diff --git a/lib/node_modules/@stdlib/math/base/special/ldexp/lib/main.js b/lib/node_modules/@stdlib/math/base/special/ldexp/lib/main.js
index 3c033a9f105f..85bbf11d0595 100644
--- a/lib/node_modules/@stdlib/math/base/special/ldexp/lib/main.js
+++ b/lib/node_modules/@stdlib/math/base/special/ldexp/lib/main.js
@@ -44,8 +44,8 @@ var fromWords = require( '@stdlib/number/float64/base/from-words' );
 
 // VARIABLES //
 
-// 1/(1<<52) = 1/(2**52) = 1/4503599627370496
-var TWO52_INV = 2.220446049250313e-16;
+// 1/(1<<54) = 1/(2**54) = 1/18014398509481984
+var TWO54_INV = 5.551115123125783e-17;
 
 // Exponent all 0s: 1 00000000000 11111111111111111111 => 2148532223
 var CLEAR_EXP_MASK = 0x800fffff>>>0; // asm type annotation
@@ -113,8 +113,8 @@ function ldexp( frac, exp ) {
 	// Extract the exponent from `frac` and add it to `exp`:
 	exp += floatExp( frac );
 
-	// Check for underflow/overflow...
-	if ( exp < MIN_SUBNORMAL_EXPONENT ) {
+	// Check for underflow/overflow (note: values on the interval `[2^-1075, 2^-1074)` may round to the smallest subnormal)...
+	if ( exp < MIN_SUBNORMAL_EXPONENT-1 ) {
 		return copysign( 0.0, frac );
 	}
 	if ( exp > MAX_EXPONENT ) {
@@ -125,8 +125,8 @@ function ldexp( frac, exp ) {
 	}
 	// Check for a subnormal and scale accordingly to retain precision...
 	if ( exp <= MAX_SUBNORMAL_EXPONENT ) {
-		exp += 52;
-		m = TWO52_INV;
+		exp += 54;
+		m = TWO54_INV;
 	} else {
 		m = 1.0;
 	}
diff --git a/lib/node_modules/@stdlib/math/base/special/ldexp/src/main.c b/lib/node_modules/@stdlib/math/base/special/ldexp/src/main.c
index 6c209d16baad..482de7105f2e 100644
--- a/lib/node_modules/@stdlib/math/base/special/ldexp/src/main.c
+++ b/lib/node_modules/@stdlib/math/base/special/ldexp/src/main.c
@@ -32,8 +32,8 @@
 #include "stdlib/constants/float64/exponent_bias.h"
 #include <stdint.h>
 
-// 1/(1<<52) = 1/(2**52) = 1/4503599627370496
-static const double TWO52_INV = 2.220446049250313e-16;
+// 1/(1<<54) = 1/(2**54) = 1/18014398509481984
+static const double TWO54_INV = 5.551115123125783e-17;
 
 // Exponent all 0s: 1 00000000000 11111111111111111111 => 2148532223
 static const uint32_t CLEAR_EXP_MASK = 2148532223;
@@ -64,8 +64,8 @@ double stdlib_base_ldexp( const double frac, const int32_t exp ) {
 	// Extract the exponent from `frac` and add it to `exp`:
 	e += exp + stdlib_base_float64_exponent( y );
 
-	// Check for underflow/overflow...
-	if ( e < STDLIB_CONSTANT_FLOAT64_MIN_BASE2_EXPONENT_SUBNORMAL ) {
+	// Check for underflow/overflow (note: values on the interval `[2^-1075, 2^-1074)` may round to the smallest subnormal)...
+	if ( e < STDLIB_CONSTANT_FLOAT64_MIN_BASE2_EXPONENT_SUBNORMAL-1 ) {
 		return stdlib_base_copysign( 0.0, y );
 	}
 	if ( e > STDLIB_CONSTANT_FLOAT64_MAX_BASE2_EXPONENT ) {
@@ -76,8 +76,8 @@ double stdlib_base_ldexp( const double frac, const int32_t exp ) {
 	}
 	// Check for a subnormal and scale accordingly to retain precision...
 	if ( e <= STDLIB_CONSTANT_FLOAT64_MAX_BASE2_EXPONENT_SUBNORMAL ) {
-		e += 52;
-		m = TWO52_INV;
+		e += 54;
+		m = TWO54_INV;
 	} else {
 		m = 1.0;
 	}
diff --git a/lib/node_modules/@stdlib/math/base/special/ldexp/test/test.js b/lib/node_modules/@stdlib/math/base/special/ldexp/test/test.js
index 349f4a22bcd3..929072166e0c 100644
--- a/lib/node_modules/@stdlib/math/base/special/ldexp/test/test.js
+++ b/lib/node_modules/@stdlib/math/base/special/ldexp/test/test.js
@@ -113,6 +113,30 @@ tape( 'the function multiplies a number by an integer power of two (subnormals)'
 	t.end();
 });
 
+tape( 'the function rounds results on the interval `[2^-1075, 2^-1074)` to the nearest representable value', function test( t ) {
+	var v;
+
+	v = ldexp( 0.75, -1074 );
+	t.strictEqual( v, 5.0e-324, 'returns expected value' );
+
+	v = ldexp( -0.75, -1074 );
+	t.strictEqual( v, -5.0e-324, 'returns expected value' );
+
+	v = ldexp( 3.0, -1076 );
+	t.strictEqual( v, 5.0e-324, 'returns expected value' );
+
+	v = ldexp( 0.5000000000000001, -1074 );
+	t.strictEqual( v, 5.0e-324, 'returns expected value' );
+
+	// Ties round to even:
+	v = ldexp( 0.5, -1074 );
+	t.strictEqual( isPositiveZero( v ), true, 'returns expected value' );
+
+	v = ldexp( -0.5, -1074 );
+	t.strictEqual( isNegativeZero( v ), true, 'returns expected value' );
+	t.end();
+});
+
 tape( 'if provided a fraction equal to `+0`, the function returns `+0`', function test( t ) {
 	var v = ldexp( 0.0, 10 );
 	t.strictEqual( isPositiveZero( v ), true, 'returns expected value' );
diff --git a/lib/node_modules/@stdlib/math/base/special/ldexp/test/test.native.js b/lib/node_modules/@stdlib/math/base/special/ldexp/test/test.native.js
index bad8c91cc85d..a4e29769621b 100644
--- a/lib/node_modules/@stdlib/math/base/special/ldexp/test/test.native.js
+++ b/lib/node_modules/@stdlib/math/base/special/ldexp/test/test.native.js
@@ -122,6 +122,30 @@ tape( 'the function multiplies a number by an integer power of two (subnormals)'
 	t.end();
 });
 
+tape( 'the function rounds results on the interval `[2^-1075, 2^-1074)` to the nearest representable value', opts, function test( t ) {
+	var v;
+
+	v = ldexp( 0.75, -1074 );
+	t.strictEqual( v, 5.0e-324, 'returns expected value' );
+
+	v = ldexp( -0.75, -1074 );
+	t.strictEqual( v, -5.0e-324, 'returns expected value' );
+
+	v = ldexp( 3.0, -1076 );
+	t.strictEqual( v, 5.0e-324, 'returns expected value' );
+
+	v = ldexp( 0.5000000000000001, -1074 );
+	t.strictEqual( v, 5.0e-324, 'returns expected value' );
+
+	// Ties round to even:
+	v = ldexp( 0.5, -1074 );
+	t.strictEqual( isPositiveZero( v ), true, 'returns expected value' );
+
+	v = ldexp( -0.5, -1074 );
+	t.strictEqual( isNegativeZero( v ), true, 'returns expected value' );
+	t.end();
+});
+
 tape( 'if provided a fraction equal to `+0`, the function returns `+0`', opts, function test( t ) {
 	var v = ldexp( 0.0, 10 );
 	t.strictEqual( isPositiveZero( v ), true, 'returns expected value' );
diff --git a/lib/node_modules/@stdlib/math/base/special/pow/test/test.js b/lib/node_modules/@stdlib/math/base/special/pow/test/test.js
index 8e196a8addf1..af423e6f178a 100644
--- a/lib/node_modules/@stdlib/math/base/special/pow/test/test.js
+++ b/lib/node_modules/@stdlib/math/base/special/pow/test/test.js
@@ -389,7 +389,7 @@ tape( 'the function evaluates the exponential function (near underflow)', functi
 	x = 1.9999999999999998;
 	y = -1075;
 	v = pow( x, y );
-	t.strictEqual( v, 0.0, 'pow('+x+','+y+') returns 0' );
+	t.strictEqual( v, 5.0e-324, 'pow('+x+','+y+') returns 5e-324' );
 
 	x = 1.9999999999999999;
 	y = -1075;
diff --git a/lib/node_modules/@stdlib/math/base/special/pow/test/test.native.js b/lib/node_modules/@stdlib/math/base/special/pow/test/test.native.js
index 77da65e94d9b..a215d5de1adc 100644
--- a/lib/node_modules/@stdlib/math/base/special/pow/test/test.native.js
+++ b/lib/node_modules/@stdlib/math/base/special/pow/test/test.native.js
@@ -401,7 +401,7 @@ tape( 'the function evaluates the exponential function (near underflow)', opts,
 	x = 1.9999999999999998;
 	y = -1075;
 	v = pow( x, y );
-	t.strictEqual( v, 0.0, 'pow('+x+','+y+') returns 0' );
+	t.strictEqual( v, 5.0e-324, 'pow('+x+','+y+') returns 5e-324' );
 
 	x = 1.9999999999999999;
 	y = -1075;