I'd like to be able to control the decimal accuracy exactly while iterating a chaotic map. In particular, I'd like to make sure that Matlab uses no more / no less than the value p I specify in the following function:
function out = chaotic_logistic(seed,n,p,varargin)
%Returns a vector containing the seed and n iterations
%of the logistic map: u*x*(1-x), which is chaotic on the
%unit interval when u is contained in (~3.56995,4)
%If a third argument is passed, this will be used as the u parameter.
%Otherwise u=4 by default.
%Finally, if 'last' is passed as the fourth argument, only the last
%iteration will be returned.
%NOTE: As currently designed, this function can be broken
%by passing 'last' as the THIRD argument, etc. SO BE CAREFUL OR FIX IT.
digits(p) %sets desried accuracy to p decimal places??
if length(varargin)>=1 %default param
param = varargin{1};
else
param = 4;
end
out = [seed zeros(1,n-1)]; %preallocate
apply_fcn = vpa(seed); %see vpa doc??
for i=1:n-1
apply_fcn = vpa(param.*apply_fcn.*(1-apply_fcn)); %note vpa again
out(i+1) = apply_fcn;
end
if length(varargin)==2 && strcmp(varargin{2},'last')
out = out(end);
end
If I understand correctly, digits() should do the trick, but I'm not sure and it's crucial that I have the accuracy as specified when I call the function (as anyone familiar with chaos will know!).
Thanks in advance!
Related
I have to construct the following function in MATLAB and am having trouble.
Consider the function s(t) defined for t in [0,4) by
{ sin(pi*t/2) , for t in [0,1)
s(t) = { -(t-2)^3 , for t in [1,3)*
{ sin(pi*t/2) , for t in [3,4)
(i) Generate a column vector s consisting of 512 uniform
samples of this function over the interval [0,4). (This
is best done by concatenating three vectors.)
I know it has to be something of the form.
N = 512;
s = sin(5*t/N).' ;
But I need s to be the piecewise function, can someone provide assistance with this?
If I understand correctly, you're trying to create 3 vectors which calculate the specific function outputs for all t, then take slices of each and concatenate them depending on the actual value of t. This is inefficient as you're initialising 3 times as many vectors as you actually want (memory), and also making 3 times as many calculations (CPU), most of which will just be thrown away. To top it off, it'll be a bit tricky to use concatenate if your t is ever not as you expect (i.e. monotonically increasing). It might be an unlikely situation, but better to be general.
Here are two alternatives, the first is imho the nice Matlab way, the second is the more conventional way (you might be more used to that if you're coming from C++ or something, I was for a long time).
function example()
t = linspace(0,4,513); % generate your time-trajectory
t = t(1:end-1); % exclude final value which is 4
tic
traj1 = myFunc(t);
toc
tic
traj2 = classicStyle(t);
toc
end
function trajectory = myFunc(t)
trajectory = zeros(size(t)); % since you know the size of your output, generate it at the beginning. More efficient than dynamically growing this.
% you could put an assert for t>0 and t<3, otherwise you could end up with 0s wherever t is outside your expected range
% find the indices for each piecewise segment you care about
idx1 = find(t<1);
idx2 = find(t>=1 & t<3);
idx3 = find(t>=3 & t<4);
% now calculate each entry apprioriately
trajectory(idx1) = sin(pi.*t(idx1)./2);
trajectory(idx2) = -(t(idx2)-2).^3;
trajectory(idx3) = sin(pi.*t(idx3)./2);
end
function trajectory = classicStyle(t)
trajectory = zeros(size(t));
% conventional way: loop over each t, and differentiate with if-else
% works, but a lot more code and ugly
for i=1:numel(t)
if t(i)<1
trajectory(i) = sin(pi*t(i)/2);
elseif t(i)>=1 & t(i)<3
trajectory(i) = -(t(i)-2)^3;
elseif t(i)>=3 & t(i)<4
trajectory(i) = sin(pi*t(i)/2);
else
error('t is beyond bounds!')
end
end
end
Note that when I tried it, the 'conventional way' is sometimes faster for the sampling size you're working on, although the first way (myFunc) is definitely faster as you scale up really a lot. In anycase I recommend the first approach, as it is much easier to read.
The function is supposed to return a value. However, when I type:
val(1d4)
it returns
ans =
[]
However, if I copied line by line, and set all parameters (V eps_today etc.), and run in the command window, it works fine...Have no clue where the problem is. Thanks:)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%This function calculates the value function of both small and big%
%investment %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function negval=val(k_next)
global V eps_today k_today eps_gv k_gv z_today delta xi
theta=0.5327; beta=0.95; rho=0.7675; sigma_eta=0.7704;
%identify the grid point that falls just below
klo=max(sum(k_next>k_gv),1);
%above
khi=klo+1;
%Linear Interpolation, effectively only on the dimension of k
if klo<size(k_gv,2)
V_temp=V(:,klo)+(k_next-k_gv(klo))/(k_gv(khi)-k_gv(klo))*(V(:,khi)-V(:,klo));
else
%when k_next> maximum point of k_gv
V_temp=V(:,size(k_gv,2));
end
EV=mean(V_temp,1);
negval=-(exp(z_today+eps_today)*k_today^theta-xi*k_today-(k_next-(1-delta)*k_today)+beta*EV);
end
I tried to replicate your problem but I couldn't. If I do not set the global variables Matlab shouts and refuses to proceed through the function. Setting some of them to empty values was the same (did not try them all). If I do set the globals then the function always gives me a value.
But I strongly suspect it has something to do with the globals anyway (can you show the code where they are initialized?). As I mentioned in the comment I would get rid of them, the code could look like this
Main programme
%set parameters
Pars.beta = beta=0.95;
Pars.theta = 0.5327;
Pars.rho=0.7675;
Pars.sigma_eta=0.7704;
Pars.delta = ...;
Pars.xi = ...;
%set grid
eps_gv = ...
k_gv = ...
ne = length(eps_gv);
nk = length(k_gv);
V = zeros(ne,nk);
k_next = ...;
value = val(k_next,V,eps_gv,k_gv,Pars);
Val function
negval = function val(k_next,V,eps_gv,k_gv,Pars)
theta=Pars.theta; beta=Pars.beta; rho=Pars.rho; etc.
You don't even need eps_today and k_today as these should be the values on the grid (ie eps_gv and k_gv). I don't know what z_today is but there should either be a grid for z or z_today should be a parameter, in which case it should be in Pars. Of course if you use eps_gv and k_gv instead of eps_today and k_today the negval = ... line needs to be modified to take account of them being arrays not scalars.
One last comment, there is a bug on the EV=mean(V_temp,1); line. The mean function calculates the average of (the columns of) V. To calculate the expected value you need to do a weighted average where you weight each row of V by the probability of eps being eps_gv(1) (ie sum(V(i,:)*prob_eps(i)), sum going over i) . What you did works only if all shocks have equal probability. Pointing out since I am not sure you are aware of that.
My main script contains following code:
%# Grid and model parameters
nModel=50;
nModel_want=1;
nI_grid1=5;
Nth=1;
nRow.Scale1=5;
nCol.Scale1=5;
nRow.Scale2=5^2;
nCol.Scale2=5^2;
theta = 90; % degrees
a_minor = 2; % range along minor direction
a_major = 5; % range along major direction
sill = var(reshape(Deff_matrix_NthModel,nCell.Scale1,1)); % variance of the coarse data matrix of size nRow.Scale1 X nCol.Scale1
%# Covariance computation
% Scale 1
for ihRow = 1:nRow.Scale1
for ihCol = 1:nCol.Scale1
[cov.Scale1(ihRow,ihCol),heff.Scale1(ihRow,ihCol)] = general_CovModel(theta, ihCol, ihRow, a_minor, a_major, sill, 'Exp');
end
end
% Scale 2
for ihRow = 1:nRow.Scale2
for ihCol = 1:nCol.Scale2
[cov.Scale2(ihRow,ihCol),heff.Scale2(ihRow,ihCol)] = general_CovModel(theta, ihCol/(nCol.Scale2/nCol.Scale1), ihRow/(nRow.Scale2/nRow.Scale1), a_minor, a_major, sill/(nRow.Scale2*nCol.Scale2), 'Exp');
end
end
%# Scale-up of fine scale values by averaging
[covAvg.Scale2,var_covAvg.Scale2,varNorm_covAvg.Scale2] = general_AverageProperty(nRow.Scale2/nRow.Scale1,nCol.Scale2/nCol.Scale1,1,nRow.Scale1,nCol.Scale1,1,cov.Scale2,1);
I am using two functions, general_CovModel() and general_AverageProperty(), in my main script which are given as following:
function [cov,h_eff] = general_CovModel(theta, hx, hy, a_minor, a_major, sill, mod_type)
% mod_type should be in strings
angle_rad = theta*(pi/180); % theta in degrees, angle_rad in radians
R_theta = [sin(angle_rad) cos(angle_rad); -cos(angle_rad) sin(angle_rad)];
h = [hx; hy];
lambda = a_minor/a_major;
D_lambda = [lambda 0; 0 1];
h_2prime = D_lambda*R_theta*h;
h_eff = sqrt((h_2prime(1)^2)+(h_2prime(2)^2));
if strcmp(mod_type,'Sph')==1 || strcmp(mod_type,'sph') ==1
if h_eff<=a
cov = sill - sill.*(1.5*(h_eff/a_minor)-0.5*((h_eff/a_minor)^3));
else
cov = sill;
end
elseif strcmp(mod_type,'Exp')==1 || strcmp(mod_type,'exp') ==1
cov = sill-(sill.*(1-exp(-(3*h_eff)/a_minor)));
elseif strcmp(mod_type,'Gauss')==1 || strcmp(mod_type,'gauss') ==1
cov = sill-(sill.*(1-exp(-((3*h_eff)^2/(a_minor^2)))));
end
and
function [PropertyAvg,variance_PropertyAvg,NormVariance_PropertyAvg]=...
general_AverageProperty(blocksize_row,blocksize_col,blocksize_t,...
nUpscaledRow,nUpscaledCol,nUpscaledT,PropertyArray,omega)
% This function computes average of a property and variance of that averaged
% property using power averaging
PropertyAvg=zeros(nUpscaledRow,nUpscaledCol,nUpscaledT);
%# Average of property
for k=1:nUpscaledT,
for j=1:nUpscaledCol,
for i=1:nUpscaledRow,
sum=0;
for a=1:blocksize_row,
for b=1:blocksize_col,
for c=1:blocksize_t,
sum=sum+(PropertyArray((i-1)*blocksize_row+a,(j-1)*blocksize_col+b,(k-1)*blocksize_t+c).^omega); % add all the property values in 'blocksize_x','blocksize_y','blocksize_t' to one variable
end
end
end
PropertyAvg(i,j,k)=(sum/(blocksize_row*blocksize_col*blocksize_t)).^(1/omega); % take average of the summed property
end
end
end
%# Variance of averageed property
variance_PropertyAvg=var(reshape(PropertyAvg,...
nUpscaledRow*nUpscaledCol*nUpscaledT,1),1,1);
%# Normalized variance of averageed property
NormVariance_PropertyAvg=variance_PropertyAvg./(var(reshape(...
PropertyArray,numel(PropertyArray),1),1,1));
Question: Using Matlab, I would like to optimize covAvg.Scale2 such that it matches closely with cov.Scale1 by perturbing/varying any (or all) of the following variables
1) a_minor
2) a_major
3) theta
I am aware I can use fminsearch, however, how I am not able to perturb the variables I want to while using this fminsearch.
I won't pretend to understand everything that you are doing. But it sounds like a typical minimization problem. What you want to do is to come up with a single function that takes a_minor, a_major and theta as arguments, and returns the square of the difference between covAvg.Scale2 and cov.Scale1. Something like this:
function diff = minimize_me(a_minor, a_major, theta)
... your script goes here
diff = (covAvg.Scale2 - cov.Scale1)^2;
end
Then you need matlab to minimize this function. There's more than one option here. Since you only have three variables to minimize over, fminsearch is a good place to start. You would call it something like this:
opts = optimset('display', 'iter');
x = fminsearch( #(x) minimize_me(x(1), x(2), x(3)), [a_minor_start a_major_start theta_start], opts)
The first argument to fminsearch is the function you want to optimize. It must take a single argument: a vector of the variables that will be perturbed in order to find the minimum value. Here I use an anonymous function to extract the values from this vector and pass them into minimize_me. The second argument to fminsearch is a vector containing the values to start searching at. The third argument are options that affect the search; it's a good idea to set display to iter when you first start optimizing, so that you can get an idea of well the optimizer is converging.
If your parameters have restricted domains (e.g. they must all be positive) take a look at fminsearchbnd on the file exchange.
If I have misunderstood your problem, and this doesn't help at all, try posting code that we can run to reproduce the problem ourselves.
I am new to Matlab. I was reading this code snippet, but in some parts (marked with asterisks) I don't understand what it means, so if anybody could help would be very much appreciated
function [A1nmb] = moran(initsize, popsize)
% MORAN generates a trajectory of a Moran type process
% which gives the number of genes of allelic type A1 in a population
% of haploid individuals that can exist in either type A1 or type A2.
% The population size is popsize and the initial number of type A1
% individuals os initsize.
% Inputs: initsize - initial number of A1 genes
% popsize - the total population size (preserved)
if (nargin==0)
initsize=10;
popsize=30;
end
A1nmb=zeros(1,popsize);
A1nmb(1)=initsize;
**lambda = inline('(x-1).*(1-(x-1)./N)', 'x', 'N');
mu = inline('(x-1).*(1-(x-1)./N)', 'x', 'N');**
x=initsize;
i=1;
while (x>1 & x<popsize+1)
if (lambda(x,popsize)/(lambda(x,popsize)+mu(x,popsize))>rand)
x=x+1;
A1nmb(i)=x;
else
x=x-1;
A1nmb(i)=x;
end;
i=i+1;
end;
nmbsteps=length(A1nmb);
***rate = lambda(A1nmb(1:nmbsteps-1),popsize) ...
+mu(A1nmb(1:nmbsteps-1),popsize);***
**jumptimes=cumsum(-log(rand(1,nmbsteps-1))./rate);**
jumptimes=[0 jumptimes];
stairs(jumptimes,A1nmb);
axis([0 jumptimes(nmbsteps) 0 popsize+1]);
The first line you marked
lambda = inline('(x-1).*(1-(x-1)./N)', 'x', 'N');
creates something called an inline function. It is equivalent to defining a mathematical function. Example:
y = inline('x^2')
would allow you to do
>> y(2)
4
This immediately explains the second line you marked.
rate = lambda(A1nmb(1:nmbsteps-1),popsize) ...
+mu(A1nmb(1:nmbsteps-1),popsize);
will compute the value of the function lambda(x,N) at x = A1nmb(1:nmbsteps-1) and N = popsize.
I will say immediately here that you should take a look at anonymous functions, a different format used to accomplish the same as inline. Only, anonymous functions are generally better supported, and usually a lot faster than inline functions.
Then, for the final line,
jumptimes = cumsum(-log(rand(1,nmbsteps-1))./rate);
is a nested command. rand will create a matrix containing pseudorandom numbers, log is the natural logarithm ("ln"), and cumsum creates a new matrix, where all the elements in the new matrix are the cumulative sum of the elements in the input matrix.
You will find the commands doc and help very useful. Try typing
doc cumsum
or
help inline
on the Matlab command prompt. Try that again with the commands forming the previous statement.
As a general word of advice: spend an insane lot of time reading through the documentation. Really, for each new command you encounter, read about it and play with it in a sandbox until you feel you understand it. Matlab only becomes powerful if you know all its commands, and there are a lot to get to know.
It defines an inline function object. For example this
lambda = inline('(x-1).*(1-(x-1)./N)', 'x', 'N')
defines lambda as a function with 2 variables. When you call lambda(A,n) Matlab simply expands the function you define in the first string. Thus lambda(A,n) using the variables you provide in the function call. lambda(A,n) would will evaluate to:
(A-1).*(1-(A-1)./n)
it just expands the function using the parameters you supply. Take a look at this link for more specific details http://www.mathworks.co.uk/help/techdoc/ref/inline.html
The cumsum function just returns the cumulative sum of a matrix along a particular dimension. Say we call cumsum on a vector X, then the value at element i in the result is equal to the sum of elements in X from index 1 to i. For example X = [1 2 1 3] we would get
AA = cumsum(X);
we would have
AA = [1 3 5 8]
See this link for more details and examples http://www.mathworks.co.uk/help/techdoc/ref/cumsum.html
I used nlfilter for a test function of mine as follows:
function funct
clear all;
clc;
I = rand(11,11);
ld = input('Enter the lag = ') % prompt for lag distance
A = nlfilter(I, [7 7], #dirvar);
% Subfunction
function [h] = dirvar(I)
c = (size(I)+1)/2
EW = I(c(1),c(2):end)
h = length(EW) - ld
end
end
The function works fine but it is expected that nlfilter progresses element by element, but in first two iterations the values of EW will be same 0.2089 0.4162 0.9398 0.1058. But then onwards for all iterations the next element is selected, for 3rd it is 0.4162 0.9398 0.1058 0.1920, for 4th it is 0.9398 0.1058 0.1920 0.5201 and so on. Why is it so?
This is nothing to worry about. It happens because nlfilter needs to evaluate your function to know what kind of output to create. So it uses feval once before starting to move across the image. The output from this feval call is what you see the first time.
From the nlfilter code:
% Find out what output type to make.
rows = 0:(nhood(1)-1);
cols = 0:(nhood(2)-1);
b = mkconstarray(class(feval(fun,aa(1+rows,1+cols),params{:})), 0, size(a));
% Apply fun to each neighborhood of a
f = waitbar(0,'Applying neighborhood operation...');
for i=1:ma,
for j=1:na,
x = aa(i+rows,j+cols);
b(i,j) = feval(fun,x,params{:});
end
waitbar(i/ma)
end
The 4th line call to eval is what you observe as the first output from EW, but it is not used to anything other than making the b matrix the right class. All the proper iterations happen in the for loop below. This means that the "duplicate" values you observe does not affect your final output matrix, and you need not worry.
I hope you know what the length function does? It does not give you the Euclidean length of a vector, but rather the largest dimension of a vector (so in your case, that should be 4). If you want the Euclidean length (or 2-norm), use the function norm instead. If your code does the right thing, you might want to use something like:
sz = size(I,2);
h = sz - (sz+1)/2 - ld;
In your example, this means that depending on the lag you provide, the output should be constant. Also note that you might want to put semicolons after each line in your subfunction and that using clear all as the first line of a function is useless since a function will always be executed in its own workspace (that will however clear persistent or global variables, but you don't use them in your code).