fix(deps): update dependencies - #4

v0.2.0

Compare Source

Release backfilled on 2026-10-26; tag created 2026-10-17.

Commits since v0.1.0:

• 057b0e6 Renovate: extend the organisation preset (#​5)
• ed92ca8 README: brand banner, docs badge, links to the site and the docs (#​6)
• 4c68ceb fix(deps): update dependencies (#​2)
• 2bf87b1 perf: faster I/O and operations (all but Parquet write and filter now at or above pandas/pyarrow) (#​7)

v0.8.5

Compare Source

What's Changed

• chore(deps): update dependency node to v26 by @​tannevaled in #​839
• internal/vm: cover the three Inspect/ToS bindings #​704 said were invisible by @​tannevaled in #​815
• conformance: census the version guards by shape, before anything is re-recorded by @​tannevaled in #​836
• internal/vm: Kernel#to_s prints the address for a plain object, and Thread::* carry their qualified names by @​tannevaled in #​818
• fix(deps): update dependencies by @​tannevaled in #​792
• vm: the Ruby 4.0 namespace, as one object under two names by @​tannevaled in #​844
• conformance: derive CUR_VERSION from RUBY_VERSION — one fact, one place by @​tannevaled in #​845
• vm: Array#rfind, and it has to index the live array by @​tannevaled in #​849
• vm: Math.log1p and Math.expm1, the two functions Ruby 4.0 added to Math by @​tannevaled in #​846
• vm: p on a self-referential object or Struct killed the process by @​tannevaled in #​852
• conformance: cmd/phantom — is a pass the behaviour, or the method's absence? by @​tannevaled in #​851
• Port to plan9 (386, amd64, arm) by @​tannevaled in #​855

Full Changelog: go-embedded-ruby/ruby@v0.8.4...v0.8.5

v0.32.0

Compare Source

What's Changed

• fix: float32 Exp and Log correctly rounded, proven over every float32 by @​tannevaled in #​87

Full Changelog: go-ndarray/ndarray@v0.31.0...v0.32.0

v0.31.0

Compare Source

What's Changed

• docs: N1 diagnosis — fresh quiet-host ratios, counters, causes ruled out by @​tannevaled in #​85
• perf: float32 Exp and Log on the float64 kernels (2-3.5x faster) by @​tannevaled in #​86

Full Changelog: go-ndarray/ndarray@v0.30.0...v0.31.0

v0.30.0

Compare Source

What's Changed

• docs: products slow down on a busy N1 host, OpenBLAS does not (open defect) by @​tannevaled in #​83
• chore(deps): update dependency go to v1.27.3 by @​tannevaled in #​60
• feat(random): allocation-free scalar draws (Float64, FillUniform, IntN, FillIntN, ShuffleInt64) by @​tannevaled in #​84

Full Changelog: go-ndarray/ndarray@v0.29.1...v0.30.0

v0.29.1

Compare Source

What's Changed

• perf(gemm): read a row-major B in place on the small-product path by @​tannevaled in #​82

Full Changelog: go-ndarray/ndarray@v0.29.0...v0.29.1

v0.29.0

Compare Source

What's Changed

• fix(deps): update dependencies by @​tannevaled in #​73
• perf(gemm): native float32 GEMM on s390x by @​tannevaled in #​79

Full Changelog: go-ndarray/ndarray@v0.28.0...v0.29.0

v0.28.0

Compare Source

What's Changed

• perf(gemm): native float32 GEMM on ppc64le by @​tannevaled in #​78

Full Changelog: go-ndarray/ndarray@v0.27.0...v0.28.0

v0.27.0

Compare Source

What's Changed

• perf(gemm): read a row-major A in place (BLIS sup kernels) by @​tannevaled in #​77

Full Changelog: go-ndarray/ndarray@v0.26.1...v0.27.0

v0.26.1

Compare Source

What's Changed

• perf(gemm): pack full B panels with whole-row array moves by @​tannevaled in #​75

Full Changelog: go-ndarray/ndarray@v0.26.0...v0.26.1

v0.26.0

Compare Source

What's Changed

• docs: the allocating operations' cost, measured and split by @​tannevaled in #​72
• perf(linalg): native float32 GEMM on amd64 and arm64 by @​tannevaled in #​74

Full Changelog: go-ndarray/ndarray@v0.25.0...v0.26.0

v0.25.0

Compare Source

What's Changed

• perf: loong64 Max/Min (float64, float32) in LSX integer arithmetic by @​tannevaled in #​71

Full Changelog: go-ndarray/ndarray@v0.24.0...v0.25.0

v0.24.0

Compare Source

What's Changed

• perf: ppc64le Max/Min (float64, float32) with an integer NaN mask by @​tannevaled in #​70

Full Changelog: go-ndarray/ndarray@v0.23.0...v0.24.0

v0.23.0

Compare Source

What's Changed

• ci: run the loong64 qemu lane with LASX, and require it by @​tannevaled in #​67
• ci: run the s390x qemu lane with VXE, and require it by @​tannevaled in #​68
• perf: s390x float32 Max/Min (VFMAXSB, z14+) by @​tannevaled in #​69

Full Changelog: go-ndarray/ndarray@v0.22.0...v0.23.0

v0.22.0

Compare Source

Release backfilled on 2026-10-26; tag created 2026-10-26.

Commits since v0.21.0:

• 151c4c1 perf: RVV Max/Min (float64, float32); fix a small-n RVV sum; run RVV in CI (#​66)

v0.21.0

Compare Source

Release backfilled on 2026-10-26; tag created 2026-10-26.

Commits since v0.20.0:

• 543d1d0 perf: float32 MatMul on the float64 GEMM (#​65)

v0.20.0

Compare Source

Release backfilled on 2026-10-26; tag created 2026-10-26.

Commits since v0.19.1:

• f4ab9ab perf: float32 Max/Min read float32 directly; SSE2 and NEON kernels (#​64)

v0.19.1

Compare Source

Release backfilled on 2026-10-26; tag created 2026-10-26.

Commits since v0.19.0:

• 5759867 ci: install qemu-user, not the virtual qemu-user-static (Ubuntu 26.04) (#​63)
• 725ee5d fix: amd64 Max/Min over signed zeros depended on element order (#​62)

v0.19.0

Compare Source

Release backfilled on 2026-10-26; tag created 2026-10-25.

Commits since v0.18.0:

• 2432766 fix(deps): update dependencies (#​57)
• 8388dc6 perf: float32 Sqrt kernels on six targets; float32 survey against NumPy (#​61)

v0.18.0

Compare Source

Release backfilled on 2026-10-26; tag created 2026-10-19.

Commits since v0.17.0:

• 8979001 perf: float32 Add/Sub/Mul/Div kernels on six targets; *Into writes typed results in place (#​59)

v0.17.0

Compare Source

Release backfilled on 2026-10-26; tag created 2026-10-18.

Commits since v0.16.0:

• 0906fd8 perf: float32 pairwise sum blocks on s390x (VFASB, z14+) (#​58)

v0.16.0

Compare Source

Release backfilled on 2026-10-26; tag created 2026-10-18.

Commits since v0.15.0:

• cfbca31 perf: float32 pairwise sum blocks on VSX (ppc64le) (#​56)

v0.15.0

Compare Source

Release backfilled on 2026-10-26; tag created 2026-10-18.

Commits since v0.14.0:

• 365f1d1 perf: float32 pairwise sum blocks on RVV (riscv64) and LSX (loong64) (#​55)

v0.14.0

Compare Source

Release backfilled on 2026-10-26; tag created 2026-10-18.

Commits since v0.13.0:

• 8bd10b2 perf: narrow integer sums widen as they read, without an int64 copy (#​52)
• 1af734c ci: an A/B benchmark workflow on the hosted amd64 and arm64 runners (#​53)
• c44fc7b perf: float32 sum blocks on SSE2 and NEON; integer widening folds without bounds checks (#​54)

v0.13.0

Compare Source

Release backfilled on 2026-10-26; tag created 2026-10-18.

Commits since v0.12.0:

• 7a0a00c fix(deps): update module github.com/go-fft/fft to v0.27.1 (#​48)
• a1f82eb test: FuzzMixedDTypes, the values of mixed-dtype operations (#​50)
• 0659ebc perf: fold contiguous reductions directly; NumPy's pairwise sum for float32/complex (#​51)

v0.5.0

Compare Source

v0.5.0 - quadratic programming (Quadprog)

Quadprog solves min ½xᵀPx + qᵀx subject to equalities, inequalities, two-sided rows and bounds, with dense or sparse (go-sparse) P and A.

• Goldfarb-Idnani dual active set (QPActiveSet) for a positive definite P: infeasibility certificates, dependent and redundant rows, working-set warm starts.
• ADMM of OSQP v1.0.0 (QPADMM) for a singular P, an LP or sparse constraints: Ruiz scaling, adaptive rho, polishing, infeasibility and unboundedness certificates, non-convexity detection. Status, iteration and rho-update counts equal OSQP's on 38 of 38 planted cases, 12 of 12 sparse cases and 25 of 25 generated benchmark instances.
• QPSolver keeps the factors between solves: UpdateQ, UpdateBounds, UpdateMatrixValues, WarmStart / WorkingSet.
• QPAuto picks the active set up to n + m = 2500 (measured crossover with ADMM near 2600).
• Checked against OSQP, Clarabel and the quadprog package, and on 33 Maros-Meszaros problems (ADMM reaches Clarabel's objective on 27 and ends with OSQP's own status on the other 6).
• Performance (tools/qpbench, docs/perf.md): ADMM takes 0.5-0.65x OSQP's time on dense n >= 500 and 0.8-0.9x on larger portfolios, but 1.04-1.49x in the solve phase and 2.2-4.9x in the setup on small sparse problems, where go-sparse (AMD, symbolic analysis, LDL solves) is the cause - an open defect there. The active set is 0.68-0.76x the quadprog package from n = 500, 1.09x at n = 200, 2.0x at n = 50.
• Tools: qporacle (Maros-Meszaros fetch/convert, sparse fixture form, linux/amd64), qpbench.
• go-sparse v0.2.1.

v0.4.1

Compare Source

Fixes

• trust-constr: projected_cg's rt_g and the tangential trust radius are now norm(x)**2 (the square of the rounded norm) as in SciPy v1.18.1, not the dot product x.x; its boundary step is x + (theta*alpha)*p, not x + theta*(alpha*p). HS21/jac and HS35/jac now take SciPy's exact path (exact runs of the 100-run fixture: 59 to 65 on arm64, 68 to 69 on amd64).

Tools and docs

• tools/slsqporacle: new trace and tracediff subcommands locate where a run leaves SciPy's path, iterate by iterate; TestTraceDump is the Go side.
• docs/design.md: the four remaining runs that differ from SciPy (HS48/fd, Rosen-free/fd, SingularEq3/jac and /fd) are explained with their first divergence (last-bit rounding of fused against unfused multiply-adds, amplified by finite differences; the SVD of a rank-deficient Jacobian), and their iteration count is now asserted.

v0.4.0

Compare Source

What's Changed

• feat: trust-constr interior-point method (Minimize method trust-constr) by @​tannevaled in #​13

Full Changelog: go-numerics/optimize@v0.3.1...v0.4.0

v0.3.1

Compare Source

What's Changed

• test(tools): tests for slsqporacle and slsqpbench, run in CI by @​tannevaled in #​12

Full Changelog: go-numerics/optimize@v0.3.0...v0.3.1

v0.3.0

Compare Source

What's Changed

• fix(deps): update module github.com/go-ndarray/ndarray to v0.18.0 by @​tannevaled in #​8
• ci: install qemu-user, not the virtual qemu-user-static (Ubuntu 26.04) by @​tannevaled in #​9
• fix(deps): update module github.com/go-ndarray/ndarray to v0.28.0 by @​tannevaled in #​10
• feat: SLSQP constrained minimisation (Minimize method SLSQP) by @​tannevaled in #​11

Full Changelog: go-numerics/optimize@v0.2.0...v0.3.0

v0.2.0

Compare Source

Release backfilled on 2026-10-26; tag created 2026-10-18.

Commits since v0.1.0:

• 6b29dcf README: brand banner, docs badge, links to the site and the docs (#​4)
• dc2a15a Renovate: extend the organisation preset (#​3)
• 08bb0fe fix(deps): update module github.com/go-ndarray/ndarray to v0.12.0 (#​5)
• 7caa2de perf: incomplete gamma/beta, Bessel and elliptic: faster, more accurate, no budget raised (#​6)
• 1839669 perf: elementary families on float64 kernels (Gamma, Erf, inverses, logistic, zeta) at unchanged accuracy (#​7)
• 66d0333 README: name the scipy function each one is measured against, and gate the table against the harness (#​8)

v0.3.1

Compare Source

What's Changed

• fix(deps): update dependencies by @​tannevaled in #​12
• chore(deps): update dependency go to v1.27.3 by @​tannevaled in #​11
• Bump go-gfx/gfx to v0.37.0 by @​tannevaled in #​16

Full Changelog: go-plot/plot@v0.3.0...v0.3.1

v0.3.0

Compare Source

go-plot/plot v0.3.0

New: text from data, without TeX

This follows matplotlib's parse_math.

• plot.ParseMath(false) (FigOption) and Figure.SetParseMath(bool) set the figure's default, as rcParams["text.parse_math"] does.
• plot.PlainText() and plot.MathText() on one call act as parse_math=False and parse_math=True on that text.
• The calls that take them:
  • SetTitle(s, opts...), SetXLabel(s, opts...), SetYLabel(s, opts...);
  • the new SetXTickLabels(locs, labels, opts...) and SetYTickLabels(locs, labels, opts...), for category ticks;
  • Legend (every entry), Text, Annotate and Pie.
• A plain string is drawn character for character as outlines. It never reaches the TeX typesetter.
• The formatters' own tick labels stay math.

SetTitle, SetXLabel and SetYLabel gained variadic options. Calls compile unchanged, but method values change type.

Tests use $\input{/etc/passwd}$, <script> and 100$. A spy renderer checks that each string goes through plain layout. A control checks that the default still sends them to TeX. The SVG stays CSP-safe.

v0.2.0

Compare Source

go-plot/plot v0.2.0

New

• Axes.Pie and Axes.Donut: matplotlib's ax.pie.
  • Options: Labels, AutoPct/AutoPctFunc, PctDistance, LabelDistance, StartAngle, CounterClock, Explode, Radius, Center, WedgeWidth, RotateLabels, Normalize, TextColor, and the existing Colors, EdgeColor, Alpha, FontSize.
  • Each labelled wedge gets a legend entry.
  • Negative, NaN, infinite or all-zero sizes return ErrPieNegative, ErrPieNonFinite or ErrPieZero, in matplotlib's order.
  • Two go-plot additions: PlainText() keeps labels out of TeX, so $ is a dollar; SpreadLabels() moves the labels of thin wedges apart and draws leader lines.
• text.PlainRenderer (LayoutPlain).
• colors.Tableau10, the palette of go-widgets/toolkit's PieChart.

Changed: SVG output is safe to inline under a strict CSP

• Paint is written as presentation attributes (fill="…"), no longer as style="…".
• Clip ids are derived from their rectangle (c<hash>), not counted (c1, c2…). Two figures inlined in one page no longer clip each other.
• There is no script, no event attribute and no external reference. Text is glyph outlines.

Fixed

• cost \$5 drew its backslash.

Judges

Against matplotlib 3.11.2 (tools/gen_pie.py):

• wedge paths match vertex for vertex on 10 cases;
• every label's text, position and alignment match;
• the Axes box and the refusals match;
• 99.1–99.9% of the pixels agree with matplotlib's PNG, and a negative control is included.

Also:

• golden SVGs;
• FuzzPie;
• 100% coverage on 3 OSes and 6 architectures.

Not measured against matplotlib's speed: the load was above the gate. tools/compare/perf.py has the pie cases.