Skip to content

Fomo.jl Benchmark

English

Deepwave’s forward modeling module is highly efficient and offers excellent accuracy. It supports scalar acoustic, acoustic, and elastic wave equations (using a staggered grid + PML).
I conducted a benchmark comparing my Julia package, Fomo.jl, against both Deepwave and Specfem2D. The results are shown below:
Fomo.jl

Elastic Wave Module Comparison Shot Gather Comparison

Benchmark Results Shots
Benchmark Results Shots

Waveform/Trace Comparison

Benchmark Results Traces
Benchmark Results Traces

Computational Time Comparison

DeepwaveFomo.jlSpecfem2D
Runtime per shot0.205s0.381s10.3s

中文

Deepwave的正演模块效率很高,准确度也不错,支持标量声波方程、声波方程还有弹性波方程(staggered grid + PML)。
我把我的julia包Fomo.jl和他还有specfem2D做过一个Benchmark,结果如下:
Fomo.jl

弹性波模块对比 炮集对比

Benchmark Results Shots
Benchmark Results Shots

波形对比

Benchmark Results Traces
Benchmark Results Traces

时间对比

DeepwaveFomo.jlSpecfem2D
单炮运行时间0.205s0.381s10.3s

从AE到VQ-VAE——模型的矛盾
How Much Information Does a Shot Gather Contain? / 一个共炮点道集蕴含了多少信息