I have the following system
$$-a_1^2 + 2 a_0 a_2 - 2 b_0 b_2 =0\quad \text{and } \quad a_2^2 - 2 a_1 a_3 + 2 a_0 a_4 - b_2^2 - 2 b_0 b_4 =0$$
where the following $a_i, b_i$ are integral of some heavy function defined before in the notebook. I can give some example, but they basically depend on hypertrigonometric function.
$$ a_i= \int dk f_i(k,c,s) \quad b_i= \int dk g_i(k,c,s) $$
And I am interested in solving the system for $c,s$.
F[m_?NumericQ, p_?NumericQ] := Quiet[Chop[
-NIntegrate[f1[k,c,s]/. { c -> m, s -> p}, {k, 0,
Infinity} , WorkingPrecision -> 20, AccuracyGoal -> 50 ]*
NIntegrate[
f1[k,c,s]/. {c -> m, s -> p}, {k, 0,
Infinity}, WorkingPrecision -> 20, AccuracyGoal -> 50 ] +
2 NIntegrate[
f0[k,c,s]/. { c -> m, s -> p}, {k, 0,
Infinity}, WorkingPrecision -> 20, AccuracyGoal -> 50 ] *
NIntegrate[
f2[k,c,s] /. {c -> m, s -> p}, {k, 0,
Infinity}, WorkingPrecision -> 20, AccuracyGoal -> 50 ] -
2 NIntegrate[
g0[k,c,s] /. { c -> m, s -> p}, {k, 0,
Infinity}, WorkingPrecision -> 20, AccuracyGoal -> 50 ] *
NIntegrate[
g2[k,c,s] /. {c -> m, s -> p}, {k, 0,
Infinity}, WorkingPrecision -> 20, AccuracyGoal -> 50 ]
]];
G[m_?NumericQ, p_?NumericQ] := Quiet[Chop[
NIntegrate[
f2[k,c,s] /. {c -> m, s -> p}, {k, 0,
Infinity}, WorkingPrecision -> 20, AccuracyGoal -> 50]*
NIntegrate[
f2[k,c,s] /. { c -> m, s -> p}, {k, 0,
Infinity}, WorkingPrecision -> 20, AccuracyGoal -> 50] -
2 NIntegrate[
f1[k,c,s]/. { c -> m, s -> p}, {k, 0,
Infinity}, WorkingPrecision -> 20, AccuracyGoal -> 50 ] *
NIntegrate[
f3[k,c,s] /. { c -> m, s -> p}, {k, 0,
Infinity}] +
2 NIntegrate[
f0[k,c,s] /. {c -> m, s -> p}, {k, 0,
Infinity}, WorkingPrecision -> 20, AccuracyGoal -> 50] *
NIntegrate[
f4[k,c,s] /. { c -> m, s -> p}, {k, 0,
Infinity}] -
NIntegrate[
g2[k,c,s] /. { c -> m, s -> p}, {k, 0,
Infinity}, WorkingPrecision -> 20, AccuracyGoal -> 50]*
NIntegrate[
g2[k,c,s] /. { c -> m, s -> p}, {k, 0,
Infinity}, WorkingPrecision -> 20, AccuracyGoal -> 50] -
2*NIntegrate[
g0[k,c,s]/. { c -> m, s -> p}, {k, 0,
Infinity}, WorkingPrecision -> 20, AccuracyGoal -> 50]*
NIntegrate[
g4[k,c,s]/. {c -> m, s -> p}, {k, 0,
Infinity}, WorkingPrecision -> 30, AccuracyGoal -> 50]
]];
FindRoot[{Chop[F[c, s]] == 0, Chop[G[c, s]] ==0}, {{c, 0.8}, {s, 0.01}}
but FindRoot struggle to find a solution (but I know one should exist around 0.81 and 0.0128), I think it is because of the operations between numerical integrations. Unfortunately, it is not possible to solve analytically, is there a way to combine FindRoot with multiplication and soustraction of NIntegrate?
FindRoot: The line search decreased the step size to within tolerance specified \ by AccuracyGoal and PrecisionGoal but was unable to find a sufficient \ decrease in the merit function. You may need more than
MachinePrecision digits of working precision to meet these tolerances.
I already used FindRoot with a single NIntegrate==0 inside, and it worked well, but here product of NIntegrate must propagate a big error.
f[1][k,m,p]orf1[k,m,p]instead off_1(k,c,s)/. {c -> m, s -> p}After you've fixed that, do a single example to make sure that your numerical integrations are returning something sensible. Also, try plotting yourF[c,s]to see if FindRoot has a chance of finding a root. $\endgroup$