Index exceeds array bounds error

Dear all,

I implemented this simple function:

function [SP_Simulation]=get_survivalprobability(a,b,w,dt,m,k)

for i=1:k

for j=1:m

hzrd(i,j)=exp(a(i)+b*w(i,j)) ;

end

end

h1=hzrd(1,:) ; % find a for the first row vector

SP1=(sum(exp(h1*-dt)))/m ;

for i=1:k-1

h=(hzrd(i,:)+hzrd(i+1,:))*-dt;

SP_others=sum(exp(h),2)/m ;

SP_Simulation=[SP1;SP_others];

end

end

whose main code to test is:

SP_termstruc=repmat(0.98,253,1);

m=3;

k=2;

b=0.01;

dt=0.5;

w=rand(253,100);

a0=[0 0]; % guess per fsolve

fun=@(a) (get_survivalprobability(a,b,w,dt,m,k)-SP_termstruc);

[a] = fsolve(fun,a0); % find f(a)=0

for i=1:k % find hazard rates with the new parameters a

for j=1:m

h(i,j)=exp(a(i)+b*w(i,j))

end

end

For the previous example, where m=3 (number of columns) and k=2(number of rows) , everything works. However, when I want to test this function for m=100 and k=253 it shows the error " Matrix dimensions must agree. Error in Tesi>@(a)(get_survivalprobability(a,b,w,dt,m,k)-SP_termstruc)

Could someone help me to understand what is wrong, please?

Thank you in advance

Martina

Answer 1

[David K.]

The problem is a0. In the line

hzrd(i,j)=exp(a(i)+b*w(i,j)) ;

a needs to be have at least k elements for it to work. You set the first a as a0 so a0 needs to have k elements. It works when k = 2 because a0 = [0 0] but wont work for any larger values of k. Replace a0 with

a0 = zeros(1,k);

assuming you want a0 to be zeros.

[Martina Della Monica]

Sorry David , but still shows the same error :(

[David K.]

Ah in my testing I did not change repmat. And since the new error is caused by a mistake in the function output itself my testing did not catch that error.

To fix this second problem, replace SP1 with SP_Simulation as such:

SP1=(sum(exp(h1*-dt)))/m ;
to
SP_Simulation=(sum(exp(h1*-dt)))/m ;

SP_Simulation=[SP1;SP_others];
to
SP_Simulation=[SP_Simulation;SP_others];

[Martina Della Monica]

Thank you so much, David. I really appreciate your help!