About a multivariate normal distribution integration with upper limit are variables

Dear all:

        I got the program from Alan Genz file and I used the qsimvn one. The upper limit in this program was a constant but I want to integral some values with variables. The variables y1, y2 and y3 are chi-square. I want to integral the MVN first then times the chi-square pdf and integral all the y variables. The symbol a is the lower limit matrix and b is the upper limit matrix which the y1, y2 and y2, y3 . r is the correlation matrix. My program

See blow:

syms y1 ; syms y2 ; syms y3 ;

r = [1 0.56694671 ; 0.56694671 1];

a = -inf*[1 1 ]';

b = @(y1,y3,y2) [3.16517./sqrt(19.*(1./y1+1./y3)) 3.16517./sqrt(19.*(1./y2+1./y3)) ]';

[p e] = qsimvn( 5000, r, a, b );

f = @(y1,y2,y3) [p e].*[chi2pdf(y1,9) 0].*[chi2pdf(y2,9) 0].*[chi2pdf(y3,9) 0];

integral3(f,0,50,0,50,0,50);

the error terms :

Error using @(y1,y3,y2)[3.16517./sqrt(19.*(1./y1+1./y3)),3.16517./sqrt(19.*(1./y2+1./y3))]' Not enough input arguments.

Error in qsimvn>chlrdr (line 116) ap(i) = ap(i)/d(i); bp(i) = bp(i)/d(i);

Error in qsimvn (line 54) [n, n] = size(r); [ ch as bs ] = chlrdr( r, a, b );

How do i fix the problem? I need it. REALLY thank u all!