xz)

(0.178, 0.5)

(0.119, 0.5)

(0.066, 0.5)

(0.0379, 0.5)

\(w_{ref}\)

3.697

19.617

68.59

219.36

\(w_{pred}\)

3.691

19.590

68.64

220.52

Loss order of magnitude

\(10^{-4}\)

\(10^{-4}\)

\(10^{-7}\)

\(10^{-6}\)

Runtime

24 min

24 min

4 h 2 min

3 h 55 min