ar4 / deepwave

Wave propagation modules for PyTorch.
MIT License
209 stars 47 forks source link

question #79

Open yulongovo opened 2 weeks ago

yulongovo commented 2 weeks ago

Hello, I would like to know if the source is loaded to Vz when P/SV simulation?

ar4 commented 2 weeks ago

Hello and thank you for your question.

Sources for the elastic propagator that you specify using source_amplitudes_y and source_locations_y will be added to the velocity in the first spatial dimension of your model, while sources that you specify using source_amplitudes_x and source_locations_x will be added to the velocity in the second spatial dimension. Whether the first or second spatial dimension corresponds to depth depends on how you orient your model. The usage documentation provides more details, but please let me know if it is still not clear.

When using the scalar wave propagator, the source is added to the wavefield as a monopole source. The description of what Deepwave calculates provides more details.

yulongovo commented 2 weeks ago

Have you developed a program to estimate wavelet?

ar4 commented 2 weeks ago

Gradients backpropagate into the source amplitudes in Deepwave, so you can optimise/invert for them (as in this example). I do not have any other program to estimate wavelets. Note that since Deepwave currently only supports 2D propagation, inverting real data may be difficult.

yulongovo commented 2 weeks ago

Thank you very much!

一场空 @.***

 

------------------ 原始邮件 ------------------ 发件人: "ar4/deepwave" @.>; 发送时间: 2024年10月17日(星期四) 晚上11:28 @.>; @.**@.>; 主题: Re: [ar4/deepwave] question (Issue #79)

Gradients backpropagate into the source amplitudes in Deepwave, so you can optimise/invert for them (as in this example). I do not have any other program to estimate wavelets. Note that since Deepwave currently only supports 2D propagation, inverting real data may be difficult.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>