```

function SimulatedAnnealing(initialState, temperature, coolingRate)

currentSolution = initialState

bestSolution = currentSolution

currentEnergy = calculateEnergy(currentSolution)

bestEnergy = currentEnergy

while temperature > threshold do

newSolution = generateNeighbor(currentSolution)

newEnergy = calculateEnergy(newSolution)

if newEnergy < currentEnergy then

currentSolution = newSolution

currentEnergy = newEnergy

if newEnergy < bestEnergy then

bestSolution = newSolution

bestEnergy = newEnergy

else

if acceptProbability(currentEnergy, newEnergy, temperature) > random(0, 1) then

currentSolution = newSolution

currentEnergy = newEnergy

temperature = temperature coolingRate

return bestSolution

end function

function generateNeighbor(solution)

// 根据当前解生成一个邻近解

end function

function acceptProbability(energy1, energy2, temperature)

return exp((energy1 - energy2) / temperature)

end function

```

在这个伪代码中，`initialState` 是算法开始时的初始解，`temperature` 是算法的初始温度，`coolingRate` 是每次迭代后温度降低的比例。`threshold` 是一个很小的值，当温度低于这个值时，算法停止。

模拟退火算法通过逐渐降低温度来减少接受较差解的概率，从而在最终收敛到一个较为满意的解。这种方法在解决复杂的组合优化问题时表现出了良好的性能。