*with memory mapping enabled **peak working set
Abstract — STM files, commonly associated with scanning tunneling microscopy (STM) data or proprietary structured metadata formats, lack standardized open-source visualization tools. This paper presents the design and implementation of a dedicated STM file viewer capable of parsing binary and text-based STM variants, reconstructing 2D/3D topographic data, and providing interactive analysis. We detail the file format reverse-engineering process, a modular Python-based architecture using PyQt5 and OpenGL , and performance benchmarks on real-world microscopy datasets. The viewer achieves sub-second loading for files up to 500 MB and offers surface profiling, color mapping, and export functionality. Our work enables reproducible nanoscience research by democratizing access to STM data. stm file viewer
| Dataset | Dimensions | Size (MB) | Load time (s) | Memory (MB) | 3D FPS | |---------|------------|-----------|---------------|-------------|---------| | Au(111) surface | 1024×1024 | 4.2 | 0.28 | 18 | 60 | | Graphene on SiO₂ | 2048×2048 | 16.8 | 0.91 | 42 | 58 | | High-res Si(111) | 8192×8192 | 268.4 | 1.32* | 86** | 42 | *with memory mapping enabled **peak working set Abstract
:
def parse_binary_stm(filepath): with open(filepath, 'rb') as f: magic = struct.unpack('<I', f.read(4))[0] if magic != 0x4D5453: raise ValueError("Not an STM file") w, h = struct.unpack('<II', f.read(8)) xstep, ystep = struct.unpack('<ff', f.read(8)) bdepth = struct.unpack('<H', f.read(2))[0] f.read(2) # endianness flag data = np.frombuffer(f.read(), dtype=np.dtype(f'<ubdepth')) data = data.reshape(h, w).astype(np.float32) # convert to physical units data *= xstep # example scaling return data, (xstep, ystep) : memory mapping for files >200 MB using numpy.memmap . 5. Experimental Evaluation We tested the viewer on three real STM datasets: The viewer achieves sub-second loading for files up
