FAQ

If you have any questions about DeepXDE, first read the papers/slides and watch the video at the DeepXDE homepage and also check the following list of frequently asked DeepXDE questions. To get further help, you can open a discussion in the GitHub Discussions.

General usage

• Q: DeepXDE failed to run.

  A: #5
• Q: By default, DeepXDE uses float32. How can I use float64?

  A: #28
• Q: I want to set the global random seeds.

  A: #353
• Q: GPU.

  A: #464
• Q: How can I use a trained model for new predictions?

  A: #669, #676, #93, #177
• Q: How can I save a trained model and then load the model later?

  A: #54, #57, #58, #63, #103, #206, #254, #468
• Q: More details about DeepXDE source code, and want to modify DeepXDE.

  A: #68, #346

PINN

• Q: What is the output of DeepXDE? How can I visualize the results?

  A: #668, #9, #675, #48, #53, #73, #77, #171, #217, #218, #223, #274, #276
• Q: More details and examples about geometry.

  A: #32, #38, #161, #264, #278, #332
• Q: How can I implement new ODEs/PDEs, e.g., compute derivatives, complicated PDEs, complicated coefficients?

  A: #670, #21, #22, #74, #78, #79, #124, #172, #185, #193, #194, #246, #302, #377, #421, #451, #465, #478, #480
• Q: More details and examples about initial conditions.

  A: #677, #75, #104, #134, #398, #485
• Q: More details and examples about boundary conditions.

  A: #6, #669, #673, #674, #22, #26, #33, #38, #40, #44, #49, #115, #140, #156, #352, #365, #457, #475
• Q: By default, initial/boundary conditions are enforced in DeepXDE as soft constraints. How can I enforce them as hard constraints?

  A: #36, #90, #92, #252
• Q: How can I use a dataset of the solution?

  A: #362, #487
• Q: I failed to train the network or get the right solution, e.g., large training loss, unbalanced losses.

  A: #673, #22, #33, #41, #61, #62, #80, #84, #85, #108, #126, #141, #188, #247, #305, #321
• Q: L-BFGS.

  A: #364, #395
• Q: Implement certain features for the input, such as Fourier features.

  A: #277, #477
• Q: A standard network doesn’t work for me, and I want to implement a network with a special structure/property.

  A: #355, #363, #466
• Q: Implement new losses or constraints.

  A: #174, #286, #311, #356, #460, #467, #504
• Q: How can I implement new IDEs?

  A: #95, #198
• Q: Solve PDEs with complex numbers.

  A: #284
• Q: Solve inverse problems with unknown parameters/fields in the PDEs or initial/boundary conditions.

  A: #55, #76, #86, #114, #120, #125, #178, #208, #235, #380
• Q: Solve parametric PDEs.

  A: #273, #299, #379, #482
• Q: How does DeepXDE choose the training points? How can I use some specific training points?

  A: #32, #57, #64, #345
• Q: How can I give different weights to different residual points?

  A: #45
• Q: Residual-based adaptive refinement (RAR).

  A: #63
• Q: I want to customize network training/optimization, e.g., mini-batch.

  A: #166, #307, #320, #331
• Q: I want to get more information about the network, such as the values and gradients of hidden layers.

  A: #149, #490
• Q: More details about DeepXDE source code, or I want to modify DeepXDE.

  A: #35, #39, #66, #69, #91, #99, #131, #163, #175, #202, #357
• Q: Examples collected from users.

  A: Lotka–Volterra, Potential flow around a cylinder, Laminar Incompressible flow passing a step, Shallow water equations, 2D time-dependent heat equation

DeepONet

• Q:

  A:

Multi-fidelity NN

• Q: How is MFNN trained?

  A: #94, #450
• Q: Transfer learning of MFNN.

  A: #195
• Q: How can I use a trained model for new predictions?

  A: #324