Coverage for sympy/solvers/polysys.py : 98%

"""Solvers of systems of polynomial equations. """ 

from __future__ import print_function, division 

from sympy.core import S 

from sympy.polys import Poly, groebner, roots 

from sympy.polys.polytools import parallel_poly_from_expr 

from sympy.polys.polyerrors import (ComputationFailed, 

    PolificationFailed, CoercionFailed) 

from sympy.simplify import rcollect 

from sympy.utilities import default_sort_key, postfixes 

class SolveFailed(Exception): 

    """Raised when solver's conditions weren't met. """ 

def solve_poly_system(seq, *gens, **args): 

    """ 

    Solve a system of polynomial equations. 

    Examples 

    ======== 

    >>> from sympy import solve_poly_system 

    >>> from sympy.abc import x, y 

    >>> solve_poly_system([x*y - 2*y, 2*y**2 - x**2], x, y) 

    [(0, 0), (2, -sqrt(2)), (2, sqrt(2))] 

    """ 

    try: 

        polys, opt = parallel_poly_from_expr(seq, *gens, **args) 

    except PolificationFailed as exc: 

        raise ComputationFailed('solve_poly_system', len(seq), exc) 

    if len(polys) == len(opt.gens) == 2: 

        f, g = polys 

        a, b = f.degree_list() 

        c, d = g.degree_list() 

        if a <= 2 and b <= 2 and c <= 2 and d <= 2: 

            try: 

                return solve_biquadratic(f, g, opt) 

            except SolveFailed: 

                pass 

    return solve_generic(polys, opt) 

def solve_biquadratic(f, g, opt): 

    """Solve a system of two bivariate quadratic polynomial equations. 

    Examples 

    ======== 

    >>> from sympy.polys import Options, Poly 

    >>> from sympy.abc import x, y 

    >>> from sympy.solvers.polysys import solve_biquadratic 

    >>> NewOption = Options((x, y), {'domain': 'ZZ'}) 

    >>> a = Poly(y**2 - 4 + x, y, x, domain='ZZ') 

    >>> b = Poly(y*2 + 3*x - 7, y, x, domain='ZZ') 

    >>> solve_biquadratic(a, b, NewOption) 

    [(1/3, 3), (41/27, 11/9)] 

    >>> a = Poly(y + x**2 - 3, y, x, domain='ZZ') 

    >>> b = Poly(-y + x - 4, y, x, domain='ZZ') 

    >>> solve_biquadratic(a, b, NewOption) 

    [(-sqrt(29)/2 + 7/2, -sqrt(29)/2 - 1/2), (sqrt(29)/2 + 7/2, -1/2 + \ 

      sqrt(29)/2)] 

    """ 

    G = groebner([f, g]) 

    if len(G) == 1 and G[0].is_ground: 

        return None 

    if len(G) != 2: 

        raise SolveFailed 

    p, q = G 

    x, y = opt.gens 

    p = Poly(p, x, expand=False) 

    q = q.ltrim(-1) 

    p_roots = [ rcollect(expr, y) for expr in roots(p).keys() ] 

    q_roots = list(roots(q).keys()) 

    solutions = [] 

    for q_root in q_roots: 

        for p_root in p_roots: 

            solution = (p_root.subs(y, q_root), q_root) 

            solutions.append(solution) 

    return sorted(solutions, key=default_sort_key) 

def solve_generic(polys, opt): 

    """ 

    Solve a generic system of polynomial equations. 

    Returns all possible solutions over C[x_1, x_2, ..., x_m] of a 

    set F = { f_1, f_2, ..., f_n } of polynomial equations,  using 

    Groebner basis approach. For now only zero-dimensional systems 

    are supported, which means F can have at most a finite number 

    of solutions. 

    The algorithm works by the fact that, supposing G is the basis 

    of F with respect to an elimination order  (here lexicographic 

    order is used), G and F generate the same ideal, they have the 

    same set of solutions. By the elimination property,  if G is a 

    reduced, zero-dimensional Groebner basis, then there exists an 

    univariate polynomial in G (in its last variable). This can be 

    solved by computing its roots. Substituting all computed roots 

    for the last (eliminated) variable in other elements of G, new 

    polynomial system is generated. Applying the above procedure 

    recursively, a finite number of solutions can be found. 

    The ability of finding all solutions by this procedure depends 

    on the root finding algorithms. If no solutions were found, it 

    means only that roots() failed, but the system is solvable. To 

    overcome this difficulty use numerical algorithms instead. 

    References 

    ========== 

    .. [Buchberger01] B. Buchberger, Groebner Bases: A Short 

    Introduction for Systems Theorists, In: R. Moreno-Diaz, 

    B. Buchberger, J.L. Freire, Proceedings of EUROCAST'01, 

    February, 2001 

    .. [Cox97] D. Cox, J. Little, D. O'Shea, Ideals, Varieties 

    and Algorithms, Springer, Second Edition, 1997, pp. 112 

    Examples 

    ======== 

    >>> from sympy.polys import Poly, Options 

    >>> from sympy.solvers.polysys import solve_generic 

    >>> from sympy.abc import x, y 

    >>> NewOption = Options((x, y), {'domain': 'ZZ'}) 

    >>> a = Poly(x - y + 5, x, y, domain='ZZ') 

    >>> b = Poly(x + y - 3, x, y, domain='ZZ') 

    >>> solve_generic([a, b], NewOption) 

    [(-1, 4)] 

    >>> a = Poly(x - 2*y + 5, x, y, domain='ZZ') 

    >>> b = Poly(2*x - y - 3, x, y, domain='ZZ') 

    >>> solve_generic([a, b], NewOption) 

    [(11/3, 13/3)] 

    >>> a = Poly(x**2 + y, x, y, domain='ZZ') 

    >>> b = Poly(x + y*4, x, y, domain='ZZ') 

    >>> solve_generic([a, b], NewOption) 

    [(0, 0), (1/4, -1/16)] 

    """ 

    def _is_univariate(f): 

        """Returns True if 'f' is univariate in its last variable. """ 

        for monom in f.monoms(): 

            if any(m > 0 for m in monom[:-1]): 

                return False 

        return True 

    def _subs_root(f, gen, zero): 

        """Replace generator with a root so that the result is nice. """ 

        p = f.as_expr({gen: zero}) 

        if f.degree(gen) >= 2: 

            p = p.expand(deep=False) 

        return p 

    def _solve_reduced_system(system, gens, entry=False): 

        """Recursively solves reduced polynomial systems. """ 

        if len(system) == len(gens) == 1: 

            zeros = list(roots(system[0], gens[-1]).keys()) 

            return [ (zero,) for zero in zeros ] 

        basis = groebner(system, gens, polys=True) 

        if len(basis) == 1 and basis[0].is_ground: 

            if not entry: 

                return [] 

            else: 

                return None 

        univariate = list(filter(_is_univariate, basis)) 

        if len(univariate) == 1: 

            f = univariate.pop() 

        else: 

            raise NotImplementedError("only zero-dimensional systems supported (finite number of solutions)") 

        gens = f.gens 

        gen = gens[-1] 

        zeros = list(roots(f.ltrim(gen)).keys()) 

        if not zeros: 

            return [] 

        if len(basis) == 1: 

            return [ (zero,) for zero in zeros ] 

        solutions = [] 

        for zero in zeros: 

            new_system = [] 

            new_gens = gens[:-1] 

            for b in basis[:-1]: 

                eq = _subs_root(b, gen, zero) 

                if eq is not S.Zero: 

                    new_system.append(eq) 

            for solution in _solve_reduced_system(new_system, new_gens): 

                solutions.append(solution + (zero,)) 

        return solutions 

    try: 

        result = _solve_reduced_system(polys, opt.gens, entry=True) 

    except CoercionFailed: 

        raise NotImplementedError 

    if result is not None: 

        return sorted(result, key=default_sort_key) 

    else: 

        return None 

def solve_triangulated(polys, *gens, **args): 

    """ 

    Solve a polynomial system using Gianni-Kalkbrenner algorithm. 

    The algorithm proceeds by computing one Groebner basis in the ground 

    domain and then by iteratively computing polynomial factorizations in 

    appropriately constructed algebraic extensions of the ground domain. 

    Examples 

    ======== 

    >>> from sympy.solvers.polysys import solve_triangulated 

    >>> from sympy.abc import x, y, z 

    >>> F = [x**2 + y + z - 1, x + y**2 + z - 1, x + y + z**2 - 1] 

    >>> solve_triangulated(F, x, y, z) 

    [(0, 0, 1), (0, 1, 0), (1, 0, 0)] 

    References 

    ========== 

    1. Patrizia Gianni, Teo Mora, Algebraic Solution of System of 

    Polynomial Equations using Groebner Bases, AAECC-5 on Applied Algebra, 

    Algebraic Algorithms and Error-Correcting Codes, LNCS 356 247--257, 1989 

    """ 

    G = groebner(polys, gens, polys=True) 

    G = list(reversed(G)) 

    domain = args.get('domain') 

    if domain is not None: 

        for i, g in enumerate(G): 

            G[i] = g.set_domain(domain) 

    f, G = G[0].ltrim(-1), G[1:] 

    dom = f.get_domain() 

    zeros = f.ground_roots() 

    solutions = set([]) 

    for zero in zeros: 

        solutions.add(((zero,), dom)) 

    var_seq = reversed(gens[:-1]) 

    vars_seq = postfixes(gens[1:]) 

    for var, vars in zip(var_seq, vars_seq): 

        _solutions = set([]) 

        for values, dom in solutions: 

            H, mapping = [], list(zip(vars, values)) 

            for g in G: 

                _vars = (var,) + vars 

                if g.has_only_gens(*_vars) and g.degree(var) != 0: 

                    h = g.ltrim(var).eval(dict(mapping)) 

                    if g.degree(var) == h.degree(): 

                        H.append(h) 

            p = min(H, key=lambda h: h.degree()) 

            zeros = p.ground_roots() 

            for zero in zeros: 

                if not zero.is_Rational: 

                    dom_zero = dom.algebraic_field(zero) 

                else: 

                    dom_zero = dom 

                _solutions.add(((zero,) + values, dom_zero)) 

        solutions = _solutions 

    solutions = list(solutions) 

    for i, (solution, _) in enumerate(solutions): 

        solutions[i] = solution 

    return sorted(solutions, key=default_sort_key)