Autism Spectrum Disorder (ASD) impacts social interaction, communication, and cognitive functioning. In recent years, virtual reality (VR) technology, with its advantages such as strong immersive capabilities, has been increasingly applied to ASD ergotherapy. However, existing systems often overlook human factors considerations, including users' perceptual preferences and attentional load during system interactions. This study proposes a Personalized Multisensory Adaptive Roaming System (P-MARS), which integrates large-scale models to analyze users' sensory preferences and generates personalized VR scenarios combined with visual, auditory, and olfactory stimuli, aiming to enhance ergotherapy outcomes for children with ASD. A longitudinal study employing functional near-infrared spectroscopy (fNIRS) was conducted to assess participants' neural states. The results demonstrate a significant reduction in central neural activation levels among children during the later stages of training. These findings suggest that periodic VR-based training can substantially alleviate cognitive load and anxiety in children with ASD, thereby accelerating ergotherapy progress. The study further validates the critical importance and efficacy of incorporating human factors into ergotherapy system design.

Recently researchers have proposed modeling register allocation as an integer linear programming (IP) problem and solving it optimally for general purpose processors and for dedicated embedded systems. Compared with traditional graph-coloring approaches, the IP-based allocators can improve a program's performance. However the solution times are much slower This paper presents an IP-based optimal register allocator which is much faster than previous work. We present several local and global reduction techniques to identify locations in a program's control-flow graph where spill decisions and register deallocation decisions are unnecessary for optimal register allocation. We propose a hierarchical reduction approach to efficiently remove the corresponding redundant decisions and constraints from the IP model. This allocator is built into the Gnu C Compiler and is evaluated experimentally using the SPEC92INT benchmarks. The results show that the improved IP model is much simpler The number of constraints produced is almost linear with the function size. The optimal allocation time is much faster with a speedup factor of about 150 for hard allocation problems.