@yarchik's answer is correct that it is not a bug. I assumed it to be a bug too quickly without really looking what expressions contained and what documentation says about possible issues.
But anyway Reduce
can in general solve even such systems but we need to add that variables are real and we are solving over Complexes
.
The same is with FindInstance
.
This returns {}
because no solutions could be found.
FindInstance[(1/(2 a (-2 r + c^2 (1 + r)) (-1 + x)) (r -
a c (2 + r) - 3 a r (-1 + x) + c^2 (-r + a (2 + r) x) -
Sqrt[-1 + c] Sqrt[
r] Sqrt[(-1 + c) (1 + c)^2 r +
2 a (r + c (2 - (2 + c) r) + (-1 + c) (1 + c)^2 r x) +
a^2 (-8 + 4 c - r + 5 c r +
2 (r + c (2 - (2 + c) r)) x + (-1 +
c) (1 + c)^2 r x^2)])) > 1 && (c - a (1 - c))/c > r >
a/(1 + a - c) && 1 > c > a > 0, {a, c, r, x}]
{}
But if we add ...&& {a, c, r, x} \[Element] Reals, {a, c, r, x}, Complexes]
a solution is found.
FindInstance[(1/(2 a (-2 r + c^2 (1 + r)) (-1 + x)) (r -
a c (2 + r) - 3 a r (-1 + x) + c^2 (-r + a (2 + r) x) -
Sqrt[-1 + c] Sqrt[
r] Sqrt[(-1 + c) (1 + c)^2 r +
2 a (r + c (2 - (2 + c) r) + (-1 + c) (1 + c)^2 r x) +
a^2 (-8 + 4 c - r + 5 c r +
2 (r + c (2 - (2 + c) r)) x + (-1 +
c) (1 + c)^2 r x^2)])) > 1 && (c - a (1 - c))/c > r >
a/(1 + a - c) &&
1 > c > a > 0 && {a, c, r, x} \[Element] Reals, {a, c, r,
x}, Complexes]
{{a -> 1/15, c -> 1/15 (1 + Sqrt[31]), r -> 46/125, x -> -17}}
So same should be possible with Reduce
. (it took some time as the system is complicated but solution was found).
Reduce[(1/(2 a (-2 r + c^2 (1 + r)) (-1 + x)) (r - a c (2 + r) -
3 a r (-1 + x) + c^2 (-r + a (2 + r) x) -
Sqrt[-1 + c] Sqrt[
r] Sqrt[(-1 + c) (1 + c)^2 r +
2 a (r + c (2 - (2 + c) r) + (-1 + c) (1 + c)^2 r x) +
a^2 (-8 + 4 c - r + 5 c r +
2 (r + c (2 - (2 + c) r)) x + (-1 +
c) (1 + c)^2 r x^2)])) > 1 && (c - a (1 - c))/c > r >
a/(1 + a - c) &&
1 > c > a > 0 && {a, c, r, x} \[Element] Reals, {a, c, r,
x}, Complexes]
(0 < a < 1/
4 && ((a < c <= a + Sqrt[2 a + a^2] &&
a/(1 + a - c) < r < (-a + c + a c)/
c && (-a + a c - c r + a c r)/(a c r) < x <
1) || (a + Sqrt[2 a + a^2] < c <
1 && ((a/(1 + a - c) <
r < -(c^2/(-2 + c^2)) && ((-a + a c - c r + a c r)/(
a c r) <
x <= (-2 a c + r - a r + c r + 2 a c r - c^2 r +
a c^2 r - c^3 r)/(a (-1 + c) (1 + c)^2 r) -
2 Sqrt[-((-c^2 + 2 r - c^2 r + c^4 r - 2 c^2 r^2 +
c^4 r^2)/((-1 + c)^2 (1 + c)^4 r^2))] ||
x > 1)) || (-(c^2/(-2 + c^2)) < r < (-a + c + a c)/
c && (-a + a c - c r + a c r)/(a c r) < x < 1))))) || (1/
4 <= a < 1 && a < c < 1 &&
a/(1 + a - c) < r < (-a + c + a c)/c && (-a + a c - c r + a c r)/(
a c r) < x < 1)
Reduce
is reliable most of the time. $\endgroup$Reduce
works over the reals. However, there may be complex solutions. To find these, the domainComplexes
should be explicitly indicated. $\endgroup$