How to create a vector based on previous value of same vector

Question

Russell Senior 2019-6-5

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/465736-how-to-create-a-vector-based-on-previous-value-of-same-vector

编辑： dpb 2019-6-11

I am trying to create a vector where the value of a particular element depends on the value of a previous element. Specifically, I have a vector of torque values, and when the values are above a threshold, I want my new vector to hold true until the torque level drops below the threshold plus some hysteresis.

An example. In this case Active = 1 when Torque > 6, but Active doesn't = 0 again until Torque < 4.

Torque = [0 1 2 3 4 5 6 7 8 9 8 7 6 5 4 3 2 1 0];
Active = [0 0 0 0 0 0 0 1 1 1 1 1 1 1 1 0 0 0 0];

Is there a way to do this with vector operations? I am interested in speed, so while a for loop will work easily, I want it to be as fast as possible.

Thanks.

3 个评论
显示 1更早的评论隐藏 1更早的评论

dpb 2019-6-5

Use the loop and go on--then profile the end result to see if this ends up being the bottleneck. I'm guessing you'll find your actual choke points will be something else.

Don't try to micro-optimize; wait until you know where the real issues are--until then, write most "dead ahead" code you can until it is shown to somehow be inadequate.

Russell Senior 2019-6-10

I appreciate your response. However, my code was already finished, and I'm trying to optimize it as much as possible. My function itself takes about 54ms after changing as much as I can to vector functions.

The slowest part of my function is a call to the interpn function, which has proven difficult to improve.

The last low hanging fruit would be an answer to my original question. If that, too, proves difficult / impossible to improve, then I will consider 54ms as good as it gets.

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

dpb 2019-6-10

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/465736-how-to-create-a-vector-based-on-previous-value-of-same-vector#answer_378576

编辑：dpb 2019-6-11

在 MATLAB Online 中打开

I suspect overall performance will not be significantly improved as removing a small fraction of a total run time entirely will still be only a small fraction of the total.

But to the original question alone...

On=6; Off=4;                                       % or whatever limits are
S=[false sign(diff(Torque))];                      % accel/decel?
Active=(Torque>=On & S>=0) | (Torque>=Off & S<0);

will work for patterns such as the example--whether will be sufficient in general for real data is another Q?