Visual Analysis of Large Multi-Field AMR Data on GPUs Using Interactive Volume Lines

Stefan Zellmann,Serkan Demirci,Uğur Güdükbay

2023-06-20

Abstract:To visually compare ensembles of volumes, dynamic volume lines (DVLs) represent each ensemble member as a 1D polyline. To compute these, the volume cells are sorted on a space-filling curve and scaled by the ensemble's local variation. The resulting 1D plot can augment or serve as an alternative to a 3D volume visualization free of visual clutter and occlusion. Interactively computing DVLs is challenging when the data is large, and the volume grid is not structured/regular, as is often the case with computational fluid dynamics simulations. We extend DVLs to support large-scale, multi-field adaptive mesh refinement (AMR) data that can be explored interactively. Our GPU-based system updates the DVL representation whenever the data or the alpha transfer function changes. We demonstrate and evaluate our interactive prototype using large AMR volumes from astrophysics simulations.

Graphics,Information Retrieval,Performance

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is: how to efficiently implement interactive Dynamic Volume Lines (DVLs) on GPU to support the visualization and analysis of large - scale, multi - field Adaptive Mesh Refinement (AMR) data. Specifically, the paper addresses the following key issues:

1. **Extending DVLs to support unstructured or hierarchical grid - type data**:
- Traditional DVL methods are mainly for smaller, structured regular volume data sets. When the data scale increases and the grid is unstructured or hierarchical, calculating DVL becomes challenging. The paper proposes a method to handle AMR data, where the size of each cell depends on its refinement level.

2. **Improving computational performance and interactivity**:
- When the data volume is large and the grid is irregular, real - time updating of DVL is a difficult problem. The paper achieves efficient DVL calculation through GPU acceleration and allows users to edit transfer functions in real - time, thus improving interactivity.

3. **Handling the impact of transfer function updates on spatial arrangement**:
- The update of the transfer function will affect the spatial arrangement of the volume set and its local changes. The paper proposes a method that can quickly recalculate the DVL representation when the user updates the transfer function.

4. **Achieving the linkage between 3D views and 1D graphs**:
- The paper develops an application that allows users to interact between 3D views and 1D graphs by brushing and linking. For example, a user can select an area of interest in the 1D graph, and the corresponding 3D cells will be highlighted in the 3D view.

### Formula Explanation

To understand the technical details in the paper more clearly, the following are several key formulas and their explanations:

- **Local change calculation**:
\[
V_h=\max_{\forall m\in M}I(m, h)-\min_{\forall m\in M}I(m, h)
\]
where \( h \) is the Hilbert code of the cell, \( M \) is the volume set, and \( I(m, h) \) is the intensity of the set member \( m \) and the AMR cell corresponding to \( h \).

- **Importance scaling factor**:
\[
f(h)=\frac{V_h}{\max V_h\cdot2^{L_h}}^P
\]
where \( L_h\in\mathbb{N}_0 \) is the AMR level corresponding to the Hilbert code \( h \), and \( P \) is a user - defined parameter used to control the steepness of the resulting curve.

- **Cumulative importance calculation**:
\[
F(h)=\sum_{i = 0}^{h}f(i)
\]

Through these improvements, the paper successfully extends the DVL method to large - scale AMR data and achieves efficient GPU acceleration, enabling users to explore complex scientific data in an interactive environment.