public static function fe_sq2(ParagonIE_Sodium_Core_Curve25519_Fe $f)
    {
        $f = self::fe_normalize($f);
        $f0 = (int) $f[0];
        $f1 = (int) $f[1];
        $f2 = (int) $f[2];
        $f3 = (int) $f[3];
        $f4 = (int) $f[4];
        $f5 = (int) $f[5];
        $f6 = (int) $f[6];
        $f7 = (int) $f[7];
        $f8 = (int) $f[8];
        $f9 = (int) $f[9];

        $f0_2 = $f0 << 1;
        $f1_2 = $f1 << 1;
        $f2_2 = $f2 << 1;
        $f3_2 = $f3 << 1;
        $f4_2 = $f4 << 1;
        $f5_2 = $f5 << 1;
        $f6_2 = $f6 << 1;
        $f7_2 = $f7 << 1;
        $f5_38 = self::mul($f5, 38, 6); /* 1.959375*2^30 */
        $f6_19 = self::mul($f6, 19, 5); /* 1.959375*2^30 */
        $f7_38 = self::mul($f7, 38, 6); /* 1.959375*2^30 */
        $f8_19 = self::mul($f8, 19, 5); /* 1.959375*2^30 */
        $f9_38 = self::mul($f9, 38, 6); /* 1.959375*2^30 */
        $f0f0 = self::mul($f0, $f0, 24);
        $f0f1_2 = self::mul($f0_2, $f1, 24);
        $f0f2_2 = self::mul($f0_2, $f2, 24);
        $f0f3_2 = self::mul($f0_2, $f3, 24);
        $f0f4_2 = self::mul($f0_2, $f4, 24);
        $f0f5_2 = self::mul($f0_2, $f5, 24);
        $f0f6_2 = self::mul($f0_2, $f6, 24);
        $f0f7_2 = self::mul($f0_2, $f7, 24);
        $f0f8_2 = self::mul($f0_2, $f8, 24);
        $f0f9_2 = self::mul($f0_2, $f9, 24);
        $f1f1_2 = self::mul($f1_2,  $f1, 24);
        $f1f2_2 = self::mul($f1_2,  $f2, 24);
        $f1f3_4 = self::mul($f1_2,  $f3_2, 24);
        $f1f4_2 = self::mul($f1_2,  $f4, 24);
        $f1f5_4 = self::mul($f1_2,  $f5_2, 24);
        $f1f6_2 = self::mul($f1_2,  $f6, 24);
        $f1f7_4 = self::mul($f1_2,  $f7_2, 24);
        $f1f8_2 = self::mul($f1_2,  $f8, 24);
        $f1f9_76 = self::mul($f9_38, $f1_2, 24);
        $f2f2 = self::mul($f2,  $f2, 24);
        $f2f3_2 = self::mul($f2_2,  $f3, 24);
        $f2f4_2 = self::mul($f2_2,  $f4, 24);
        $f2f5_2 = self::mul($f2_2,  $f5, 24);
        $f2f6_2 = self::mul($f2_2,  $f6, 24);
        $f2f7_2 = self::mul($f2_2,  $f7, 24);
        $f2f8_38 = self::mul($f8_19, $f2_2, 25);
        $f2f9_38 = self::mul($f9_38, $f2, 24);
        $f3f3_2 = self::mul($f3_2,  $f3, 24);
        $f3f4_2 = self::mul($f3_2,  $f4, 24);
        $f3f5_4 = self::mul($f3_2,  $f5_2, 24);
        $f3f6_2 = self::mul($f3_2,  $f6, 24);
        $f3f7_76 = self::mul($f7_38, $f3_2, 24);
        $f3f8_38 = self::mul($f8_19, $f3_2, 24);
        $f3f9_76 = self::mul($f9_38, $f3_2, 24);
        $f4f4 = self::mul($f4,  $f4, 24);
        $f4f5_2 = self::mul($f4_2,  $f5, 24);
        $f4f6_38 = self::mul($f6_19, $f4_2, 25);
        $f4f7_38 = self::mul($f7_38, $f4, 24);
        $f4f8_38 = self::mul($f8_19, $f4_2, 25);
        $f4f9_38 = self::mul($f9_38, $f4, 24);
        $f5f5_38 = self::mul($f5_38, $f5, 24);
        $f5f6_38 = self::mul($f6_19, $f5_2, 24);
        $f5f7_76 = self::mul($f7_38, $f5_2, 24);
        $f5f8_38 = self::mul($f8_19, $f5_2, 24);
        $f5f9_76 = self::mul($f9_38, $f5_2, 24);
        $f6f6_19 = self::mul($f6_19, $f6, 24);
        $f6f7_38 = self::mul($f7_38, $f6, 24);
        $f6f8_38 = self::mul($f8_19, $f6_2, 25);
        $f6f9_38 = self::mul($f9_38, $f6, 24);
        $f7f7_38 = self::mul($f7_38, $f7, 24);
        $f7f8_38 = self::mul($f8_19, $f7_2, 24);
        $f7f9_76 = self::mul($f9_38, $f7_2, 24);
        $f8f8_19 = self::mul($f8_19, $f8, 24);
        $f8f9_38 = self::mul($f9_38, $f8, 24);
        $f9f9_38 = self::mul($f9_38, $f9, 24);

        $h0 = (int) ($f0f0 + $f1f9_76 + $f2f8_38 + $f3f7_76 + $f4f6_38 + $f5f5_38) << 1;
        $h1 = (int) ($f0f1_2 + $f2f9_38 + $f3f8_38 + $f4f7_38 + $f5f6_38) << 1;
        $h2 = (int) ($f0f2_2 + $f1f1_2  + $f3f9_76 + $f4f8_38 + $f5f7_76 + $f6f6_19) << 1;
        $h3 = (int) ($f0f3_2 + $f1f2_2  + $f4f9_38 + $f5f8_38 + $f6f7_38) << 1;
        $h4 = (int) ($f0f4_2 + $f1f3_4  + $f2f2    + $f5f9_76 + $f6f8_38 + $f7f7_38) << 1;
        $h5 = (int) ($f0f5_2 + $f1f4_2  + $f2f3_2  + $f6f9_38 + $f7f8_38) << 1;
        $h6 = (int) ($f0f6_2 + $f1f5_4  + $f2f4_2  + $f3f3_2  + $f7f9_76 + $f8f8_19) << 1;
        $h7 = (int) ($f0f7_2 + $f1f6_2  + $f2f5_2  + $f3f4_2  + $f8f9_38) << 1;
        $h8 = (int) ($f0f8_2 + $f1f7_4  + $f2f6_2  + $f3f5_4  + $f4f4    + $f9f9_38) << 1;
        $h9 = (int) ($f0f9_2 + $f1f8_2  + $f2f7_2  + $f3f6_2  + $f4f5_2) << 1;

        $carry0 = ($h0 + (1 << 25)) >> 26;
        $h1 += $carry0;
        $h0 -= $carry0 << 26;
        $carry4 = ($h4 + (1 << 25)) >> 26;
        $h5 += $carry4;
        $h4 -= $carry4 << 26;