1 files changed, 78 insertions, 0 deletions
diff --git a/math/aarch64/experimental/erfinvf_4u7.c b/math/aarch64/experimental/erfinvf_4u7.c
new file mode 100644
index 000000000000..152994f6336a
--- /dev/null
+++ b/math/aarch64/experimental/erfinvf_4u7.c
@@ -0,0 +1,78 @@
+/*
+ * Single-precision inverse error function.
+ *
+ * Copyright (c) 2023-2024, Arm Limited.
+ * SPDX-License-Identifier: MIT OR Apache-2.0 WITH LLVM-exception
+ */
+#include "poly_scalar_f32.h"
+#include "math_config.h"
+#include "test_sig.h"
+#include "test_defs.h"
+
+const static struct
+{
+  /*  We use P_N and Q_N to refer to arrays of coefficients, where P_N is the
+      coeffs of the numerator in table N of Blair et al, and Q_N is the coeffs
+      of the denominator.  */
+  float P_10[3], Q_10[4], P_29[4], Q_29[4], P_50[6], Q_50[3];
+} data = { .P_10 = { -0x1.a31268p+3, 0x1.ac9048p+4, -0x1.293ff6p+3 },
+	   .Q_10 = { -0x1.8265eep+3, 0x1.ef5eaep+4, -0x1.12665p+4, 0x1p+0 },
+	   .P_29
+	   = { -0x1.fc0252p-4, 0x1.119d44p+0, -0x1.f59ee2p+0, 0x1.b13626p-2 },
+	   .Q_29 = { -0x1.69952p-4, 0x1.c7b7d2p-1, -0x1.167d7p+1, 0x1p+0 },
+	   .P_50 = { 0x1.3d8948p-3, 0x1.61f9eap+0, 0x1.61c6bcp-1,
+		     -0x1.20c9f2p+0, 0x1.5c704cp-1, -0x1.50c6bep-3 },
+	   .Q_50 = { 0x1.3d7dacp-3, 0x1.629e5p+0, 0x1p+0 } };
+
+/* Inverse error function approximation, based on rational approximation as
+   described in
+   J. M. Blair, C. A. Edwards, and J. H. Johnson,
+   "Rational Chebyshev approximations for the inverse of the error function",
+   Math. Comp. 30, pp. 827--830 (1976).
+   https://doi.org/10.1090/S0025-5718-1976-0421040-7
+   Largest error is 4.71 ULP, in the tail region:
+   erfinvf(0x1.f84e9ap-1) got 0x1.b8326ap+0
+			 want 0x1.b83274p+0.  */
+float
+erfinvf (float x)
+{
+  if (x == 1.0f)
+    return __math_oflowf (0);
+  if (x == -1.0f)
+    return __math_oflowf (1);
+
+  float a = fabsf (x);
+  if (a > 1.0f)
+    return __math_invalidf (x);
+
+  if (a <= 0.75f)
+    {
+      /* Greatest error in this region is 4.60 ULP:
+	 erfinvf(0x1.0a98bap-5) got 0x1.d8a93ep-6
+			       want 0x1.d8a948p-6.  */
+      float t = x * x - 0.5625f;
+      return x * horner_2_f32 (t, data.P_10) / horner_3_f32 (t, data.Q_10);
+    }
+  if (a < 0.9375f)
+    {
+      /* Greatest error in this region is 3.79 ULP:
+	 erfinvf(0x1.ac82d6p-1) got 0x1.f8fc54p-1
+			       want 0x1.f8fc5cp-1.  */
+      float t = x * x - 0.87890625f;
+      return x * horner_3_f32 (t, data.P_29) / horner_3_f32 (t, data.Q_29);
+    }
+
+  /* Tail region, where error is greatest (and sensitive to sqrt and log1p
+     implementations.  */
+  float t = 1.0 / sqrtf (-log1pf (-a));
+  return horner_5_f32 (t, data.P_50)
+	 / (copysignf (t, x) * horner_2_f32 (t, data.Q_50));
+}
+
+#if USE_MPFR
+# warning Not generating tests for erfinvf, as MPFR has no suitable reference
+#else
+TEST_SIG (S, F, 1, erfinv, -0.99, 0.99)
+TEST_ULP (erfinvf, 4.09)
+TEST_SYM_INTERVAL (erfinvf, 0, 1, 40000)
+#endif