amirmardan / pyfwi Goto Github PK

This repository is for PyFWI, a Python package for seismic FWI and reservoir monitoring (time-lapse FWI)

Home Page: https://pyfwi.readthedocs.io/en/latest/

License: GNU General Public License v3.0

Python 28.25% C 3.64% Jupyter Notebook 68.11%

seismic full-waveform-inversion fwi tl-fwi wave-modelling monitoring seismic-inversion seismic-monitoring time-lapse time-lapse-imaging

pyfwi's Introduction

Hi there 👋

I'm Amir, a postdoctoral researcher at Polytechnique Montréal working on applications of AI for processing and inversion of near-surface seismic data. I'm based in Quebec City, Canada.

🔭 I’m currently working on

processing near-surface seismic data using AI,
velocity modelling using deep learning,
velocity modelling using FWI.

📚 These days, I'm mostly reading about

self-supervised learning,
uncertainty analysis and decision-making under uncertainty.

You can see my research at my personal page

📫 How to reach me: [email protected]

My main public repositories

Repository	Description
PyFWI	A Python package for performing FWI and TL-FWI
pinn_fwi	An implementation of PINN for performing FWI
ml_course	Prepared materials for my machine learning course at AUT
first_break_picking	A Python package for performing automatic first-break picking

⚡ Technologies

pyfwi's People

Contributors

Stargazers

Watchers

Forkers

1902912604 csupzw zy911k24 elseviersoftwarex jedfang rickli92 omid-b zj15001 yingchangliu doziej84

pyfwi's Issues

example

Hi Amir, thanks a lot for the work you do.

Could you please provide an example of the marmousi2 model? I'm interested in understanding its application in seismic imaging.

PyFWI on Windows

Hello, can this tool be used on Windows?

Issue in the example code

Hi Amir, In the example code:

# Create the wave object
W = wave.WavePropagator(inpa, src, rec_loc, model_shape,
                        n_well_rec=n_well_rec,
                        components=seisout, chpr=0)

# Call the forward modelling 
d_obs = W.forward_modeling(model, show=False)  # show=True can show the propagation of the wave

When I run the code, I keep receiving this error:

Cell In[6], line 7
      2 W = wave.WavePropagator(inpa, src, rec_loc, model_shape,
      3                         n_well_rec=n_well_rec,
      4                         components=seisout, chpr=0)
      6 # Call the forward modelling 
----> 7 d_obs = W.forward_modeling(model, show=False)  # show=True can show the propagation of the wave

File ~/anaconda3/envs/env3/lib/python3.10/site-packages/PyFWI/wave_propagation.py:898, in WavePropagator.forward_modeling(self, model0, show)
    896 self.pml_preparation(model['vp'].max())
    897 self.elastic_buffers(model)
--> 898 seismo = self.forward_propagator(model)
    899 data = acq.seismic_section(seismo, self.components, shape=self.seismo_shape)
    900 return data

File ~/anaconda3/envs/env3/lib/python3.10/site-packages/PyFWI/wave_propagation.py:648, in WavePropagator.forward_propagator(self, model)
    642     self.prg.MakeAllZero(self.queue, (self.tnz, self.tnx), None,
    643                          self.vx_b, self.vz_b,
    644                          self.taux_b, self.tauz_b, self.tauxz_b
    645                          )
    646     # print(s)
    647     # print(coeff)
--> 648     self.__kernel(s, coeff)
    650 if self.acq_type == 2:
    651     for par in self.seismogram:

File ~/anaconda3/envs/env3/lib/python3.10/site-packages/PyFWI/wave_propagation.py:714, in WavePropagator.__kernel(self, s, coeff)
    712 for t in np.arange(self.nt):
    713     if t % self.dt_scale == 0:
--> 714         self.__injSrc(t_src, s)
    715         # t_src += 1.0
    716     print(t)

File ~/anaconda3/envs/env3/lib/python3.10/site-packages/PyFWI/wave_propagation.py:661, in WavePropagator.__injSrc(self, t, s)
    656 src_kv_x, src_kv_z, src_kt_x, src_kt_z, src_kt_xz = np.float32(self.src(t))
    657 # print(t.astype(np.float32).type)
    658 # print(t)
    659 # print(type(src_kt_xz))
    660 # t = np.float32(t)
--> 661 self.prg.injSrc(self.queue, (self.tnz, self.tnx), None,
    662                 self.vx_b, self.vz_b,
    663                     self.taux_b, self.tauz_b, self.tauxz_b,
    664                 self.seismogramid_vx_b, self.seismogramid_vz_b,
    665                 self.seismogramid_taux_b, self.seismogramid_tauz_b, self.seismogramid_tauxz_b,
    666                 self.dxr,
    667                 self.srcx[s], self.srcz[s],
    668                 src_kv_x, src_kv_z,
    669                 src_kt_x, src_kt_z, 
    670                 t
    671                 )

File ~/anaconda3/envs/env3/lib/python3.10/site-packages/pyopencl/__init__.py:901, in _add_functionality.<locals>.kernel_call(self, queue, global_size, local_size, *args, **kwargs)
    895 def kernel_call(self, queue, global_size, local_size, *args, **kwargs):
    896     # __call__ can't be overridden directly, so we need this
    897     # trampoline hack.
    898 
    899     # Note: This is only used for the generic __call__, before
    900     # kernel_set_scalar_arg_dtypes is called.
--> 901     return self._enqueue(self, queue, global_size, local_size, *args, **kwargs)

File <pyopencl invoker for 'injSrc'>:7, in enqueue_knl_injSrc(self, queue, global_size, local_size, arg0, arg1, arg2, arg3, arg4, arg5, arg6, arg7, arg8, arg9, arg10, arg11, arg12, arg13, arg14, arg15, arg16, arg17, global_offset, g_times_l, allow_empty_ndrange, wait_for)

LogicError: clSetKernelArg failed: INVALID_ARG_SIZE - when processing arg#18 (1-based):

Tried to run this but ran into problems. I'm on a M1 MacOS laptop. I tried to run it via both standard m1 modes and Rosetta mode to mimic Intel-based mac behavior.

The error I get (not doing any Rosetta fanciness) is

(pyfwi) JustinGosses_per:PyFWI-test justingosses$ pip install PyFWI
Collecting PyFWI
  Using cached PyFWI-0.1.7-py3-none-any.whl (68 kB)
Collecting datetime
  Downloading DateTime-4.5-py2.py3-none-any.whl (52 kB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 52.0/52.0 kB 1.3 MB/s eta 0:00:00
Collecting scipy
  Using cached scipy-1.8.1-cp310-cp310-macosx_12_0_arm64.whl (28.7 MB)
Collecting segyio
  Using cached segyio-1.9.6.tar.gz (1.1 MB)
  Preparing metadata (setup.py) ... error
  error: subprocess-exited-with-error
  
  × python setup.py egg_info did not run successfully.
  │ exit code: 1
  ╰─> [6 lines of output]
      Traceback (most recent call last):
        File "<string>", line 2, in <module>
        File "<pip-setuptools-caller>", line 34, in <module>
        File "/private/var/folders/ry/tr7t3nh122z_0n6rz2cl3lv40000gn/T/pip-install-guz0ovaa/segyio_20210f3369a9411ba04344ec9a238dd0/setup.py", line 3, in <module>
          import skbuild
      ModuleNotFoundError: No module named 'skbuild'
      [end of output]
  
  note: This error originates from a subprocess, and is likely not a problem with pip.
error: metadata-generation-failed

× Encountered error while generating package metadata.
╰─> See above for output.

note: This is an issue with the package mentioned above, not pip.
hint: See above for details.

Googled this error and seems to be related to errors others have had with segyio. Maybe someone with experience with segyio would be able to help?

Originally posted by @JustinGOSSES in softwareunderground/awesome-open-geoscience#179 (comment)

error while running example.py

Hi Amir,
As your suggestion for adding argument "set_env_variable=False" in wave.WavePropagator, then I got this error

PS C:\Users\ASUS\Desktop\bismillah> & C:/Users/ASUS/AppData/Local/Microsoft/WindowsApps/python3.10.exe c:/Users/ASUS/Desktop/bismillah/example.py
Device 0 is chosen.
Choose platform:
[0] <pyopencl.Platform 'Intel(R) OpenCL HD Graphics' at 0x1cf38ace6a0>
[1] <pyopencl.Platform 'NVIDIA CUDA' at 0x1cf38ae0a20>
[2] <pyopencl.Platform 'Intel(R) OpenCL' at 0x1cf2e529808>
[3] <pyopencl.Platform 'Intel(R) FPGA Emulation Platform for OpenCL(TM)' at 0x1cf2e52abf8>
[4] <pyopencl.Platform 'Intel(R) FPGA SDK for OpenCL(TM)' at 0x7ffdeca44f20>
Choice [0]:1
Set the environment variable PYOPENCL_CTX='1' to avoid being asked again.
Device 0 is chosen.
Choose platform:
[0] <pyopencl.Platform 'Intel(R) OpenCL HD Graphics' at 0x1cf38ace6a0>
[1] <pyopencl.Platform 'NVIDIA CUDA' at 0x1cf38ae0a20>
[2] <pyopencl.Platform 'Intel(R) OpenCL' at 0x1cf2e529808>
[3] <pyopencl.Platform 'Intel(R) FPGA Emulation Platform for OpenCL(TM)' at 0x1cf2e52abf8>
[4] <pyopencl.Platform 'Intel(R) FPGA SDK for OpenCL(TM)' at 0x7ffdeca44f20>
    self.__kernel(s, coeff)
  File "C:\Users\ASUS\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.10_qbz5n2kfra8p0\LocalCache\local-packages\Python310\site-packages\PyFWI\wave_propagation.py", line 712, in __kernel
    cl.enqueue_copy(self.queue, copy_purpose, self.vx_b)
  File "C:\Users\ASUS\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.10_qbz5n2kfra8p0\LocalCache\local-packages\Python310\site-packages\pyopencl\__init__.py", line 2009, in enqueue_copy    return _cl._enqueue_read_buffer(queue, src, dest, **kwargs)
pyopencl._cl.RuntimeError: clEnqueueReadBuffer failed: OUT_OF_RESOURCES

Implementation on circular geometry

Discussed in #6

^{Originally posted by makeabhishek March 21, 2023}
Hi, This is a discussion rather than an issue. I want to know if it is possible to implement the code for circular geometry. Is any example available?

Any answers and suggestions would be appreciated.

Consider sources and geophones all around the model.

fn shouldn't be required in FWI

User gets the following error for running FWI.

self.fn = inpa['fn']
KeyError: 'fn'