Replication Data for: Super-resolution reconstruction of scalar fields from the pyrolysis of pulverised biomass using deep learning
收藏DataCite Commons2025-11-14 更新2026-05-07 收录
下载链接：
https://darus.uni-stuttgart.de/citation?persistentId=doi:10.18419/DARUS-5519
下载链接
链接失效反馈官方服务：
资源简介：
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8"/>
<meta name="viewport" content="width=device-width, initial-scale=1"/>
<title>README</title>
<style>
body { font-family: system-ui, -apple-system, Segoe UI, Roboto, Helvetica, Arial, sans-serif; line-height: 1.5; padding: 1rem; max-width: 900px; margin: auto; }
code { font-family: ui-monospace, SFMono-Regular, Menlo, Consolas, "Liberation Mono", monospace; }
pre { background: #f6f8fa; padding: 0.75rem; overflow: auto; border-radius: 6px; }
table { border-collapse: collapse; margin: 1rem 0; width: 100%; }
th, td { border: 1px solid #ddd; padding: 0.5rem; text-align: left; }
thead th { background: #fafbfc; }
h1, h2, h3, h4, h5, h6 { line-height: 1.25; }
</style>
</head>
<body>
<h1>Repository for publication: A. Shamooni et al., Super-resolution reconstruction of scalar fields from the pyrolysis of pulverised biomass using deep learning, Proc. Combust. Inst. (2025)</h1>
<h1>Containing</h1>
<h2>torch_code</h2>
<p>
The main Pytorch source code used for training/testing is provided in <em>torch_code.tar.gz</em> file. <br/>
</p>
<h2>torch_code_tradGAN</h2>
<p>
To compare with traditional GAN, we use the code in torch_code_tradGAN with similar particle-laden datasets. The source code is <em>torch_code_tradGAN.tar.gz</em> file.<br/>
</p>
<h2>datasets</h2>
<p>
The training/validation/testing datasets have been provided in lmdb format which is ready to use in the code. The datasets in <em>datasets.tar.gz</em> contain:<br/>
</p>
<ul>
<li><p>Training dataset:</p>
<p>data_train_OF-mass_kinematics_mk0x_1x_2x_FHIT_particle_128_Re52-2D_20736_lmdb.lmdb </p>
</li>
<li><p>Test dataset:</p>
<p>data_valid_inSample_OF-mass_kinematics_mk0x_1x_2x_FHIT_particle_128_Re52-2D_3456_lmdb.lmdb </p>
</li>
</ul>
</li>
</ul>
<p>Note that the samples from 9 DNS cases are collected in order (each case 2304 samples for training and 384 samples for testing) which can be recognized using the provided metadata file in each folder. </p>
<ul>
<li><p>Out of distribution test datasets: </p>
<ul>
<li><p>Out of distribution test dataset (used in Fig 10 of the paper):</p>
<p>data_valid_inSample_OF-mass_kinematics_mk3x_FHIT_particle_128_Re52-2D_nonUniform_1024_lmdb.lmdb |</p>
</li>
</ul>
</li>
</ul>
<p>We have two separate OOD DNS cases and from each we select 512 samples. </p>
<h2 id="experiments">experiments</h2>
<p>The main trained models are provided in <em>experiments.tar.gz</em> file. Each experiment contains the log file of the training, the last training state (for restart) and the model wights used in the publication. </p>
<ul>
<li><p>Trained model using the main dataset (used in Figs 2-10 of the paper):</p>
<p>h_oldOrder_mk_700-11-c_PFT_Inp4TrZk_outTrZ_RRDBNetCBAM-4Prt_DcondPrtWav_f128g64b16_BS16x4_LrG45D5_DS-mk012-20k_LStandLog</p>
</li>
</ul>
</code></pre><p>To compare with traditional GAN, we use the code in torch_code_tradGAN with similar particle-laden datasets as above. The training consists of one pre-training step and two separate fine-tuning. One fine-tuning with the loss weights from the litreature and one fine-tuning with tuned loss weights. The final results are in experiments/trad_GAN/experiments/</p>
<ul>
<li><p>Pre-trained traditional GAN model (used in Figs 8-9 of the paper):</p>
<p>train_RRDB_SRx4_particle_PSNR </p>
</li>
<li><p>Fine-tuned traditional GAN model with loss weights from lit. (used in Figs 8-9 of the paper) </p>
<p>train_ESRGAN_SRx4_particle_Nista_oneBlock </p>
</li>
<li><p>Fine-tuned traditional GAN model with optimized loss weights (used in Figs 8-9 of the paper) </p>
<p>train_ESRGAN_SRx4_particle_oneBlock_betaA </p>
</li>
</ul>
<h2 id="inference_notebooks">inference_notebooks</h2>
<p>The inference_notebooks folder contains example notebooks to do inference. The folder contains &quot;torch_code_inference&quot; and &quot;torch_code_tradGAN_inference&quot;. The &quot;torch_code_inference&quot; is the inference of main trained model. The &quot;torch_code_tradGAN_inference&quot; is the inference for traditional GAN approach.
Move the inference folders in each of these folders into the corresponding torch_code roots. Also create softlinks of datasets and experiments in the main torch_code roots. Note that in each notebook you must double check the required paths to make sure they are set correctly. </p>
</p>
<h1>How to</h1>
<h2>Build the environment</h2>
<p>
To build the environment required for the training and inference you need Anaconda. Go to the torch_code folder and <br/>
</p>
<pre><code class="language-bash">
conda env create -f environment.yml
</code></pre>
<p>
Then create ipython kernel for post processing, <br/>
</p>
<pre><code class="language-bash">
conda activate torch_22_2025_Shamooni_PCI
python -m ipykernel install --user --name ipyk_torch_22_2025_Shamooni_PCI --display-name &quot;ipython kernel for post processing of PCI2025&quot;
</code></pre>
<h2>Perform training</h2>
<p>
It is suggested to create softlinks to the dataset folder directly in the torch_code folder:<br/>
</p>
<pre><code class="language-bash">
cd torch_code
ln -s &lt;path to the dataset folder&gt; datasets
</code></pre>
<p>
You can also simply move the datasets and inference forlders in the torch_code folder beside the cfd_sr folder and other files. <br/>
</p>
<p>
In general, we prefer to have a root structure as below:<br/>
</p>
<p>
root files and directories:<br/>
</p>
<p>
cfd_sr<br/>
datasets<br/>
experiments<br/>
inference<br/>
options<br/>
__init__.py<br/>
test.py<br/>
train.py <br/>
version.py<br/>
</p>
<p>
Then activate the conda environment <br/>
</p>
<pre><code class="language-bash">
conda activate torch_22_2025_Shamooni_PCI
</code></pre>
<p>
An example script to run on single node with 2 GPUs:<br/>
</p>
<pre><code class="language-bash">
torchrun --standalone --nnodes=1 --nproc_per_node=2 train.py -opt options/train/condSRGAN/use_h_mk_700-011_PFT.yml --launcher pytorch
</code></pre>
<p>
<em>Make sure that the paths to datasets &quot;dataroot_gt&quot; and &quot;meta_info_file&quot; for both training and validation data in option files are set correctly.</em> <br/>
</p>
</body>
</html>
提供机构：
DaRUS
创建时间：
2025-11-06
