deepxde
deepxde copied to clipboard
how to get the 'variables' of new predictions using the trained model
Hello, LuLu
I read your paper and tried your code for apoptosis. During training, I was able to get the 'variables' of the ode model well. However, I would like to know how to get the 'variables' of new predictions using the trained model. I couldn't find it in the deepxde FAQ.
I don't get your question.
I don't get your question.
Oh....I'm sorry for asking a question without a detailed explanation. The 'variable' I was talking about was a parameter of the ode system and meant a trainable variable (in Lorenz system C1, C2, C3 ). In the inverse problem, I could verify the ode parameter ('C1, C2, C3') through callback during training. However, what I want is a way to get the ode parameters (like 'C1, C2, C3') of a new prediction using an already trained model. Please let me know if this is possible and, if so, how to do it.
https://deepxde.readthedocs.io/en/latest/demos/pinn_inverse/lorenz.inverse.html
Hello @DonnaShin , please can share your code with me ? I have a first order nonlinear ODE and I want to infer one parameter A and I tried with many different hyperparameters and still the results are unstable and wrong and I am stuck now. I used the Lorenz inverse problem as the basis of the code. def Emsley_system(t, DP):
k=A*(tf.math.exp(-E / (R * (T+273))))
dy_x = dde.grad.jacobian(DP, t, i=0)
return [
dy_x + k * (DP**2),
] and I am trying the infer the value of A. The true value of A is 7.8*(10**8) and the time domain according to the data that I have is between 1 and 2000 days and DP value range from 1000 to 200 , I tried to include 10^8 in the expression of the ODE and use the log(A) instead of A and scale the time domain between 0 and 5 but still I have MSE in the range of 10^4 and wrong A value.
At this moment, there is no existing API for this. But you can look at the source code of dde.callbacks.VariableValue
, and you can find the code there.
At this moment, there is no existing API for this. But you can look at the source code of
dde.callbacks.VariableValue
, and you can find the code there.
Understood. Thank you for providing a good library.