How to optimize code/avoid a loops

Hello,

I have terrible knowledge of cell arrays and I realized some loops to read and comparing the cell arrays.

If I use big arrays — calculating for every loop takes up to minute for every dataset. It's very inefficient. So, I try to describe my code part:

I have multiple cell-arrays, and I'm read while any time point t_ref_cnt the frequency values in cell M_interpl and Traj_interpl.

If the right time position is chosen, I'm writing values to from M_interpl to motor_actual and Traj_interpl to motorspeed_detected arrays.

After this operation, I have while every t_ref_cnt two Arrays that I can compare (by TP-FN-FP).

Structure of M_interpl is:

1×5 cell array with 1 x 3001 doubles

1. cell is a time t_ref_cnt 0 0,01 0,02 0,03 ... 30,0

2.,3.,4.,5. cells are frequency from 0 to x

Structure of Traj_interpl (depending on of dataset):

example 1×184 cell array with y x 2 doubles (1<y<1000)

1. cell is a time t_ref_cnt 0,03 0,04 0,05 (just small time blocks - not from 0 to 30)

2. cell are frequencys from 1 to 500

I have atatsched both files for example

for t_ref_cnt = 1:length(t_ref) %for every 10ms time step
    %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
    for m_interpl_cnt = 2:length(M_interpl) %for any motor-array
        for m_row_cnt = 1:length(M_interpl{m_interpl_cnt}) %for any row of single motor
            if t_ref(t_ref_cnt) == M_interpl{1}(m_row_cnt) %if right row of motor is choosen
                motor_actual(m_interpl_cnt-1) = M_interpl{m_interpl_cnt}(m_row_cnt); %write actual value of motorspeed to array
                break;
            else
                motor_actual(m_interpl_cnt-1) = 0; 
            end            
        end
    end
    %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%               
    for traj_cnt = 1:length(Traj_interpl) %for every Traj-Cell
        for t_row_cnt = 1:length(Traj_interpl{traj_cnt}) %for any row of Traj 
            if t_ref(t_ref_cnt) == Traj_interpl{traj_cnt}(t_row_cnt,1) %if right row of Traj is choosen 
                motorspeed_detected(traj_cnt) = Traj_interpl{traj_cnt}(t_row_cnt,2); %write detected value to array
                break;
            else
                motorspeed_detected(traj_cnt) = 0; 
            end
        end
    end
end

I think there are a better way to solve it!?

Thank you very much in advice!

For better understanding:

Best

Nik

[darova]

I dont' understand the question. Can you collaborate?

Here is how M_interpl looks

Here is Traj

Can you explain what are you trying to interpolate?

[Nik Rocky]

Hello Darova,

Thanks for your response!

Values of Traj in your graph can be right — you see a lot of “detections” on 120 Hz — this is a second harmonic of Ground truth M_interpl (61,4 · 2) (in this case there are no True Positive values — depending on you count just F0 or F1).

Data of Traj_interpl is already interpolated on the same x-axis (time).

(I see, I uploaded a not interpolated Traj before, here is a right one)

The question is not about interpolation — question is about — is it possible to replace loops (that takes too much time while searching inside all cells for right value) with one MATLAB function?

Thank you!

[darova]

• The question is not about interpolation

Well, maybe you should take a look into it. You have some data (time and value) and want to have some values having time1. interp1 can easily handle this task