Modeling and analyzing multiphysics effects has become one of the most challenging issues in integrated circuit design. Equivalent thermal circuit method is one of the most commonly used method in circuit design to simulate the electrothermal coupling effects, since it is fast and compatible with SPICE. However, it is still difficult to realize electro-thermalmechanical coupling simulation based on equivalent circuit method due to a lack of equivalent mechanical circuit method, which brings difficulties to do electro-thermal-mechanical analysis by SPICE. The equivalent mechanical circuit method is proposed based on solid mechanics equilibrium equation by deriving the electro-mechanical equivalent relation, equivalent circuit elements, equivalent circuit structure, equivalent circuit boundary condition, and the solving algorithm. The equivalent multiphysics circuit of TSV and FinFET are then further constructed to simulate the electro-thermal-mechanical coupling effects and verified with simulation results obtained from the finite element method (FEM). The results show that our proposed equivalent multiphysics circuit framework is able to simulate the electro-thermal-mechanical coupling effects by SPICE in integrated circuits.

Experience Replay (ER) allows Deep Reinforcement Learning (RL) agent to reuse past experience, as though recall the same Experience repeatedly. ER enables RL algorithm to be trained by reusing previous states, so that RL agent can obtain more accurate value estimations and action selections. Current policy algorithms either have a rule-based replay policy or uniformly replay past experiences, which may be suboptimal. The agent is updated based on the replay data to maximize the cumulative reward, and the replay policy is updated to provide the more valuable experience for the agent. In this work, we propose a novel experience replay algorithm Regret Minimization Experience Replay (RMER), to improve the immediate reward via sampling and ensure certain exploration capability of agent. Finally, we prove the ascendency of RMER with different off- policy algorithms on the suite of Open AI gym continuous control tasks.