Divide a scalar by all non-zero values in a matrix row and input results into a new matrix

I have a matrix cA. I want to input values into a matrix A that are a scalar (1E-14) divided by each element in a row of cA that is not equal to zero.

I have so far:

A = (1E-14)./cA(i,:);

but this uses every column in the ith row. Some of the cell values in the columns are zero so the result is infinity in some cells/elements of A. I would only like to input a value into an element of A if cA(i,j) > 0. Alternatively, if the result is infinity, then just input a zero into that element of A.

[Robert Demyanovich]

It looks like the following should work:

A = (1E-14)./cA(i,:);

A(~isfinite(A)) = 0;

[dpb]

Equivalent, but alternative...and sometimes useful to have the index array for later...

MAGICCONSTANT=1E-14;
ix=(cA(i,:)~=0);
A=zeros(1,size(cA,2));
A(ix)=MAGICCONSTANT./cA(i,ix);

[Meg Noah]

Another Trick:

verySmallNumber = 1.0e-32;
denom = cA;
denom(abs(denom) < verySmallNumber & denom > 0) = verySmallNumber;
denom(abs(denom) < verySmallNumber & denom < 0) = -verySmallNumber;
A=MAGICCONSTANT./denom;

[dpb]

OBTW,

A(isinf(A)) = 0;

also works by definition that

isinf(-inf) == isinf(+inf)