벡터화를 통한 최적화 기법

소개

인도 MIT 인턴쉽 과정 중에 했던 mini project 자료입니다.
Matlab 프로그래밍 자료까지 첨부되어 있습니다.

목차

My Details
Vectorization for optimization
Introduction
Problem1. Sinusoidal waveform optimization
Problem2. The optimization of the signal altered by amplitude and location

본문내용

Generally, we would be used to do iterative loop for optimization. It is because the iterative loop is easy to implement. However, it is useless in the aspect of efficiency. It takes a long time to calculate the loop, if we continue to extend the loop numbers. For solving this problem, we should use the preallocation as well as the vectorization.
Preallocation seems to be not necessary, but you can see the faster data processing speed than before. For example, we can do initialization using the command ‘zeros’, ‘ones’, and so on.
Vectirization means to replace the parallel computation with vector operation. It can improve the processing speed ten-fold., Even if you should implement that one using the algorithm which is harder than iterative loop. We will look into the vectorization with priority in the next part.




0.4530
Problem2. The optimization of the signal altered by amplitude and location
Statement
Firstly, the impulse pass through the digital filter called all-poles filter, and then, the impulse is manipulated to the signal which looks like a mountain having a amplitude and starting location. Actually, we want to find a similar waveform like that fixed waveform using vectorization.
Methodology
We should make a fixed signal by using impulse composed by location index 7 and amplitude 3. Once, the one cycle of ECG is sampled easily. And then, the impulse denoted `x` pass through the filter which has a coefficient made by command arcov. Successfully, we obtain the ytest_v using command `meshgrid` after the fixed signal denoted `ytest` is determined
for i=1:1:158
new_ecg(i)=ecg(i)
end

참고자료

· 없음