AS_NUMERIC in C more closely corresponds to as.double() in R
All that as.numeric() in R does is assure that the result is of a numeric mode. It does not coerce integer variables to doubles.
Check, for example, str(as.numeric(1:100))
Daniel Hoppe wrote:
Dear List,
I stepped into a strange effect which I can't explain to myself
(probably due to lack of knowledge on R internals).
I have four vectors a,b,c and z of size 10000 each. With these vectors I call
.Call("hyp2f1forrey", a, b, b, z, PACKAGE = "hyp2f1")
to access
SEXP hyp2f1forrey(SEXP a, SEXP b, SEXP c, SEXP x) { int i,n; double *xa, *xb, *xc, *xx, *xresr, *xresi; SEXP resr, resi;
n = LENGTH(a);
PROTECT(a = AS_NUMERIC(a)); PROTECT(b = AS_NUMERIC(b)); PROTECT(c = AS_NUMERIC(c)); PROTECT(x = AS_NUMERIC(x)); PROTECT(resr = NEW_NUMERIC(n)); PROTECT(resi = NEW_NUMERIC(n));
xa = NUMERIC_POINTER(a); xb = NUMERIC_POINTER(b); xc = NUMERIC_POINTER(c); xx = NUMERIC_POINTER(x); xresr = NUMERIC_POINTER(resr); xresi = NUMERIC_POINTER(resi);
for (i = 0; i < n; i++) F77_CALL(hyp)(&xx[i], &xa[i], &xb[i], &xc[i], &xresr[i], &xresi[i]); UNPROTECT(6); return(resr); }
I read the documentation in the way that I could either do the call from
above or
.Call("hyp2f1forrey", as.numeric(a), as.numeric(b), as.numeric(b), as.numeric(z), PACKAGE = "hyp2f1")
and remove the
PROTECT(a = AS_NUMERIC(a)); PROTECT(b = AS_NUMERIC(b)); PROTECT(c = AS_NUMERIC(c)); PROTECT(x = AS_NUMERIC(x));
in the C-Code. What happens now is that in my resulting vector of length
10000 some values are incorrect depending on the exact way of calling
the C-Function.
.Call("hyp2f1forrey", a, b, b, z, PACKAGE = "hyp2f1")[11] --> incorrect result .Call("hyp2f1forrey", a[11], b[11], b[11]), z[11]), PACKAGE = "hyp2f1") --> works .Call("hyp2f1forrey", as.numeric(a), as.numeric(b), as.numeric(c), as.numeric(z), PACKAGE = "hyp2f1")[11] -- works .Call("hyp2f1forrey", a[1:10000], b[1:10000], b[1:10000]), z[1:10000]), PACKAGE = "hyp2f1") --> works too!
Could someone explain to me where the difference in these calls comes from?
Thanks in advance,
Daniel
______________________________________________ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-devel
______________________________________________ [EMAIL PROTECTED] mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-devel
