Change Log

All notable changes to this project will be documented in this file. This change log follows the conventions of keepachangelog.com.

[3.0.0 (alpha)]

First release without SMILE dependency. Also a lot of refactoring [breaking] was done.

Added

• kringing interpolation is enhanced to work with various variograms
• semi-variograms for kringing
• rbf interpolation accepts polynomial term
• barycentric interpolation
• :frequencies and :bins-maps data in histogram result
• JTransform fft,dct,dst,dht transformations
• Data transformations (Power/Box-Cox/Yeo-Johnson)
• all *mean functions accept optional weights
• skewness-test, kurtosis-test, normality-test
• durbin-watson stat
• shifting and scaling matrix columns/rows
• lm and glm in fastmath.ml.regression with full analysis
• fastmath.ml.clustering - (fuzzy) kmeans++ and dbscan clustering algorithms
• dissimilarity and similarity can accept data for histograms
• bessel-j0, bessel-j1, bessel-j, bessel-i0, bessel-i1, bessel-i, bessel-y0, bessel-y1, bessel-y, bessel-k0, bessel-k1, bessel-k add to fastmath.special
• ince-C and ince-S (with radial versions) polynomials

Changed

Removed all functions based on SMILE library, most of them are replaced/rewritten.

• [breaking] fastmath.clustering namespace is moved to the separate fastmath-clustering library.
• [breaking] fastmath.gp moved to the fastmath.interpolation.gp
• [breaking] fastmath.kernel is rewritten with changed API for rbf and kernel
• [breaking] fastmath.interpolation is rewritten with changed API, part of the methods are implemented directly in Clojure, names are unified
• [breaking] fastmath.distance rewritten
• integrate-pdf rewritten to use :gauss-kronrod integration (safer and faster)
• fastmath.core functionality based on functions with inline code not macros
• [breaking] polynomial funcitons moved to fastmath.polynomials
• [breaking] special funcitons (gamma, beta, erf, kummers, bessel, Si/Ci etc) moved to fastmath.special

[2.4.0]

Added

• Power divergence tests can work on histograms / distribution
• dissimilarity and similarity distance function of PDF (stats)

Changed

• SMILE downgraded to 2.6.0 due to licence issue (3.0.2 is GPL, 2.6.0 is LGPL)
• [breaking] correlation-matrix can use only correlation methods (:pearson, :kendall and :spearman)
• some SMILE based methods are rewritten to Java or Clojure

Deprecated

• kullback-leibler-divergence and jensen-shannon-divergence, use dissimilarity function instead.

Fixed

• stats distances type inconsistency on some cases

[2.3.0]

Added

• Apache Commons Math RealMatrix and double[][] matrix interfaces
• vec->array converts any vector to double array
• Rotation matrix, quaternions, vectors consistency

Changed

• SMILE deps updated

[2.2.2-SNAPSHOT]

Added

• mat->float-array2d, mat->array, mat->float-array - row-wise matrix representations as arrays
• debug? option for :lbfgsb optimizer
• rint accepts scale argument (same as floor and ceil)
• roll-a-dice to simulate dice rolling
• double-exponent, double-significand and ulp functions
• fastmath.solver/quadratic solver of quadratic equation ax^2+bx+c=0
• fastmath.quaternion namespace
• Cholesky decomposition for fixed size matrices
• Eigenvectors and eigenvalues for matrices
• negative-zero? tests for -0.0 value
• linear-optimizer - linear optimization
• mat, rows->mat and cols->mat - additional matrix creators
• norm - various matrix norms (0, 1, Inf, max, generalized L_pq, Shatten p-norm, Frobenius)
• condition - condition number of the matrix
• smooth-maximum - boltzmann operator, lse, smu, p-norm, mellowmax
• fastmath.calculus - integration (VEGAS+ MC, h-adaptive Cubature and Gauss-Kronrod, ACM wrappers) and differentiations: derivative, gradient and hessian + Richardson extrapolation

Changed

• fastmath.complex reimplemented

Fixed

• matrix object name in printing, indentation
• speed of element matrix accessing is improved
• eastwood boxed math and reflection warnings (removed problematic macros, shadowing global vars, etc.)
• ensuring working with Clojure 1.10.3
• problem with field methods propagation in uberjar #25, thanks @pfeodrippe

[2.2.1]

Added

• :lbfgsb java implementation of L-BFGS-B optimization algorithm, use it instead of :bfgs (buggy SMILE implementation)

[2.2.0]

2.1.9-SNAPSHOT renamed to be 2.2.0 - see changes below

[2.1.9-SNAPSHOT]

Added

• fma - fused multiply and add (+ (* x y) z) (for java 9+)
• fastmath.efloat - floating point operations with error bounds
• project - vector projection
• Apache Commons Math ArrayRealVector extended with VectorProto
• fastmath.matrix - matrix 2x2, 3x3 and 4x4 basic operations (unrolled)
• fields refactored
• binomial-ci - confidence interval methods for binomial distribution
• new distributions (based on R package gamlss.dist):
  • :zaga - zero adjusted gamma distribution
  • :nbi - negative binomial type I
  • :zinbi - zero inflated negative binomial type I
  • :zanbi - zero altered negative binomial type I
  • :zip - poisson zero inflated
  • :zip2 - zero inflated poisson 2
  • :exgaus - ex-Gaussian, exponentially modified gaussian
  • :zabi - zero adjusted binomial
  • :zibi - binomial zero inflated
  • :bb - beta binomial
  • :zabb - zero adjusted beta binomial
  • :zibb - zero inflated beta binomial
• other distributions
  • :kolmogorov - Kolmogorov distribution
  • :half-normal - half-normal distribution
  • :truncated and :mixture distributions
  • :multinomial distribution
• hpdi-extent and pi-extent - new extents based on HPDI and PI from rethinking R package
• distribution-parameters accept a keyword additionally
• fastmath.solver/find-root - root solver for univariate functions
• delta-eq and delta= for checking eqality with given accuracy.
• coefficient-matrix and correlation-matrix
• grid functions accept separate coordinates
• many additional log-exp functions
• weighted mean (wmean), median (wmedian) and quantiles (wquantile,wquantiles)
• Vec2/3/4 implement IReduce and IReduceInit
• historical trigonometric functions (versine with variants, chord, exsec/excsc) with their inverses
• Clerk notebook for: fastmath.core
• Added PDF to :cramer-von-misses distribution (based on finite difference)
• integrate-pdf converted from private to public
• contingency-2x2-measures - selection of 2x2 contingency measures
• clj-kondo hooks
• fastmath.stats.bootstrap introduced as the replacement of fastmath.stats/bootstrap function (now deprecated)

Tests

• p-value - helper for calculating p-value
• contingency-table - calculate frequencies
• binomial-test
• t-test-one-sample, t-test-two-samples (more information returned)
• z-test-one-sample, z-test-two-samples
• f-test
• power-divergence-test + special cases
  • chisq-test
  • multinomial-likelihood-ratio-test
  • minimum-discrimination-information-test
  • neyman-modified-chisq-test
  • freeman-tukey-test
  • cressie-read-test
• distribution tests
  • ad-test-one-sample - Anderson Darling test
  • ks-test-one-sample, ks-test-two-samples - Kolmogorov Smirnov test
• ANOVA tests
  • one-way-anova-test
  • levene-test
  • brown-forsythe-test
  • fligner-killeen-test
• kruskal-test - Kruskal-Wallis rank sum test

Changed

• (set! *warn-on-reflection* true) removed (motivation)
• names changed: ttest- -> t-test-
• RidgedMulti.java refactored
• effect size functions revisited

Fixed

• a nasty bug with primitive macro generation, how could I missed that?
• :histogram issues with low number of samples
• fast-max and fast-min had wrong inline operation
• wrong primitive hinting and protocol extensions
• set-seed calls protocol for PRNGs
• integrate-pdf accuracy increased
• nrd fixed when bandwidth was 0
• kde estimator revisited

[2.1.8]

Added

• use-primitive-operators accepts optional set of symbols which shouldn't be imported

Fixed

• abs redefinition in vector was not properly implemented (problem was visible when vector was precompiled with javac for Clojure2d)

[2.1.7 - do not use]

Added

• ball-random - unit ball random vector
• sequence-generator :ball added - unit ball random sampling
• some more fields
• ceil and floor can snap to the nearest multiply of scale parameter (optional)

Fixed

• Vec3 rotation
• [breaking] ensure proper behaviour for Clojure interfaces for fastmath.vector custom types (see: https://github.com/nextjournal/clerk/issues/64)
• next-double and prev-double to properly cross 0.0
• Clojure 1.11.0 fixes (abs)

[2.1.6]

Added

• jinc - besselj1(x)/x
• muladd macro
• evalpoly and mevalpoly - evaluate polynomial (from Julia)
• makepoly create polynomial function
• Si and Ci - https://dlmf.nist.gov/6.2#ii
• much more types of skewness and kurtosis
• much more effect size functions: r2-determination, eta-sq, omega-sq, epsilon-sq, cohens-f2, cohens-q, cramers-v, cohens-w, tschuprows-t
• trim and winsor data based on quantiles
• distances:
  • dist-ang - angular distance
  • sim-cos - cosine similarity
  • angular - refers to angular distance
• ->seq in fastmath.random can accept sampling scheme (:uniform, :stratified and :systematic)
• new vector functions: softmax, logsoftmax, logsumexp, logmeanexp, shift (adds a value to all elements), average (mean / weighted average)

Fixed

• moment didn't work properly for certain cases
• effect-size methods fixed
• [breaking] fixes around cosine distance and similarity

Changed

• fastmath.distance/cosine - refers to cosine similarity now
• changes around data based distributions:
  • continuous-distribution is the same as kde
  • integer-discrete-distribution and real-discrete-distribution are sorted version of enumerated-int and enumerated-real

Removed

• [breaking] fastmath.vector/dist-cos - wrong implementation

[2.1.5]

Fixed

• grid: rhombus is now a real rhombus, triangle is equilateral now (cont. #8)

[2.1.4]

Fixed

• cell->mid for :triangle grid returned wrong mid point for down triangles (#8)

Changed

• [breaking] :triangle grid anchor is 3d now, last coordinate indicates triangle position (up/down)

[2.1.3]

Fixed

• better histogram split

[2.1.2]

Added

• bias corrected and accelerated percentile method for boostrap (PR7)

[2.1.1]

Added

• bias corrected percentile method for boostrap (PR6)

[2.1.0]

Added

• l-bfgs-b optimizer as :bfgs

Changed

• SMILE version bump to 2.6.0
• fastmath.gp relies now on SMILE backend (openblas/mkl)

[2.0.5]

Added

• set-seed! accept :smile RNG which calls smile.math.MathEx/setSeed function. Also, without RNG function will set seed to both: Smile and default-rng.

[2.0.4]

Added

• :kde distribution - create distribution based on kernel density estimation

[2.0.3]

Changed

• SMILE version bumped to 2.5.0
• netlib replaced by mkl (smile)

[2.0.2]

Added

• moving-average-filter - moving average smoothing
• kernel-smoothing-filter - signal smoothing filter using kernel methods

[2.0.1]

Added

• savgol-filter - Savitzky-Golay smoothing filter

[2.0.0-alpha1]

Breaking change: due to significant change of SMILE API I decided to remove three ML namespaces: classification and regression. They probably be back as a separated bindings or incorporeted into SMILE directly.

Added

• lloyd k-means variants
• spectral clustering
• gp namespace (GaussianProcesses) - moved from regression namespace

Removed

• neural-gas clustering

Changed

• [breaking] removed namespaces: classification and regression
• :outliers? key is removed from regrouped data for clusters, now :outliers key is created when outliers are present

[1.5.3]

Added

• round-even - even (or IEEE/IEC) rounding
• cut - cut range into even intervals
• slice-range - cut range into even steps
• :dense rank ties method

Changed

• R2 sequence is able to generate up to dimensions=15
• [breaking] rank and order chagned to be zero based.
• Licence changed to MIT

Fixed

• co-intervals works the same as in R now
• group-by-intervals checks all intervals instead the first one

[1.5.2]

Fixed

• moment when center is nil should use mean

[1.5.1]

Added

• moment function

Removed

• second-moment (invalid implementation)

[1.5.0]

Changed

• removed slf4j deps

[1.5.0-alpha4]

Changed

• some defaults for scan optimization

[1.5.0-alpha3]

Added

• B-Spline and polynomial interpolation
• cb as cube of the number

[1.5.0-alpha2]

Added

• mnorm - macro version of norm

Changed

• protocol's wrapping functions shouldn't have primitive math hints (to avoid too many primitive->boxed conversions)

[1.5.0-alpha1]

Cleaned documentation with more usage examples.

Added

• demean function
• seq->vec2, seq->vec3 and seq->vec4
• warp-noise
• fastmath.signal - audio signal processing with effects + oscillators

Changed

• [possibly breaking] all protocols put in separate namespace, wrapping functions introduced (with type hints where possible)
• binary measure parameters

Fixed

• acf calculation
• MAD SMILE bindings replaced (SMILE mutates an array)

Removed

• TOLERANCE constant in fastmath.vector

[1.4.0 SNAPSHOT]

Added

• optimization package with various optimization methods
• BayesianOptimization
• NegativeBinomial distribution
• bunch of SSJ based distributions
• bootstrap datasets
• t-test

Changed

• [breaking] classification and regression refactored
• [breaking] kernel-density moved to kernel namespace

Fixed

• m/seq->double-array didn't recognize array type properly

[1.3.0 SNAPSHOT]

Added

• make-vector returns vector for given number of dimensions and optional sequence.
• vectors implements now IPersistentVector to work with vector?
• core.matrix protocols for vectors
• math functions can operate on vectors now (like [[sin]] etc.)
• new functions for vectors clamp, zero-count, nonzero-count, as-vec
• classification bindings for SMILE, ~~XGBoost~~ and LIBLINEAR
• expm1 function
• monotone interpolation
• various extent stats funcions
• haversine and haversine-dist (distance)
• predicates nan?, inf?, valid-double?, invalid-double?, between?
• calculate intervals for set of values co-intervals (same as R's function). Also group-by-intervals.
• various kernel-density methods and kernel-density-ci for kernel density with confidence interval
• various effect size functions
• binary measures like fn, tp, etc... (around 30 statistics)
• multivariate normal distribution added
• gaussian processes reimplemented
• kernels consolidated in one namespace

Changed

• use fmap instead of applyf (now deprecated)
• outliers are samples which are outside inner fence instead of outer fence
• Vectors are associative now
• [breaking] kernel-density is now multimethod
• [breaking] histogram :bins contain starting point and sample counts only

Removed

• rbf-obj converter
• rbf and mercer namespaces (use fastmath.kernel instead)

[1.2.0]

Added

• Various grid operations
• hashCode for vectors

Changed

• Breaking: sequence generator creatators return lazy sequence now (instead of function returning sequence)

[1.1.4]

Added

• lcm - least common multiplier
• Vec types implement Reversible Indexed ILookup
• negative values for some constants (PI, E etc.)

Fixed

• GAMMA constant name clash

Changed

• stats-map returns list of outliers rather than number of outliers

Removed

• :bessel vector field removed (hard to limit input range)

[1.1.3]

Added

• sample can now return pairs of [x,(f x)].

[1.1.2]

Added

• estimators for number of bins in histogram
• kernel-density function

[1.1.1]

Changed

• histogram data enhanced

[1.1.0]

Added

• fastmath.clustering namespace

Changed

• removed k-means from fastmath.stats

[1.0.3]

Fixed

• norm when domain is a point, returns range start when value is less or equal domain, range end otherwise
• small fix in one field

[1.0.2]

Added

• curl for 2d vector fields

[1.0.1]

Added

• :default (:jdk) rng is added

Fixed

• MersenneTwister is not synchronized, default RNG is :jdk now

[1.0.0]

Added

• easings namespace
• sample function
• extent stat ([min,max] pair)
• histgram
• vector field functions moved from clojure2d, api changed

Changed

• interpolator have shorter names now (without -interpolator suffix)
• rbf functions are moved to separate namespace
• changed parameters order in interpolators (xs and ys are last)
• next-float-up -> next-double
• next-float-down -> prev-double

[0.1.1]

Added

• BesselJ
• all Gamma function variants (regularized, etc)
• regularized Beta
• log1p
• estimation-strategy can be passed to any function calculating quantiles/percentiles
• ->seq can accept number of samples now
• low-exp, high-exp to find lower/greater exponent for given base and number

Changed

• stats-map doesn't contain select keys option anymore

[0.1.0]

Initial version created from Clojure2d library.