inifcns_nstdsums.cpp: S_num takes cl_N as an argument instead of numeric.
authorAlexei Sheplyakov <varg@theor.jinr.ru>
Wed, 19 Mar 2008 09:24:36 +0000 (12:24 +0300)
committerJens Vollinga <jensv@nikhef.nl>
Mon, 28 Jul 2008 19:22:23 +0000 (21:22 +0200)
Implicit conversion from cl_N to numeric considered harmful.

Using GiNaC::numeric for numerical computations incurs significant
overhead, so one might want to do these computations using proper CLN
types. Unfortunately, it's not easy due to automatic conversion from
cln::cl_N to GiNaC::numeric. Here is a simple example:

cl_N x, y;
// initialize them
return sin(x) +  y*exp(y);

The compiler complains about ambigously overloaded of functions, i.e.
cl_N cln::sin(const cl_N&) versus numeric GiNaC::sin(const numeric&).

Hence, I propose to disable *implicit* conversion from cl_N to numeric
(this can be done by marking the numeric ctor as `explicit').

However, this change happens to be a bit nontrivial, because that evil
conversion is used in GiNaC itself. So, I decided to rewrite those fragments
of code.

ginac/inifcns_nstdsums.cpp

index 43add2803c969b4107457ec8f995914d368d6021..bd61fd30e984e7941a48a0ffb1d888f018844ce4 100644 (file)
@@ -320,7 +320,7 @@ cln::cl_N Lin_do_sum_Xn(int n, const cln::cl_N& x)
 
 
 // forward declaration needed by function Li_projection and C below
-numeric S_num(int n, int p, const numeric& x);
+const cln::cl_N S_num(int n, int p, const cln::cl_N& x);
 
 
 // helper function for classical polylog Li
@@ -371,7 +371,7 @@ cln::cl_N Li_projection(int n, const cln::cl_N& x, const cln::float_format_t& pr
                } else {
                        cln::cl_N result = -cln::expt(cln::log(x), n-1) * cln::log(1-x) / cln::factorial(n-1);
                        for (int j=0; j<n-1; j++) {
-                               result = result + (S_num(n-j-1, 1, 1).to_cl_N() - S_num(1, n-j-1, 1-x).to_cl_N())
+                               result = result + (S_num(n-j-1, 1, 1) - S_num(1, n-j-1, 1-x))
                                                  * cln::expt(cln::log(x), j) / cln::factorial(j);
                        }
                        return result;
@@ -402,7 +402,7 @@ numeric Lin_numeric(int n, const numeric& x)
                cln::cl_N x_ = ex_to<numeric>(x).to_cl_N();
                cln::cl_N result = -cln::expt(cln::log(x_), n-1) * cln::log(1-x_) / cln::factorial(n-1);
                for (int j=0; j<n-1; j++) {
-                       result = result + (S_num(n-j-1, 1, 1).to_cl_N() - S_num(1, n-j-1, 1-x_).to_cl_N())
+                       result = result + (S_num(n-j-1, 1, 1) - S_num(1, n-j-1, 1-x_))
                                * cln::expt(cln::log(x_), j) / cln::factorial(j);
                }
                return result;
@@ -1720,10 +1720,10 @@ cln::cl_N C(int n, int p)
                        if (k == 0) {
                                if (n & 1) {
                                        if (j & 1) {
-                                               result = result - 2 * cln::expt(cln::pi(),2*j) * S_num(n-2*j,p,1).to_cl_N() / cln::factorial(2*j);
+                                               result = result - 2 * cln::expt(cln::pi(),2*j) * S_num(n-2*j,p,1) / cln::factorial(2*j);
                                        }
                                        else {
-                                               result = result + 2 * cln::expt(cln::pi(),2*j) * S_num(n-2*j,p,1).to_cl_N() / cln::factorial(2*j);
+                                               result = result + 2 * cln::expt(cln::pi(),2*j) * S_num(n-2*j,p,1) / cln::factorial(2*j);
                                        }
                                }
                        }
@@ -1731,23 +1731,23 @@ cln::cl_N C(int n, int p)
                                if (k & 1) {
                                        if (j & 1) {
                                                result = result + cln::factorial(n+k-1)
-                                                                 * cln::expt(cln::pi(),2*j) * S_num(n+k-2*j,p-k,1).to_cl_N()
+                                                                 * cln::expt(cln::pi(),2*j) * S_num(n+k-2*j,p-k,1)
                                                                  / (cln::factorial(k) * cln::factorial(n-1) * cln::factorial(2*j));
                                        }
                                        else {
                                                result = result - cln::factorial(n+k-1)
-                                                                 * cln::expt(cln::pi(),2*j) * S_num(n+k-2*j,p-k,1).to_cl_N()
+                                                                 * cln::expt(cln::pi(),2*j) * S_num(n+k-2*j,p-k,1)
                                                                  / (cln::factorial(k) * cln::factorial(n-1) * cln::factorial(2*j));
                                        }
                                }
                                else {
                                        if (j & 1) {
-                                               result = result - cln::factorial(n+k-1) * cln::expt(cln::pi(),2*j) * S_num(n+k-2*j,p-k,1).to_cl_N()
+                                               result = result - cln::factorial(n+k-1) * cln::expt(cln::pi(),2*j) * S_num(n+k-2*j,p-k,1)
                                                                  / (cln::factorial(k) * cln::factorial(n-1) * cln::factorial(2*j));
                                        }
                                        else {
                                                result = result + cln::factorial(n+k-1)
-                                                                 * cln::expt(cln::pi(),2*j) * S_num(n+k-2*j,p-k,1).to_cl_N()
+                                                                 * cln::expt(cln::pi(),2*j) * S_num(n+k-2*j,p-k,1)
                                                                  / (cln::factorial(k) * cln::factorial(n-1) * cln::factorial(2*j));
                                        }
                                }
@@ -1870,7 +1870,7 @@ cln::cl_N S_projection(int n, int p, const cln::cl_N& x, const cln::float_format
                                res2 = res2 + cln::expt(cln::cl_I(-1),r) * cln::expt(cln::log(1-x),r)
                                              * S_do_sum(p-r,n-s,1-x,prec) / cln::factorial(r);
                        }
-                       result = result + cln::expt(cln::log(x),s) * (S_num(n-s,p,1).to_cl_N() - res2) / cln::factorial(s);
+                       result = result + cln::expt(cln::log(x),s) * (S_num(n-s,p,1) - res2) / cln::factorial(s);
                }
 
                return result;
@@ -1881,7 +1881,7 @@ cln::cl_N S_projection(int n, int p, const cln::cl_N& x, const cln::float_format
 
 
 // helper function for S(n,p,x)
-numeric S_num(int n, int p, const numeric& x)
+const cln::cl_N S_num(int n, int p, const cln::cl_N& x)
 {
        if (x == 1) {
                if (n == 1) {
@@ -1917,11 +1917,11 @@ numeric S_num(int n, int p, const numeric& x)
        // what is the desired float format?
        // first guess: default format
        cln::float_format_t prec = cln::default_float_format;
-       const cln::cl_N value = x.to_cl_N();
+       const cln::cl_N value = x;
        // second guess: the argument's format
-       if (!x.real().is_rational())
+       if (!instanceof(realpart(value), cln::cl_RA_ring))
                prec = cln::float_format(cln::the<cln::cl_F>(cln::realpart(value)));
-       else if (!x.imag().is_rational())
+       else if (!instanceof(imagpart(value), cln::cl_RA_ring))
                prec = cln::float_format(cln::the<cln::cl_F>(cln::imagpart(value)));
 
        // [Kol] (5.3)
@@ -1934,9 +1934,9 @@ numeric S_num(int n, int p, const numeric& x)
                        cln::cl_N res2;
                        for (int r=0; r<p; r++) {
                                res2 = res2 + cln::expt(cln::cl_I(-1),r) * cln::expt(cln::log(1-value),r)
-                                             * S_num(p-r,n-s,1-value).to_cl_N() / cln::factorial(r);
+                                             * S_num(p-r,n-s,1-value) / cln::factorial(r);
                        }
-                       result = result + cln::expt(cln::log(value),s) * (S_num(n-s,p,1).to_cl_N() - res2) / cln::factorial(s);
+                       result = result + cln::expt(cln::log(value),s) * (S_num(n-s,p,1) - res2) / cln::factorial(s);
                }
 
                return result;
@@ -1951,7 +1951,7 @@ numeric S_num(int n, int p, const numeric& x)
                        for (int r=0; r<=s; r++) {
                                result = result + cln::expt(cln::cl_I(-1),s) * cln::expt(cln::log(-value),r) * cln::factorial(n+s-r-1)
                                                  / cln::factorial(r) / cln::factorial(s-r) / cln::factorial(n-1)
-                                                 * S_num(n+s-r,p-s,cln::recip(value)).to_cl_N();
+                                                 * S_num(n+s-r,p-s,cln::recip(value));
                        }
                }
                result = result * cln::expt(cln::cl_I(-1),n);
@@ -1987,12 +1987,18 @@ numeric S_num(int n, int p, const numeric& x)
 static ex S_evalf(const ex& n, const ex& p, const ex& x)
 {
        if (n.info(info_flags::posint) && p.info(info_flags::posint)) {
+               const int n_ = ex_to<numeric>(n).to_int();
+               const int p_ = ex_to<numeric>(p).to_int();
                if (is_a<numeric>(x)) {
-                       return S_num(ex_to<numeric>(n).to_int(), ex_to<numeric>(p).to_int(), ex_to<numeric>(x));
+                       const cln::cl_N x_ = ex_to<numeric>(x).to_cl_N();
+                       const cln::cl_N result = S_num(n_, p_, x_);
+                       return numeric(result);
                } else {
                        ex x_val = x.evalf();
                        if (is_a<numeric>(x_val)) {
-                               return S_num(ex_to<numeric>(n).to_int(), ex_to<numeric>(p).to_int(), ex_to<numeric>(x_val));
+                               const cln::cl_N x_val_ = ex_to<numeric>(x_val).to_cl_N();
+                               const cln::cl_N result = S_num(n_, p_, x_val_);
+                               return numeric(result);
                        }
                }
        }
@@ -2017,7 +2023,11 @@ static ex S_eval(const ex& n, const ex& p, const ex& x)
                        return Li(n+1, x);
                }
                if (x.info(info_flags::numeric) && (!x.info(info_flags::crational))) {
-                       return S_num(ex_to<numeric>(n).to_int(), ex_to<numeric>(p).to_int(), ex_to<numeric>(x));
+                       int n_ = ex_to<numeric>(n).to_int();
+                       int p_ = ex_to<numeric>(p).to_int();
+                       const cln::cl_N x_ = ex_to<numeric>(x).to_cl_N();
+                       const cln::cl_N result = S_num(n_, p_, x_);
+                       return numeric(result);
                }
        }
        if (n.is_zero()) {