Re: Bad expression parsing/evaluation

[Andrew Makhorin]

There exist at least two different definitions of the modulo operator
that lead to different results in case of negative operands; please see
https://en.wikipedia.org/wiki/Modulo_operation .

Mathematically Knuth's definition (used in glpk) is most correct.

An easy way to avoid any surprises is just not to use negative operands.




On Sun, 2021-08-15 at 21:06 +0200, Domingo Alvarez Duarte wrote:
> Definitely the way GMPL/GLPK calculates the modulo of a negative
> number 
> is wrong:
> 
> ====
> 
> param a := 4;
> param b := 3;
> param c := a mod b;
> param d := a - c;
> 
> 
> param a2 := -4;
> param b2 := 3;
> param c2 := a2 mod b2;
> param d2 := a2 - c2;
> 
> display a, b, c, d, a2, b2, c2, d2;
> 
> ====
> 
> AMPL output:
> 
> ====
> 
> ampl test4.ampl
> a = 4
> b = 3
> c = 1
> d = 3
> a2 = -4
> b2 = 3
> c2 = -1
> d2 = -3
> 
> ====
> 
> GMPL/GLPK/GLPSOL output:
> 
> ====
> 
> glpsol -m test4.ampl
> GLPSOL: GLPK LP/MIP Solver, v4.65-ex, glp_double size 8
> Parameter(s) specified in the command line:
>   -m test4.ampl
> Reading model section from test4.ampl...
> test4.ampl:12: warning: unexpected end of file; missing end statement 
> inserted
> 12 lines were read
> Display statement at line 12
> a = 4
> b = 3
> c = 1
> d = 3
> a2 = -4
> b2 = 3
> c2 = 2
> d2 = -6
> 
> ...
> 
> ====
> 
> Cheers !
> 
> On 15/8/21 16:50, Heinrich Schuchardt wrote:
> > On 8/15/21 4:06 PM, Domingo Alvarez Duarte wrote:
> > > Comparing how AMPL and GMPL calculate random expressions I found
> > > some of
> > > then where they differ or GMPL can't manage see bellow, also one
> > > of then
> > > that references itself makes glpsol segfault but ampl gives an
> > > error
> > 
> > I can't see a segfault with GLPK 5.0.
> > 
> > Please, use gdb to identify in which line of code it occurs:
> > 
> >     gdb --args glpk -m test.od
> > 
> > > message explaining the problem:
> > > 
> > > ====
> > > 
> > > param a0 := (((((((788)*(8.46))))+8342*1.803-1))*4186.4*(15));
> > > printf "%f\n", a0;
> > > 
> > > param a1 := (((22 mod 284/((7530/((2)*(((((25))-421))))))*597 mod
> > > 2663)+7283.8-9.60+167 mod ((3))))+(8871);
> > > printf "%f\n", a1;
> > > 
> > > #param a2 := a1 * a2; #glpsol segfault
> > > param a2 := a0 * a1;
> > > printf "%f\n", a2;
> > > 
> > > param a = (((((((788)*(8.46))))+8342*1.803-1))*4186.4*(15))*(((22
> > > mod
> > 
> > Your syntax is wrong. This should be:
> > 
> > param a :=
> > 
> > With correct syntax:
> > 
> > glpsol -m /tmp/test.mod
> > GLPSOL: GLPK LP/MIP Solver, v4.65
> > Parameter(s) specified in the command line:
> >  -m /tmp/test.mod
> > Reading model section from /tmp/test.mod...
> > 28 lines were read
> > 1363056632.376000
> > 17428.775299
> > 23756407765226.851562
> > 11664732388290.359375
> > 11665910614443.386719
> > 8259816920615.111328
> > 8259816920615.111328
> > Model has been successfully generated
> > 
> > Best regards
> > 
> > Heinrich
> > 
> > > 284/((7530/((2)*(((((25))-421))))))*597 mod 2663)+7283.8-9.60+167
> > > mod
> > > ((3))))+(8871);
> > > #param a := 3+2;
> > > printf "%f\n", a;
> > > 
> > > param b = (((((((788)*(8.46))))+8342*1.803-1))*4186.4*(15))*(((22
> > > mod
> > > 284/((7530/((2)*(((((25))-421))))))*597 mod 2663)+7283.8-
> > > 9.60+167.8644
> > > mod ((3))))+(8871);
> > > #param a := 3+2;
> > > printf "%f\n", b;
> > > 
> > > param c =
> > > (((((((788)*(8.46))))+8342*1.803-
> > > 1))*4186.4*(15))*(((22/((7530/((2)*(((((25))-
> > > 421))))))*597)+7283.8-9.60+167))+(8871); 
> > > 
> > > 
> > > #param a := 3+2;
> > > printf "%f\n", c;
> > > 
> > > param d =
> > > (((((((788.0)*(8.46))))+8342.0*1.803-
> > > 1.0))*4186.4*(15.0))*(((22.0/((7530.0/((2.0)*(((((25.0))-
> > > 421))))))*597.0)+7283.8-9.60+167.0))+(8871.0); 
> > > 
> > > 
> > > #param a := 3+2;
> > > printf "%f\n", d;
> > > 
> > > ====
> > > 
> > > AMPL output:
> > > 
> > > ====
> > > 
> > > ampl test.ampl
> > > 1363056632.376000
> > > 14765.775299
> > > 20126587953209.562500
> > > 8034912576273.071289
> > > 8036090802426.097656
> > > 8259816920615.111328
> > > 8259816920615.111328
> > > 
> > > ====
> > > 
> > > GLPSOL output:
> > > 
> > > ====
> > > 
> > > glpsol -m test.ampl
> > > GLPSOL: GLPK LP/MIP Solver, v4.65
> > > Parameter(s) specified in the command line:
> > >   -m test.ampl
> > > Reading model section from test.ampl...
> > > test.ampl:25: warning: unexpected end of file; missing end
> > > statement
> > > inserted
> > > 25 lines were read
> > > 1363056632.376000
> > > 17428.775299
> > > 23756407765226.851562
> > > test.ampl:13: no value for a
> > > MathProg model processing error
> > > 
> > > ====
> > > 
> > > With a incorrect expression:
> > > 
> > > ====
> > > 
> > > param a2 := a1 * a2; #glpsol segfault
> > > #param a2 := a0 * a1;
> > > 
> > > ====
> > > 
> > > AMPL output:
> > > 
> > > ====
> > > 
> > > ampl test.ampl
> > > 1363056632.376000
> > > 14765.775299
> > > 
> > > test.ampl, line 7 (offset 228):
> > >      a2 is not defined
> > > context:  param a2 := a1 *  >>> a2; <<< #glpsol segfault
> > > 
> > > ====
> > > 
> > > GLPSOL output:
> > > 
> > > ====
> > > 
> > > glpsol -m test.ampl
> > > GLPSOL: GLPK LP/MIP Solver, v4.65
> > > Parameter(s) specified in the command line:
> > >   -m test.ampl
> > > Reading model section from test.ampl...
> > > test.ampl:25: warning: unexpected end of file; missing end
> > > statement
> > > inserted
> > > 25 lines were read
> > > 1363056632.376000
> > > 17428.775299
> > > Segmentation fault (core dumped)
> > > 
> > > ====
> > > 
> > > Cheers !
> > > 
> > > 
> 
>