Skip to content

Commit

Permalink
[Jenkins] auto-formatting by clang-format version 10.0.0-4ubuntu1
Browse files Browse the repository at this point in the history
  • Loading branch information
@stan-buildbot
stan-buildbot committed Sep 28, 2024
1 parent 572dcd4 commit 00a930e
Show file tree
Hide file tree
Showing 2 changed files with 58 additions and 35 deletions.
4 changes: 2 additions & 2 deletions stan/math/prim/prob/wiener4_lccdf.hpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -196,11 +196,11 @@ inline auto wiener4_ccdf_grad_w(const T_y& y, const T_a& a, const T_v& v,
auto prob_grad_w = ret_t(1 / w);
const bool exp_sign = (v > 0) ? -1 : 1;
const auto exponent = exp_sign * 2.0 * v * a * w;
prob_grad_w = exp(LOG_TWO + log(fabs(v)) + log(a) - log1m_exp(exponent));
prob_grad_w = exp(LOG_TWO + log(fabs(v)) + log(a) - log1m_exp(exponent));
if (exp_sign == -1) {
prob_grad_w *= exp(exponent);
}

const auto cdf_grad_w = wiener4_cdf_grad_w(y, a, v, w, cdf, err);
return prob_grad_w * exp(prob) - cdf_grad_w;
}
Expand Down
89 changes: 56 additions & 33 deletions stan/math/prim/prob/wiener4_lcdf.hpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -6,7 +6,7 @@
namespace stan {
namespace math {
namespace internal {

/**
* Helper function. Log of Mill's ratio for the normal distribution
*
Expand Down Expand Up @@ -69,7 +69,7 @@ inline auto log_probability_GradAV(const T_a& a, const T_v& v,
return ret_t(-w);
}
auto nearly_one = ret_t(1.0 - 1.1 * 1.0e-5);
// auto nearly_one = ret_t(1.0 - std::numeric_limits<ret_t>::min());
// auto nearly_one = ret_t(1.0 - std::numeric_limits<ret_t>::min());
ret_t prob;
if (v < 0) {
const auto two_va_one_minus_w = (2.0 * v * a * (1.0 - w));
Expand All @@ -86,10 +86,10 @@ inline auto log_probability_GradAV(const T_a& a, const T_v& v,
auto log_quotient = log1p(-exp_two_avw) - log1p(-exp_two_av);
if (log(w) > log_quotient) {
prob += log_diff_exp(log(w), log_quotient);
return exp(prob);
return exp(prob);
} else {
prob += log_diff_exp(log_quotient, log(w));
return -exp(prob);
return -exp(prob);
}
} else {
const auto minus_two_va_one_minus_w = (-2.0 * v * a * (1.0 - w));
Expand All @@ -113,10 +113,10 @@ inline auto log_probability_GradAV(const T_a& a, const T_v& v,
}
if (log(w) > log_quotient) {
prob += log_diff_exp(log(w), log_quotient);
return -exp(prob);
return -exp(prob);
} else {
prob += log_diff_exp(log_quotient, log(w));
return exp(prob);
return exp(prob);
}
}
}
Expand Down Expand Up @@ -191,7 +191,7 @@ inline auto wiener4_distribution(const T_y& y, const T_a& a, const T_v& vn,
ans = NEGATIVE_INFTY;
}
log_distribution = ans + -v * a * w - square(v) * y / 2;
return NaturalScale ? exp(log_distribution) : log_distribution;
return NaturalScale ? exp(log_distribution) : log_distribution;
} else {
auto summand_1 = log_probability_distribution(a, v, w);

Expand Down Expand Up @@ -228,9 +228,9 @@ inline auto wiener4_distribution(const T_y& y, const T_a& a, const T_v& vn,
} else if (summand_1 < summand_2) {
log_distribution = log_diff_exp(summand_2, summand_1);
}
return NaturalScale ? exp(log_distribution) : log_distribution;
return NaturalScale ? exp(log_distribution) : log_distribution;
}
// return NaturalScale ? exp(log_distribution) : log_distribution;
// return NaturalScale ? exp(log_distribution) : log_distribution;
}

/**
Expand Down Expand Up @@ -283,33 +283,38 @@ inline auto wiener4_cdf_grad_a(const T_y& y, const T_a& a, const T_v& vn,
auto d_k = std_normal_lpdf(r_k / sqrt_y);
auto x = r_k - vy;
auto xsqrt_y = x / sqrt_y;
auto temp = min(exp(d_k + logMill(xsqrt_y)), std::numeric_limits<ret_t>::max());
auto temp
= min(exp(d_k + logMill(xsqrt_y)), std::numeric_limits<ret_t>::max());
const auto factor = (2 * k + w);
const auto factor_2 = (2 * k + 2.0 - w);
auto temp2 = exp(d_k);
auto temp3 = temp * (-vy) - sqrt_y * temp2;
const auto t1 = temp3 * factor;
x = r_k + vy;
xsqrt_y = x / sqrt_y;
temp = min(exp(d_k + logMill(xsqrt_y)), std::numeric_limits<ret_t>::max());
temp
= min(exp(d_k + logMill(xsqrt_y)), std::numeric_limits<ret_t>::max());
temp3 = temp * vy - sqrt_y * temp2;
const auto t2 = temp3 * factor;
r_k = (2 * k + 1) * a + a * (1 - w);
d_k = std_normal_lpdf(r_k / sqrt_y);
x = r_k - vy;
xsqrt_y = x / sqrt_y;
temp = min(exp(d_k + logMill(xsqrt_y)), std::numeric_limits<ret_t>::max());
temp
= min(exp(d_k + logMill(xsqrt_y)), std::numeric_limits<ret_t>::max());
temp2 = exp(d_k);
temp3 = temp * (-vy) - sqrt_y * temp2;
const auto t3 = -temp3 * factor_2;
x = r_k + vy;
xsqrt_y = x / sqrt_y;
temp = min(exp(d_k + logMill(xsqrt_y)), std::numeric_limits<ret_t>::max());
temp
= min(exp(d_k + logMill(xsqrt_y)), std::numeric_limits<ret_t>::max());
temp3 = temp * vy - sqrt_y * temp2;
const auto t4 = -temp3 * factor;
ans += (t1 + t2 + t3 + t4);
}
F_k = min(exp(v * a * w + 0.5 * square(v) * y), std::numeric_limits<ret_t>::max());
F_k = min(exp(v * a * w + 0.5 * square(v) * y),
std::numeric_limits<ret_t>::max());
const auto summands_small_y = ans / y / F_k;
return -v * w * cdf + summands_small_y;
} else {
Expand All @@ -327,7 +332,8 @@ inline auto wiener4_cdf_grad_a(const T_y& y, const T_a& a, const T_v& vn,
ans += -last * factor;
}
const auto evaw = exp(-v * a * w - 0.5 * square(v) * y);
const auto prob = min(exp(log_probability_distribution(a, v, w)), std::numeric_limits<ret_t>::max());
const auto prob = min(exp(log_probability_distribution(a, v, w)),
std::numeric_limits<ret_t>::max());
const auto dav = log_probability_GradAV(a, v, w);
const auto pia2 = 2 * pi() / square(a);
auto prob_deriv
Expand Down Expand Up @@ -363,10 +369,12 @@ inline auto wiener4_cdf_grad_v(const T_y& y, const T_a& a, const T_v& vn,
const auto log_a = log(a);
auto K_large_value = ret_t(1.0);
if (v != 0) {
const auto temp = -min(exp(log_a - LOG_PI - 0.5 * log_y), std::numeric_limits<ret_t>::max());
const auto temp = -min(exp(log_a - LOG_PI - 0.5 * log_y),
std::numeric_limits<ret_t>::max());
const auto log_v = log(fabs(v));
auto alphK_large = min(exp(factor + 0.5 * (7 * LOG_PI + log_y) - 2.5 * LOG_TWO
- 3 * log_a - log_v), std::numeric_limits<ret_t>::max());
auto alphK_large = min(exp(factor + 0.5 * (7 * LOG_PI + log_y)
- 2.5 * LOG_TWO - 3 * log_a - log_v),
std::numeric_limits<ret_t>::max());
alphK_large = fmax(0.0, fmin(1.0, alphK_large));
K_large_value
= fmax(ceil((alphK_large == 0)
Expand Down Expand Up @@ -394,27 +402,32 @@ inline auto wiener4_cdf_grad_v(const T_y& y, const T_a& a, const T_v& vn,
auto d_k = std_normal_lpdf(r_k / sqrt_y);
auto x = r_k - vy;
auto xsqrt_y = x / sqrt_y;
auto temp = min(exp(d_k + logMill(xsqrt_y)), std::numeric_limits<ret_t>::max());
auto temp
= min(exp(d_k + logMill(xsqrt_y)), std::numeric_limits<ret_t>::max());
const auto factor = 2 * k + w;
const auto factor_2 = 2 * k + 2.0 - w;
const auto t1 = -temp * x;
x = r_k + vy;
xsqrt_y = x / sqrt_y;
temp = min(exp(d_k + logMill(xsqrt_y)), std::numeric_limits<ret_t>::max());
temp
= min(exp(d_k + logMill(xsqrt_y)), std::numeric_limits<ret_t>::max());
const auto t2 = temp * x;
r_k = (2 * k + 1) * a + a * (1 - w);
d_k = std_normal_lpdf(r_k / sqrt_y);
x = r_k - vy;
xsqrt_y = x / sqrt_y;
temp = min(exp(d_k + logMill(xsqrt_y)), std::numeric_limits<ret_t>::max());
temp
= min(exp(d_k + logMill(xsqrt_y)), std::numeric_limits<ret_t>::max());
const auto t3 = temp * x;
x = r_k + vy;
xsqrt_y = x / sqrt_y;
temp = min(exp(d_k + logMill(xsqrt_y)), std::numeric_limits<ret_t>::max());
temp
= min(exp(d_k + logMill(xsqrt_y)), std::numeric_limits<ret_t>::max());
const auto t4 = -temp * x;
ans += (t1 + t2 + t3 + t4);
}
F_k = min(exp(v * a * w + 0.5 * square(v) * y), std::numeric_limits<ret_t>::max());
F_k = min(exp(v * a * w + 0.5 * square(v) * y),
std::numeric_limits<ret_t>::max());
const auto summands_small_y = ans / F_k;
return -1 * ((-w * a - v * y) * cdf + summands_small_y);
} else {
Expand All @@ -432,7 +445,8 @@ inline auto wiener4_cdf_grad_v(const T_y& y, const T_a& a, const T_v& vn,
ans += -last * factor;
}
const auto evaw = exp(-v * a * w - 0.5 * square(v) * y);
const auto prob = min(exp(log_probability_distribution(a, v, w)), std::numeric_limits<ret_t>::max());
const auto prob = min(exp(log_probability_distribution(a, v, w)),
std::numeric_limits<ret_t>::max());
const auto dav = log_probability_GradAV(a, v, w);
const auto pia2 = 2 * pi() / square(a);
auto prob_deriv = is_inf(dav * a) ? NEGATIVE_INFTY : dav * a;
Expand Down Expand Up @@ -465,9 +479,11 @@ inline auto wiener4_cdf_grad_w(const T_y& y, const T_a& a, const T_v& vn,

const auto log_y = log(y);
const auto log_a = log(a);
const auto temp = -min(exp(log_a - LOG_PI - 0.5 * log_y), std::numeric_limits<ret_t>::max());
const auto temp = -min(exp(log_a - LOG_PI - 0.5 * log_y),
std::numeric_limits<ret_t>::max());
auto alphK_large
= min(exp(factor + 0.5 * (LOG_PI + log_y) - 1.5 * LOG_TWO - log_a), std::numeric_limits<ret_t>::max());
= min(exp(factor + 0.5 * (LOG_PI + log_y) - 1.5 * LOG_TWO - log_a),
std::numeric_limits<ret_t>::max());
alphK_large = fmax(0.0, fmin(1.0, alphK_large));
const auto K_large_value
= fmax(ceil((alphK_large == 0)
Expand All @@ -481,7 +497,8 @@ inline auto wiener4_cdf_grad_w(const T_y& y, const T_a& a, const T_v& vn,
const auto K_large = fabs(v) / a * y - wdash;
const auto lv = log1p(square(v) * y);
const auto alphK_small = factor - LOG_TWO - lv;
const auto arg = fmin(min(exp(alphK_small), std::numeric_limits<ret_t>::max()), 1.0);
const auto arg
= fmin(min(exp(alphK_small), std::numeric_limits<ret_t>::max()), 1.0);
const auto K_small
= (arg == 0)
? INFTY
Expand All @@ -498,33 +515,38 @@ inline auto wiener4_cdf_grad_w(const T_y& y, const T_a& a, const T_v& vn,
auto d_k = std_normal_lpdf(r_k / sqrt_y);
auto x = r_k - vy;
auto xsqrt_y = x / sqrt_y;
auto temp = min(exp(d_k + logMill(xsqrt_y)), std::numeric_limits<ret_t>::max());
auto temp
= min(exp(d_k + logMill(xsqrt_y)), std::numeric_limits<ret_t>::max());
const auto factor = a;
const auto factor_2 = -a;
auto temp2 = exp(d_k);
auto temp3 = temp * (-vy) - sqrt_y * temp2;
const auto t1 = temp3 * factor;
x = r_k + vy;
xsqrt_y = x / sqrt_y;
temp = min(exp(d_k + logMill(xsqrt_y)), std::numeric_limits<ret_t>::max());
temp
= min(exp(d_k + logMill(xsqrt_y)), std::numeric_limits<ret_t>::max());
temp3 = temp * vy - sqrt_y * temp2;
const auto t2 = temp3 * factor;
r_k = (2 * k + 1) * a + a * (1 - w);
d_k = std_normal_lpdf(r_k / sqrt_y);
x = r_k - vy;
xsqrt_y = x / sqrt_y;
temp = min(exp(d_k + logMill(xsqrt_y)), std::numeric_limits<ret_t>::max());
temp
= min(exp(d_k + logMill(xsqrt_y)), std::numeric_limits<ret_t>::max());
temp2 = exp(d_k);
temp3 = temp * (-vy) - sqrt_y * temp2;
const auto t3 = -temp3 * factor_2;
x = r_k + vy;
xsqrt_y = x / sqrt_y;
temp = min(exp(d_k + logMill(xsqrt_y)), std::numeric_limits<ret_t>::max());
temp
= min(exp(d_k + logMill(xsqrt_y)), std::numeric_limits<ret_t>::max());
temp3 = temp * vy - sqrt_y * temp2;
const auto t4 = -temp3 * factor;
ans += (t1 + t2 + t3 + t4);
}
F_k = min(exp(v * a * w + 0.5 * square(v) * y), std::numeric_limits<ret_t>::max());
F_k = min(exp(v * a * w + 0.5 * square(v) * y),
std::numeric_limits<ret_t>::max());
const auto summands_small_y = ans / y / F_k;
return -1 * (-v * a * cdf + summands_small_y);
} else {
Expand All @@ -542,7 +564,8 @@ inline auto wiener4_cdf_grad_w(const T_y& y, const T_a& a, const T_v& vn,
ans += -last * factor;
}
const auto evaw = exp(-v * a * w - 0.5 * square(v) * y);
const auto prob = min(exp(log_probability_distribution(a, v, w)), std::numeric_limits<ret_t>::max());
const auto prob = min(exp(log_probability_distribution(a, v, w)),
std::numeric_limits<ret_t>::max());

// Calculate the probability term 'P' on log scale
auto dav = ret_t(-1 / (1.0 - w));
Expand Down

0 comments on commit 00a930e

Please sign in to comment.