net.dermetfan.gdx.math.MathUtils Maven / Gradle / Ivy
Go to download
Show more of this group Show more artifacts with this name
Show all versions of libgdx-utils Show documentation
Show all versions of libgdx-utils Show documentation
support library for libGDX
The newest version!
/** Copyright 2014 Robin Stumm ([email protected], http://dermetfan.net)
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License. */
package net.dermetfan.gdx.math;
import com.badlogic.gdx.utils.FloatArray;
/** MathUtils libGDX backend
* @author dermetfan
* @since 0.5.0
* @see net.dermetfan.utils.math.MathUtils */
public class MathUtils extends net.dermetfan.utils.math.MathUtils {
/** @see net.dermetfan.utils.math.MathUtils#clamp(float[], int, int, float, float) */
public static FloatArray clamp(FloatArray values, float min, float max) {
clamp(values.items, 0, values.size, min, max);
return values;
}
/** @see net.dermetfan.utils.math.MathUtils#abs(float[], int, int) */
public static FloatArray abs(FloatArray values) {
abs(values.items, 0, values.size);
return values;
}
/** @see net.dermetfan.utils.math.MathUtils#add(float[], int, int, float) */
public static FloatArray add(FloatArray values, float value) {
add(values.items, 0, values.size, value);
return values;
}
/** @see net.dermetfan.utils.math.MathUtils#sub(float[], int, int, float) */
public static FloatArray sub(FloatArray values, float value) {
sub(values.items, 0, values.size, value);
return values;
}
/** @see net.dermetfan.utils.math.MathUtils#mul(float[], int, int, float) */
public static FloatArray mul(FloatArray values, float factor) {
mul(values.items, 0, values.size, factor);
return values;
}
/** @see net.dermetfan.utils.math.MathUtils#div(float[], int, int, float) */
public static FloatArray div(FloatArray values, float divisor) {
div(values.items, 0, values.size, divisor);
return values;
}
/** @see net.dermetfan.utils.math.MathUtils#sum(float[], int, int) */
public static float sum(FloatArray values) {
return sum(values.items, 0, values.size);
}
/** @see net.dermetfan.utils.math.MathUtils#amplitude2(float[], int, int) */
public static float amplitude2(FloatArray f) {
return amplitude2(f.items, 0, f.size);
}
/** @see net.dermetfan.utils.math.MathUtils#max(float[], int, int) */
public static float max(FloatArray floats) {
return max(floats.items, 0, floats.size);
}
/** @see net.dermetfan.utils.math.MathUtils#min(float[], int, int) */
public static float min(FloatArray floats) {
return min(floats.items, 0, floats.size);
}
/** @see net.dermetfan.utils.math.MathUtils#nearest(float, float, float[], int, int) */
public static float nearest(float value, float range, FloatArray values) {
return nearest(value, range, values.items, 0, values.size);
}
/** @see net.dermetfan.utils.math.MathUtils#nearest(float, float[], int, int) */
public static float nearest(float value, FloatArray values) {
return nearest(value, values.items, 0, values.size);
}
/** @see net.dermetfan.utils.math.MathUtils#scale(float[], int, int, float, float) */
public static FloatArray scale(FloatArray values, float min, float max) {
scale(values.items, 0, values.size, min, max);
return values;
}
}