pypi dit 2.4
v2.4

one hour ago

What's Changed

  • Markov-order tests, bias-corrected and ordinal estimators, and null generators for dit.inference by @Autoplectic in #256
  • Move transfer entropy and significance tools to infoflow; add ordinal patterns and permutation entropies by @Autoplectic in #257
  • Fix KSG bias: count neighbours strictly inside the epsilon-ball by @Autoplectic in #258
  • Add less-noisy intrinsic mutual information and bound_information example by @Autoplectic in #262
  • Sound relaxed two-part bound for two_way_skar; fix SKAR bound docs by @Autoplectic in #261
  • Flatten reduced intrinsic MI into a single joint optimization by @Autoplectic in #259
  • Add iterated public discarding lower bound on two-way SKAR by @Autoplectic in #260
  • Bump urllib3 from 2.7.0 to 2.8.0 by @dependabot[bot] in #263
  • Add deterministic_erasure and xor_key example sources (Abin & Gohari 2026) by @Autoplectic in #266
  • Fix speaker alphabet in iterated_discarding_skar for |X| != |Y| by @Autoplectic in #267
  • Add sources with proven secret key agreement rates and tests by @Autoplectic in #268
  • CI: push builds only on main; cancel superseded runs by @Autoplectic in #271
  • Seed intrinsic MI basin hopping with constant and identity corruptions by @Autoplectic in #270
  • Add secret key cost (information of formation) by @Autoplectic in #269
  • Fix one-way SKAR and aux-var Jacobian; add PID_LN; structural seeding for nonconvex optimizers by @Autoplectic in #272
  • Add multiterminal secret key capacity with helpers and omniscience rate by @Autoplectic in #273
  • Correctness fixes from full package review by @Autoplectic in #274
  • Add order-k Kirkwood and Lancaster approximation constructors by @Autoplectic in #276
  • Exact functional common information via minimum-entropy exact cover by @Autoplectic in #277
  • Add the standard genetic code as an example distribution by @Autoplectic in #278
  • Add the privacy funnel and an IB/PF region plotter by @Autoplectic in #279
  • Add the Fejér-Riesz approximation constructor by @Autoplectic in #281
  • Add common-information-ordering witness example distributions by @Autoplectic in #280
  • Tension common information: dual-total-correlation residual, meet split, and MSS reduction by @Autoplectic in #282
  • Add PID_CI union-information synergy measure and BaseSynergyPID by @Autoplectic in #283
  • Speed up CI test runs by @Autoplectic in #284
  • Add frustration measure (order-k, with pairwise_frustration) by @Autoplectic in #285
  • Add hypothesis tests for the TV and KL data-processing inequalities by @Autoplectic in #286
  • Add ExtensionProfile: extension profile, its distinguished points, and an interactive plot by @Autoplectic in #287
  • Bump nbconvert from 7.17.1 to 7.17.2 by @dependabot[bot] in #288
  • Bump pillow from 12.3.0 to 13.0.0 by @dependabot[bot] in #289
  • Add channel_synthesis_rate (Cuff 2013) and closed-form common-information examples by @Autoplectic in #290
  • Bump filelock from 3.24.2 to 4.2.1 by @dependabot[bot] in #291
  • Bump certifi from 2026.1.4 to 2026.10.19 by @dependabot[bot] in #292
  • Bump setuptools from 83.0.0 to 84.1.0 by @dependabot[bot] in #293
  • Bump tornado from 6.5.7 to 6.5.9 by @dependabot[bot] in #264
  • Bump fsspec from 2026.2.0 to 2026.6.0 by @dependabot[bot] in #265
  • Certified, accelerated Blahut-Arimoto; add capacity-cost and Sibson capacity by @Autoplectic in #294
  • Add bias-corrected discrete entropy estimators and alphabet-aware CMI by @Autoplectic in #295
  • Add PED_SX, PhiID, GID, alpha-synergy spectra and representational complexity by @Autoplectic in #296
  • Reorganize docs, add generated API reference, and refresh README by @Autoplectic in #298
  • Fix alpha-synergy spectra for near-underflow probabilities by @Autoplectic in #300
  • Add non-signalling box variants, CHSH value, and non-signalling check by @Autoplectic in #299
  • Channel protocol: attach channels with memory to channel codes by @Autoplectic in #301
  • Feedback channel codes: erasure ARQ, stop-and-wait ARQ, Horstein by @Autoplectic in #304
  • Add undersampled, continuous and time-series entropy estimators by @Autoplectic in #305
  • Standardize keyword names, validate input pmfs, keep coalesced symbols as tuples by @Autoplectic in #306
  • Add exact channel synthesis, the exact common information rate, and the maximal cross entropy by @Autoplectic in #307
  • Bump tornado from 6.5.9 to 6.6 by @dependabot[bot] in #309
  • Support alpha = inf (the max-divergence) in renyi_divergence by @Autoplectic in #308
  • Add the higher-order common information and a rate-constrained information bottleneck by @Autoplectic in #310
  • Exact degradation redundancy (PID_Deg, PID_Prec) and corrected PID_RDR by @Autoplectic in #312
  • Add a Thanks section to the README by @Autoplectic in #313
  • Add PID property checks and literature claims by @Autoplectic in #311
  • PID audit: validate inputs, fill incomplete lattices exactly, fix measures on degenerate inputs by @Autoplectic in #314
  • Report the BROJA cone program's duality gap in consistent units (fixes #302) by @Autoplectic in #317
  • CI: test BROJA's cone solver, and measure coverage on the newest Python of every OS by @Autoplectic in #318
  • PID performance and conditional/contextual decompositions (#315 and #316, onto main) by @Autoplectic in #319
  • Add tests for 16 under-covered modules, run the symbolic tests on CI, and make MinEnt optima match the marginals by @Autoplectic in #320
  • CI: drop the unused cddlib installs by @Autoplectic in #321
  • Return non-finite vectors from MinEnt's projection instead of looping by @Autoplectic in #322
  • Add the commitment capacity of a channel by @Autoplectic in #323
  • Test BROJA on W, every bivariate gate, and random sparse supports by @Autoplectic in #325
  • Closed-form dependency PIDs, PID overlap diagnostics, and reference-family tests by @Autoplectic in #326
  • Add PID_HiFi, per-source Hi-Fi bounds, a triplet report, and a BROJA guard by @Autoplectic in #327
  • Add PIDBounds, forced overlap, and layer-cake decompositions by @Autoplectic in #328
  • Add a G² test for the highest-order connected information by @Autoplectic in #329
  • Stop prepare_dist from modifying its input, and accept a string for rv_names by @Autoplectic in #330
  • Record PID_dep_c's two-source properties, and correct the Hi-Fi references by @Autoplectic in #331
  • Add pointwise (local) forms of the Shannon-like measures by @Autoplectic in #332
  • Rename PID_dep_c to PID_PR, the pairwise-reference PID by @Autoplectic in #333
  • Credit the DeWeese measures to a personal communication by @Autoplectic in #334
  • Make PLTable optional, falling back to prettytable by @Autoplectic in #336
  • Bump soupsieve from 2.9 to 4.0 by @dependabot[bot] in #337
  • Audit the claimed properties of every PID measure, fix PID_CI's source pruning, and generalize PID_CT by @Autoplectic in #335
  • Keep small probabilities when reading BROJA solver results by @Autoplectic in #338
  • Speed up unconditioned entropies and the PID totals by @andrefcorreiagomes in #339
  • Add regions, discernible subsets, Gács–Körner and formal sums to the logarithmic decomposition by @Autoplectic in #340
  • Add causal emergence criteria and integrated-information readouts to PhiID by @Autoplectic in #341
  • Add a Gács–Körner double redundancy to PhiID by @Autoplectic in #342
  • Add the information-geometric PhiID and compute I_IG exactly on sparse supports by @Autoplectic in #343
  • Bump tornado from 6.6 to 6.6.1 by @dependabot[bot] in #344
  • Add the W- and M-information and BROJA-PhiID of Liardi et al. by @Autoplectic in #345

New Contributors

Full Changelog: v2.3...v2.4

Don't miss a new dit release

NewReleases is sending notifications on new releases.