Tungsten defines certain system properties which allow for
customization of behaviors. These properties may be set from
the Java command line as follows:
java -Dtungsten.types.numerics.foo=true
Property | Type | Default Value | Explanation |
---|---|---|---|
tungsten.types.numerics.ComplexType.extended.enable |
Boolean | false | If set to true, enables extended complex numbers. |
tungsten.types.numerics.MathUtils.prefer.native |
Boolean | true | If set to true, uses Java's inbuilt methods for computing some functions (e.g., BigDecimal.pow() ), which are typically faster but less accurate. |
tungsten.types.numerics.ComplexRectImpl.fastMagnitude.enable |
Boolean | false | If true, enables faster computation of the magnitude of complex numbers in rectangular format at the expense of less accurate identification of the result. |
tungsten.types.set.impl.FibonacciNumbers.epsilonLimit |
Real | (none) | If provided, this value is parsed as a threshold limit for epsilon values used to approximate phi. |
tungsten.types.numerics.RationalType.reduceForEquals |
Boolean | false | When true, rational values are first reduced before comparison for equality. |
tungsten.types.numerics.MathUtils.Gamma.termScale |
Integer | 2048 | Determines how many terms of the Weierstrass formula for 𝚪 will be computed. This is multiplied by the precision. |
tungsten.types.numerics.MathUtils.Gamma.blockSize |
Integer | 250 | Determines how many Weierstrass terms are computed per block, thus governs work-per-thread. |
tungsten.types.numerics.MathUtils.Gamma.zeroNeighborhood |
Real | 0.01 | Determines the maximum magnitude of a value that counts as "close enough to zero" for Gamma function approximation. |
tungsten.types.util.rendering.matrix.cell.RealCellRenderer.maxFractionDigits |
Integer | 4 | Determines the maximum number of digits to render after the decimal point. |
tungsten.types.util.rendering.matrix.cell.RealCellRenderer.useEllipses |
Boolean | true | If true, appends ellipses to values to be truncated, otherwise rounds. |
tungsten.types.Matrix.useFrobenius |
Boolean | false | If true, use the Frobenius norm instead of the max norm for matrices. |
tungsten.types.numerics.MathUtils.ln.rational.threshold |
Integer | 250 | When computing ln(x) for a rational value x, determines whether to use the integer approximation for ln(). |
tungsten.types.util.rendering.matrix.cell.ComplexCellRenderer.maxFractionDigits |
Integer | 3 | Determines the maximum number of digits to render after the decimal point. |
tungsten.types.util.rendering.matrix.cell.ComplexCellRenderer.useEllipses |
Boolean | true | If true, appends ellipses to values to be truncated, otherwise rounds. |
tungsten.types.numerics.OptionalOperations.slow.matrix.scan |
Boolean | false | When determining the element type of a matrix, setting this to true forces scanning the entire matrix. |
tungsten.types.numerics.MathUtils.lnGamma.threshold |
Integer | 7 | When computing lnGamma(z), if Re(z) < threshold, the function must be computed using a log-Gamma identity for accuracy. |
tungsten.types.numerics.ContinuedFraction.repeatInBrackets |
Boolean | false | When rendering a continued fraction, a true value renders any repeating section in angle brackets instead of using an overline. |
tungsten.types.numerics.ContinuedFraction.emitNullOnZeroTerm |
Boolean | true | When iterating over a continued fraction, emit nulls if a zero term is encountered. |