Out of Memory on Function

Question

IDN 2022-7-9

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1756495-out-of-memory-on-function

评论： dpb 2022-7-9

calc_RSI.m

Hello,

I have the below function from Matlab file exchange. I am getting an out of memory alert on RSI every time I run a large data set. How can I employ tall arrays in this particulat function? Thanks so much for the help!

function RSI = calc_RSI(data, N)
% CALC_RSI      Calculate RSI indicator given stock price vector
%   RSI = calc_RSI(data, N) calculates the Relative Strength Index (RSI)
%   for the stock price vector, data, over a period of N trading days.
%   Typically the vector 'data' is a list of sequential closing prices for
%   a stock.
%
%   RSI = calc_RSI(data) uses the default period of 14 trading days in
%   calculating the RSI.
%
%   EXAMPLES
%       RSI = calc_RSI(stock,20);
%           Returns the RSI for the data in the vector 'stock' over a
%           period of 20 days.
%
%       RSI = calc_RSI(stock);
%           Returns the RSI for the data in the vector 'stock' using the
%           default period of 14.
%
%   DATA FORMAT
%       data - The vector 'data' must be a numerical vector containing
%           stock prices over consecutive days.  The first and oldest stock
%           price must be located in the first sample (data(1)), while the
%           last and newest stock price is at the end of the array
%           (data(end)).
%
%       N - N is a numerical number specifying the number of samples to use
%           for each period.  The default value is 14.
%
%       RSI - RSI is a vector returned by the program that contains the RSI
%           values calculated by the function.  The first RSI value is
%           calculated using the first N samples.  Therefore, the returned
%           RSI vector is not the same length as the data vector, but will
%           instead have length(data)-N samples.  The last sample in the
%           RSI vector (RSI(end)) corresponds to the RSI value for the most
%           recent date.
% Created by Josiah Renfree
% February 8, 2008
% If no data is passed to the function, give error
if nargin == 0
    error('Please provide a vector of stock prices')    % error prompt
end
% If only one variable is passed, set default period to 14
if nargin == 1
    N = 14;                         % set default period
end
RSI = zeros(1,length(data) - N);    % intialize RSI vector 
Adva = zeros(1,N);                  % initialize positive gain vector
Decl = zeros(1,N);                  % intiialize negative gain vector
% Use the first N samples to calculate the intitial RSI value
for i = 1:N
    chg = data(i+1) - data(i);      % find difference between days
    if chg >= 0                     % if positive change, it advanced
        Adva(i) = chg;              % save to variable
    else                            % if negative change, it declined
        Decl(i) = abs(chg);         % save to variable
    end     
end
AvgGain = mean(Adva);               % take mean of all price increases
AvgLoss = mean(Decl);               % take mean of all price decreases
if AvgLoss == 0                     % if average loss is 0, RSI is 100
    RSI(1) = 100;                   % set RSI to 100
else
    RS = AvgGain / AvgLoss;         % calculate RS value
    RSI(1) = 100 - (100/(1+RS));    % calculate intitial RSI value
end
clear Adva Decl                     % clear variables
% Now cycle through the rest of the data using the initial RSI value to
% calculate the remaining RSI values.
for i = 1+N:length(data)-1
    chg = data(i+1) - data(i);      % calculate change between days
    
    if chg >= 0                     % if positive change, it advanced
        Adva = chg;                 % assign change to advance variable
        Decl = 0;                   % set declined variable to 0
    else                            % if negative change, it declined
        Decl = abs(chg);            % assign change to declined variable
        Adva = 0;                   % set advanced variable to 0
    end
    AvgGain = ((AvgGain*(N-1))+Adva)/N;   % calculate next average gain
    AvgLoss = ((AvgLoss*(N-1))+Decl)/N;   % calculate next average loss
    
    if AvgLoss == 0                 % if average loss is 0, RSI = 100
        RSI(i-N) = 100;             % set RSI to 100
    else
        RS = AvgGain / AvgLoss;     % calculate RS
        RSI(i+1-N) = 100 - (100/(1+RS));    % calculate latest RSI
    end
end

3 个评论
显示 1更早的评论隐藏 1更早的评论

IDN 2022-7-9

Hello, its a table that is 1151636x1

dpb 2022-7-9

在 MATLAB Online 中打开

What time frame does 1M stock prices cover? Seems absurd to me...but, whatever floats your boat. :)

But, what else do you have in memory? I don't have any problem with that size of an array here...actually, the aux vectors computed are only of N length, not the full vector so it isn't terribly inefficient with its memory use after all, when I looked more carefully.

P=randi([30 200],1151636,1);        % a randomized set of prices
RSI=calc_RSI(P);                    % compute the index
>> whos P RSI
  Name            Size              Bytes  Class     Attributes
  P         1151636x1             9213088  double              
  RSI       1151622x1             9212976  double              
>> 

I added a line to RSI that fixes up the output vector direction to match that of the input; your version unchanged will output a row vector instead (or would, if it could finish).

请先登录，再进行评论。

请先登录，再回答此问题。

Out of Memory on Function

3 个评论
显示 1更早的评论隐藏 1更早的评论

回答（0 个）

另请参阅

类别

标签

产品

版本

Community Treasure Hunt

Out of Memory on Function

3 个评论 显示 1更早的评论隐藏 1更早的评论

回答（0 个）

另请参阅

类别

标签

产品

版本

Community Treasure Hunt

3 个评论
显示 1更早的评论隐藏 1更早的评论