Repository navigation
Releases
Casimir 1.7.0
Casimir 1.7.0: solve refuses an equation whose coefficient of the variable is 0 where canon can't see it, in place of 1.6.0's SOLVED with a row that never holds. (SIN(A)^2+COS(A)^2-1)×X=1 is NOT_NUMERIC, as A×X-A×X=1 already was.
What's new since 1.6.0
- A coefficient that is 0 where canon can't see it (CAS 040). Before a letter solve answers, X's coefficient is read at six fixed points, each letter its own value (A takes 1/5, -2/3, -9/4, 19/3, 147/4 and -75/2), by the bound every printed sign rests on (CAS 022 rule 8). Shown not 0 at a point, the equation answers as at 1.6.0. Read 0 at each point where it has a value, and not 0 at none, it is refused NOT_NUMERIC, never taken as 0 from its digits:
(EXP(A+B)-EXP(A)×EXP(B))×X=C,(TAN(A)×COS(A)-SIN(A))×X=1in GRAD and(LN(A^2)-2×LN(A))×X=1are each NOT_NUMERIC, and show steps refuses them the same way. A coefficient that isn't 0 keeps its answer:(SIN(A)^2+COS(A)^2)×X=1isSOLVED: X=1÷(SIN(A)^2+COS(A)^2); SIN(A)^2+COS(A)^2≠0, and(ATAN(1÷A)+ATAN(A)-π÷2)×X=1, 0 only for A>0, keeps its line. work/cas040.md has the points, the readings and the shapes refused though not 0. - What moves, against 1.6.0:
- the golden corpus: none of its 10,822 lines;
- the documented examples: none of 1.6.0's 1,323 commands;
- the developer tests: none of tests/solve_test.c's 416 earlier checks;
- the reviewer's hidden sets of the earlier units: 27,858 records byte for byte, but the Pythagorean identity row of CAS 038's and CAS 039's sets, now NOT_NUMERIC.
- Measured: S3 code 270,410 bytes (605 more than 1.6.0's), the first release past decision 92's 270,336 B line, inside decision 94's 301,056 B (Michael's pre-approved 30 KiB raise). No accepted function's frame moved, on the host or the S3. On the ESP32-S3 the bench passes at this tag, and its worst stack is unchanged (11,360 B masked).
The guarantee is unchanged: an exact answer, or a status that says why not. Never an approximation presented as an answer.
The known wrong answers
None. CAS 040's hidden set found no wrong value, on its first run.
How it's tested (at this tag)
make test: 20 programs with 118,633 checks. They read every answer back through an independent evaluator (the calculator firmware's core) and compare values. New in 1.7.0: the identities behind the refused rows are checked to be 0 on the core at letter values of each sign, and the rows the zero test must leave alone print their 1.6.0 lines.- The golden corpus, 10,822 calls compared byte for byte; none moved.
- Frame and symbol gates: no frame over 1 KiB, no
powlinked, and every function's frame against the recorded table. No frame moved from 1.6.0, on the host or the S3 (every row diffed). The new function takes 96 B on the S3 (ls_zero). - Value checks beside the corpus judge Taylor polynomials, limits, integrals, sums and solve's steps against independent constructions.
- The docs: every
casimrow of docs/language.md and the units' example files is run (2,860 checks, each again with the series route compiled out), and every backticked answer in docs/language.md's prose must be in one of them or in the golden corpus. - Two fuzzes, against 1.6.0's command line (scripts/zero-diff.py, scripts/zero-plant.py): 20,000 random letter equations, none printing differently; 4,000 with an identity planted in the coefficient, all 2,057 zero ones refused and all 1,943 others as at 1.6.0.
- CI on every push:
make checkunder gcc and clang,make test, CMake under gcc and clang, WebAssembly, and the ESP-IDF component for the ESP32-S3. - Mutants (scripts/mutants.py): 588 deliberate faults, and at this tag's tree
make testcatches 582 (work/v1.7.0-mutants.txt). The six that survive are 1.6.0's six, and searched again at this tree each moves exactly the lines it moved at 1.6.0's (work/v1.5.0-mutants.txt names what masks each). CAS 040's 12 and the three CAS 038 mutants it re-anchored are caught. - Hidden sets kept by the project's reviewer, independent of this repository, rerun at each unit's acceptance. At CAS 040's: its own, 223 rows at two arenas, 0 of 446 runs failing, and the earlier units'.
- On the ESP32-S3, at this tag's tree (8a76a80), every one of the bench's 1,454 cases is byte-identical to the host, and the steps of the 297 derivatives, integrals, solves and partial fractions page as the host's:
- with a 16,384 B arena, the size the calculator runs it with;
- worst stack 11,360 B read with interrupts masked, plus a 400 B interrupt allowance (two frames, or the most measured over the masked reading where that is larger), 11,760 of a 12 KiB task;
- nesting deeper than 44 levels is refused (NO_ROOM);
- about 270 KB of Casimir's code at -O2 (270,410 B), 30,646 B under decision 94's 301,056 B line.
Not yet, after 1.7.0
These print a line without the note it could have, print a row the answer doesn't need, have no value where the sum has one, refuse, or print a constant as written. None of them is a wrong value.
New with the zero test (CAS 040), each a refusal or 1.6.0's line, never a wrong value:
- Refused, though X's coefficient isn't 0. The coefficient reads 0 at every one of the six points where it has a value: 0 on a region holding them all (
(ABS(A+40)-(A+40))×X=1, 0 for A≥-40, whose 1.6.0 answer held for A<-40); a factor built to vanish at the points; a part under about 1E-31 of its terms ((SIN(A)^2+COS(A)^2-1+10^-40)×X=1); or read at one point only, 0 there ((4×A-147)×SQRT(A-30)×X=1, which has a value only at A = 147/4 among the points). - A 0 coefficient no point reads prints 1.6.0's line, SOLVED with a row that never holds: functions the bound doesn't read (
(IP(A)+FP(A)-A)×X=1,(10^LOG(A)-A)×X=1), and a domain holding no point ((2×LN(A-60)-LN((A-60)^2))×X=1). - canon doesn't multiply out (A-B)×(A+B) where a sum holds it:
(A^2-(A-B)×(A+B)-B^2)×X=1printed SOLVED at 1.6.0, and 1.7.0 refuses it only because the point test reads its 0. canon's form is unchanged.
Carried from 1.6.0, each example run again on this tag's command line, where it prints as at 1.6.0. With letters as coefficients (CAS 038 and 039):
- canon's forms, not a student's:
(X+A)÷B=CisX=B×(C-A÷B); B≠0,A×X+B=CisX=-(B-C)÷A; A≠0,F=9÷5×C+32for C isC=5×F÷9-160÷9, and steps' BOTH SIDES forV=I×Ris-I×R+V=0. - A row in the expanded coefficient's form:
A=(B+C)×H÷2for H has-B÷2-C÷2≠0, and(A×B-A)×X=Bhas-A+A×B≠0. - A row that always holds, or one left unsolved:
(SQRT(A)+1)×(X+1)=BhasSQRT(A)+1≠0,(SQRT(A)-1)×X=1hasSQRT(A)-1≠0(A≠1),ACOSH(A)×X=1hasA≥1; ACOSH(A)≠0(A>1), and(A^3-1)×X=1hasA^3-1≠0. - Refused, NOT_NUMERIC as 1.5.0 refuses them: var dropping out (
A×(X-1)=A×X), degree 2 or more with a letter (X^2=A), var inside a function or a denominator (A×SIN(X)=B,A÷X=1), a condition no row writes (ASIN's ABS(u)≤1, ATANH's ABS(u)<1), model 33's variable i, a constant with neither a letter nor π (SIN(1)×X=1), and an answer where the typed text has no value (X×(X-A)÷(X-A)=A). pw's first refusal decides whether letters are read at all:X^2÷X×A=1is NOT_POLYNOMIAL, andA×X^2÷X=1isSOLVED: X=1÷A; A≠0. Past six conditions, NO_ROOM (A×B×C×D×E×F×G×X=1).
Carried from 1.5.0, printing as at 1.6.0:
- The typed domain, stricter than it needs to be (CAS 035):
- limit's letter conditions at the edge:
SQRT(Y)^2×Xat 1 printsY; Y>0, though Y = 0 has a value, andASIN(Y)×Xat 1 printsASIN(Y); ABS(Y)<1, though |Y| = 1 has one; SQRT(X^2×Y)^2at 0 gives0; Y>0, though Y = 0 has a value;- limit's degenerate divisor tries k = 0 only for k linear in one letter:
(SIN((Y-Z)×X)+1)÷(SIN((Y-Z)×X)+1)at 0 prints1; Y-Z≠0, though at Y = Z it is 1 everywhere, andY×Z-1andY^2-1(1; Y≠-1; Y≠1) the same.
- limit's letter conditions at the edge:
- The typed domain, refused where it could be answered (CAS 035):
- an X×Y-form limit at ±∞ is UNDECIDED: its sign is Y's, and no row is given (
SQRT(X×Y)÷SQRT(X×Y)at ∞); - a divisor 0 for every X at a letter's value that casim can't rule out is UNDECIDED:
(Y×SIN(X)-SIN(Y×X))÷(Y×SIN(X)-SIN(Y×X))at 0, whose divisor is 0 for every X at Y = ±1; - rows count toward the 191 bytes, so some lines with them are TOO_LONG:
K×(1+Z×G)^Kfrom 1 to ∞, where G is u written asSQRT(u)^2,(u^(1÷4))^4,EXP(LN(u))orALOG(LOG(u))and u is Y+1, Y-2 or W-1, and(K-1)×(F(Z)×Y)^K÷F(Z)from 1 to N for ACOSH, ASINH and ATANH, and LN from 2.
- an X×Y-form limit at ±∞ is UNDECIDED: its sign is Y's, and no row is given (
- Older checks beside the typed rows (found by 1.4.0's mutant search; 1.3.0 prints the same):
- which refusal a constant with no value gets beside a pole on the interval:
X^(-2)×ACOS(3)over 0..1 is IMPROPER, where UNDEFINED would be just as lawful (CAS 009's interval check answers first); - limit rows from the answer's own walk and f's (CAS 023 sites 6 and 7) that restate a typed row (
K×π+4≠0besideK×π÷2+2≠0) or can't fail (EXP(K^2)≠0), and 69 limit lines found TOO_LONG where the line without site 6's rows fits.
- which refusal a constant with no value gets beside a pole on the interval:
- A quotient that canon cancels, or whose divisor falls in K, is still refused.
X^(2×K)÷X^Kis NO_CLOSED_FORM: canon reads it asX^K, which has a value at X = 0, where its terms are 0÷0. So isX^K÷Y^(-K): its divisor has no value at Y = 0, while canon's(X×Y)^Khas one. Answering them needs a row that(X÷Y)^K-style twins don't print (X≠0,Y≠0). So areY^K÷Y^K(whereY^K×Y^(-K)printsN; Y≠0) and a divisor in K inside an exponent:(-8)^(1÷(K-1))from 0 is NO_CLOSED_FORM, though the term at K = 1 has no value (UNDEFINED would say why). - A divisor of degree 2 or more in K that canon cancels is read by no CAS 037 rule. The shapes found are refused (
((K^2-1)×K)÷(K^2-1)from 0 to N is NO_CLOSED_FORM) or UNDEFINED already. - UNDEFINED for a whole sum to N whose divisor is 0 mid-range (CAS 037's (C), as for a divisor canon keeps):
((K+1)^2-1)÷((K+1)-1)from -2 to N is UNDEFINED, though the sum has a valu...
Casimir 1.6.0
Casimir 1.6.0: solve answers an equation linear in its variable whose coefficients hold other letters, with the rows where the answer holds, and shows its steps. V=I×R for R is SOLVED: R=V÷I; I≠0, where 1.5.0 refused it.
What's new since 1.5.0
-
Letters as coefficients in solve (CAS 038, picked on Michael's "whatever is good/next for casimir"). An equation linear in var whose coefficients hold another letter or π, as typed, is solved for var:
A×X+B=C×X+DisSOLVED: X=-(B-D)÷(A-C); A-C≠0. Before it prints, the answer is checked as two identities: L − R is a×var + b, and a×x + b is 0. Its rows say where it holds:- a≠0, read as a divisor is: each letter factor (
P×V=N×R×Tfor T isT=P×V÷N÷R; N≠0; R≠0), LN(u)'s u≠1, and none for a factor never 0; - each condition of the typed text, read at the answer: SQRT's u≥0 and ACOSH's u≥1 (new in solve's rows), u>0, LN's u≠1 and TAN's COS(u)≠0.
X=SQRT(A)÷BisSOLVED: X=SQRT(A)÷B; A≥0; B≠0,LN(A)×X=1isSOLVED: X=1÷LN(A); A>0; A≠1, and u≠0 with u≥0 on one u are u>0:SQRT(A)×X=1isSOLVED: X=1÷SQRT(A); A>0; - a condition with no letter that fails is UNDEFINED, as in sum, limit and defint:
A×X=LN(-1).
A text with neither a letter nor π prints as in 1.5.0, so
X÷0=1is still NOT_NUMERIC. - a≠0, read as a divisor is: each letter factor (
-
Steps for these (CAS 039). They page as a numeric linear equation pages, with no new rule for the calculator to name: BOTH SIDES (POLYNOMIAL for an expression), L − R gathered as collect prints it, left out where it is the text as typed; LINEAR at depth 1, with the answer and the rows of var's coefficient, where the division happens; and ANSWER, solve's line with every row.
V=I×Rfor R pages asBOTH SIDES: -I×R+V=0,LINEAR: -I×R+V=0: R=V÷I; I≠0,ANSWER: SOLVED: R=V÷I; I≠0. The typed text's own rows (LN's A>0) stay in ANSWER. -
What moves, against 1.5.0:
- the golden corpus: none of its 10,822 lines;
- the documented examples: none of 1.5.0's 1,230 commands;
- the developer tests: two of tests/solve_test.c's refusals now solve,
Y×X-1(SOLVED: X=1÷Y; Y≠0) andπ×X-1(SOLVED: X=1÷π); - the reviewer's hidden sets of the earlier units: byte for byte, but one line,
(6×X+1)-Bfor X. It was NOT_NUMERIC and isSOLVED: X=B÷6-1÷6, and its steps page through POLYNOMIAL.
-
Measured: S3 code 269,805 bytes (4,611 more than 1.5.0's), 531 under decision 92's 270,336 B line. Decision 94 lets the line rise to 301,056 B, and 1.6.0 doesn't use it. On the ESP32-S3 the bench's worst stack is unchanged.
The guarantee is unchanged: an exact answer, or a status that says why not. Never an approximation presented as an answer.
The known wrong answers
None. The hidden sets of CAS 038 and 039 found no wrong value. CAS 038's first build refused a class of equations it should answer (var on both sides with a sum as its coefficient), and the build at acceptance answers them.
How it's tested (at this tag)
make test: 20 programs with 118,545 checks. They read every answer back through an independent evaluator (the calculator firmware's core) and compare values. New in 1.6.0: each letter solve's answer is put back on the core at letter values where its rows hold, and at each one-letter row's edge the equation has no value as typed or no longer depends on var.- The golden corpus, 10,822 calls compared byte for byte; none moved.
- Frame and symbol gates: no frame over 1 KiB, no
powlinked, and every function's frame against the recorded table. No frame moved from 1.5.0, on the host or the S3. The new functions take at most 304 B on the S3 (ls_steps). - Value checks beside the corpus judge Taylor polynomials, limits, integrals, sums and solve's steps against independent constructions. New in 1.6.0: solve's steps judge reads a letter solve's page by value (BOTH SIDES against the typed L − R at random letter values, LINEAR's value and rows, ANSWER).
- The docs: every
casimrow of docs/language.md and the units' example files is run (2,792 checks, each again with the series route compiled out), and every backticked answer in docs/language.md's prose must be in one of them or in the golden corpus. - CI on every push:
make checkunder gcc and clang,make test, CMake under gcc and clang, WebAssembly, and the ESP-IDF component for the ESP32-S3. - Mutants (scripts/mutants.py): 576 deliberate faults, and at this tag's tree
make testcatches 570 (work/v1.6.0-mutants.txt). The six that survive are 1.5.0's six, and searched again at this tree each moves exactly the lines it moved at 1.5.0's (work/v1.5.0-mutants.txt names what masks each). CAS 038's and 039's 33 are caught. - Hidden sets kept by the project's reviewer, independent of this repository, rerun at each unit's acceptance. At CAS 038's: its own, 225 rows at two arenas, 0 of 450 runs failing (after one hold, above), and the earlier units'. At CAS 039's: its own, 252 rows at three arenas, 0 of 756 failing, and the earlier units'.
- On the ESP32-S3, every one of the bench's 1,454 cases is byte-identical to the host (at 2da4e2f, whose code is this tag's but for the version text and a comment), and the steps of the 297 derivatives, integrals, solves and partial fractions page as the host's:
- with a 16,384 B arena, the size the calculator runs it with;
- worst stack 11,360 B read with interrupts masked, plus a 384 B interrupt allowance (two frames, or the most measured over the masked reading where that is larger), 11,744 of a 12 KiB task;
- nesting deeper than 44 levels is refused (NO_ROOM);
- about 270 KB of Casimir's code at -O2 (269,805 B), 531 B under the 270,336 B line.
Not yet, after 1.6.0
These print a line without the note it could have, print a row the answer doesn't need, have no value where the sum has one, refuse, or print a constant as written. None of them is a wrong value.
New with letters as coefficients (CAS 038 and 039):
- A coefficient that is 0 where canon can't see it prints SOLVED with a row that never holds:
(SIN(A)^2+COS(A)^2-1)×X=1isSOLVED: X=1÷(SIN(A)^2+COS(A)^2-1); SIN(A)^2+COS(A)^2-1≠0. Nothing printed is false, but the equation has no solution. - canon's forms, not a student's:
(X+A)÷B=CisX=B×(C-A÷B); B≠0,A×X+B=CisX=-(B-C)÷A; A≠0,F=9÷5×C+32for C isC=5×F÷9-160÷9, and steps' BOTH SIDES forV=I×Ris-I×R+V=0. - A row in the expanded coefficient's form:
A=(B+C)×H÷2for H has-B÷2-C÷2≠0, and(A×B-A)×X=Bhas-A+A×B≠0. - A row that always holds, or one left unsolved:
(SQRT(A)+1)×(X+1)=BhasSQRT(A)+1≠0,(SQRT(A)-1)×X=1hasSQRT(A)-1≠0(A≠1),ACOSH(A)×X=1hasA≥1; ACOSH(A)≠0(A>1), and(A^3-1)×X=1hasA^3-1≠0. - Refused, NOT_NUMERIC as 1.5.0 refuses them: var dropping out (
A×(X-1)=A×X), degree 2 or more with a letter (X^2=A), var inside a function or a denominator (A×SIN(X)=B,A÷X=1), a condition no row writes (ASIN's ABS(u)≤1, ATANH's ABS(u)<1), model 33's variable i, a constant with neither a letter nor π (SIN(1)×X=1), and an answer where the typed text has no value (X×(X-A)÷(X-A)=A). pw's first refusal decides whether letters are read at all:X^2÷X×A=1is NOT_POLYNOMIAL, andA×X^2÷X=1isSOLVED: X=1÷A; A≠0. Past six conditions, NO_ROOM (A×B×C×D×E×F×G×X=1).
Carried from 1.5.0, each example run again on this tag's command line, where it prints as at 1.5.0:
- The typed domain, stricter than it needs to be (CAS 035):
- limit's letter conditions at the edge:
SQRT(Y)^2×Xat 1 printsY; Y>0, though Y = 0 has a value, andASIN(Y)×Xat 1 printsASIN(Y); ABS(Y)<1, though |Y| = 1 has one; SQRT(X^2×Y)^2at 0 gives0; Y>0, though Y = 0 has a value;- limit's degenerate divisor tries k = 0 only for k linear in one letter:
(SIN((Y-Z)×X)+1)÷(SIN((Y-Z)×X)+1)at 0 prints1; Y-Z≠0, though at Y = Z it is 1 everywhere, andY×Z-1andY^2-1(1; Y≠-1; Y≠1) the same.
- limit's letter conditions at the edge:
- The typed domain, refused where it could be answered (CAS 035):
- an X×Y-form limit at ±∞ is UNDECIDED: its sign is Y's, and no row is given (
SQRT(X×Y)÷SQRT(X×Y)at ∞); - a divisor 0 for every X at a letter's value that casim can't rule out is UNDECIDED:
(Y×SIN(X)-SIN(Y×X))÷(Y×SIN(X)-SIN(Y×X))at 0, whose divisor is 0 for every X at Y = ±1; - rows count toward the 191 bytes, so some lines with them are TOO_LONG:
K×(1+Z×G)^Kfrom 1 to ∞, where G is u written asSQRT(u)^2,(u^(1÷4))^4,EXP(LN(u))orALOG(LOG(u))and u is Y+1, Y-2 or W-1, and(K-1)×(F(Z)×Y)^K÷F(Z)from 1 to N for ACOSH, ASINH and ATANH, and LN from 2.
- an X×Y-form limit at ±∞ is UNDECIDED: its sign is Y's, and no row is given (
- Older checks beside the typed rows (found by 1.4.0's mutant search; 1.3.0 prints the same):
- which refusal a constant with no value gets beside a pole on the interval:
X^(-2)×ACOS(3)over 0..1 is IMPROPER, where UNDEFINED would be just as lawful (CAS 009's interval check answers first); - limit rows from the answer's own walk and f's (CAS 023 sites 6 and 7) that restate a typed row (
K×π+4≠0besideK×π÷2+2≠0) or can't fail (EXP(K^2)≠0), and 69 limit lines found TOO_LONG where the line without site 6's rows fits.
- which refusal a constant with no value gets beside a pole on the interval:
- A quotient that canon cancels, or whose divisor falls in K, is still refused.
X^(2×K)÷X^Kis NO_CLOSED_FORM: canon reads it asX^K, which has a value at X = 0, where its terms are 0÷0. So isX^K÷Y^(-K): its divisor has no value at Y = 0, while canon's(X×Y)^Khas one. Answering them needs a row that(X÷Y)^K-style twins don't print (X≠0,Y≠0). So areY^K÷Y^K(whereY^K×Y^(-K)printsN; Y≠0) and a divisor in K inside an exponent:(-8)^(1÷(K-1))from 0 is NO_CLOSED_FORM, though the term at K = 1 has no value (UNDEFINED would say why). - A divisor of degree 2 or more in K that canon cancels is read by no CAS 037 rule. The shapes found are refused (
((K^2-1)×K)÷(K^2-1)from 0 to N is NO_CLOSED_FORM) or UNDEFINED already. - UNDEFINED for a whole sum to N whose divisor is 0 mid-range (CAS 037's (C), as for a divisor canon keeps):
((K+1)^2-1)÷((K+1)-1)from -2 to N is UNDEF...
Casimir 1.5.0
Casimir 1.5.0: the wrong answers found after 1.4.0 are fixed. A closed sum whose summand holds a power with K in its exponent, or a divisor in K that canon cancels, no longer prints a value where the sum as typed has none. A sum to N whose line had no value at a point where the sum has one, the empty sum, gets the row it was missing, and limit no longer adds 1.4.0's named row for a k linear in one letter. 1.5.0's code is CAS 036's and CAS 037's, accepted at ead17ba and 8b6f1d7, with the version text changed. src/casim.h's interface is 1.0's, but answers' bytes change, which is why this is 1.5.0 and not 1.4.1. A moved answer keeps its value. It gains rows that exclude a point where the line has no value, and limit's moved lines lose a row that excluded nothing. A status moves where 1.4.0 printed a value the sum doesn't have (now UNDEFINED), where a refusal now says why (NO_CLOSED_FORM to UNDEFINED), where one line once TOO_LONG now fits, and where a refusal changes kind with the arena (below).
What's new since 1.4.0
-
A row where a sum to N's line misses a value (CAS 036, Michael's cards cas-missing and cas-missing-scope). At the empty sum (N = LO-1) the sum is 0. Where the line has no value there, a row now excludes the point. Each row excludes its point at every N, though the sum has a value there at the empty sum, so it is stricter than the sum by that one N. The rows give way at the cap: where the line with them would pass 191 bytes, they go together, and the line is 1.4.0's byte for byte.
- a typed condition whose hole the answer shows:
(1+Z×SQRT(Y))^Kfrom 0 to N gainsY≥0,(1+Z×LN(Y))^KgainsY>0, andY^K÷Xfrom 1 gainsX≠0. Where the new row says a letter's row, that row goes:(1+Z×ATANH(Y+1))^Kfrom 0 printsABS(Y+1)<1in place ofY≠0; - the ratio's monomial pole from a LO of 1 or more:
Y^(-K)and(3÷Y)^Kfrom 1 to N gainY≠0; - the base of a divisor whose exponent alone holds K, and of u^K from a LO below 0:
X^K÷(Y-1)^Kfrom 1 gainsY≠1, and(SIN(2×Z)×Y)^Kfrom -2 gainsSIN(2×Z)≠0; - a stricter row that says a hard one can bring a line under the cap:
((K-1)+1)×LN(Z)^((K-1)+1)×Y^((K-1)+1)from 0 to N, TOO_LONG in 1.4.0, prints its line withZ>0.
- a typed condition whose hole the answer shows:
-
A closed sum reads a power with K in its exponent as typed (CAS 037, the fix; Michael's card cas037-wrong). Summed term by term, 1.4.0 already read each term as typed (
Y^K×Y^(-K)from -1 to 1 is3; Y≠0). Its closed form merged the powers first and could print a value where the sum has none, and 1.3.0 printed the same lines:- a factor that cancels the base's negative powers:
(Y×SQRT(Z))^K×SQRT(Z)from -1 to N printed a line that is 1 at Y = 1, Z = 0, where the term at K = -1 is SQRT(0)^(-1). It gainsZ≠0, andY^K×Yfrom -1 gainsY≠0. The same holds for SIN, COS, TAN, ABS, LN and the other functions at their zeros, and to ∞; - powers that cancel exactly, from any LO:
Y^K×Y^(-K)from 1 to N printedNand printsN; Y≠0. The exponent is read where it is negative for some K from LO, at LO, LO+1 or LO+2 or falling from there; - a divisor in K alone that canon cancels, 0 at a whole K in range:
((K+1)^2-1)÷((K+1)-1)from 0 to N printed5×N÷2+N^2÷2+2and is UNDEFINED, as1÷(K-2)from 0 to N already was.(-8)^(1÷(-K))from 0, NO_CLOSED_FORM in 1.4.0, is UNDEFINED too.
These rows are hard: the line without one would have a value where the sum has none, so they never give way at the cap.
- a factor that cancels the base's negative powers:
-
limit tries k = 0 for k linear in one letter again (CAS 036). 1.4.0 left the try out to fit its flash line, so
(SIN((Y-1)×X)+1)÷(SIN((Y-1)×X)+1)at 0 printed1; Y≠1. It prints1, as the input is 1 everywhere at Y = 1. A k in two letters, or a letter's square, keeps its row (Not yet). -
The S3 code line is gated. Michael's decision 92 set Casimir's line at 270,336 bytes.
make sizenow fails past it. -
Prose answers are checked. 1.4.0's docs/language.md kept 1.3.0's line for Σ K×(SIN(Z)×Y)^K÷SIN(Z) from 1, and its reason, after CAS 035 moved the answer, and it named
X^(1÷2)as a fraction casim prints (it prints SQRT(X)). Both are fixed (07d8a91), and the 1.4.0 release page carries the erratum. Every backticked answer in that file's prose must now be in a checked row or in the golden corpus (3d7c1d3). -
What moves, against 1.4.0:
- the golden corpus: 3 of 10,822 lines, and their 3 under →DEC (
X^(-K)from 1 to N gainsX≠0); - the documented examples: 43 of 1.4.0's commands, each by a row;
- a family of 10,271 sums: 1,253, by rows only;
- the 1.4.0 mutant search's inputs at 65,536 B: 1,764 of 35,388 sum lines and 23 of 162,486 limit lines, with defint and taylor unchanged (0 of 406,524). Of the sum lines, 1,754 move by rows, 3 go from OK to UNDEFINED and 6 from NO_CLOSED_FORM to UNDEFINED (CAS 037's (C)), and 1 goes from TOO_LONG to OK. No answer moves;
- at the calculator's arenas the same inputs move the same way, and a refusal changes kind: 3 sum lines go from TOO_LONG to NO_ROOM at 8,800 B and 1 at 16,384 B, each TOO_LONG at 65,536 B. The rows take arena.
- the golden corpus: 3 of 10,822 lines, and their 3 under →DEC (
-
The hidden sets at acceptance: CAS 036's has 12 failing runs, each ruled (decision 93). Six are under Not yet below, and six are the row's exclusion at N = 0 from a LO of 0, where the one term is u^0 = 1 (below). CAS 037's, 189 rows at three arenas, has 18, each a CAS 037 row's exclusion at N = LO of the same kind (below). In CAS 035's set, two lines lose a letter's row that a kept row says, accepted.
-
Measured: S3 code 265,194 bytes (3,158 more than 1.4.0's), 5,142 under the line. On the ESP32-S3 the bench's worst stack is unchanged.
The guarantee is unchanged: an exact answer, or a status that says why not. Never an approximation presented as an answer.
The known wrong answers
None. Those found at CAS 036's acceptance, and printed by 1.4.0 and 1.3.0 as well, print CAS 037's rows (above). make check now holds 596 sums of their shapes (scripts/wrong-values.py), each added term by term from the typed summand: a line with a value where the sum has none fails.
How it's tested (at this tag)
make test: 20 programs with 118,449 checks. They read every answer back through an independent evaluator (the calculator firmware's core) and compare values.- The golden corpus, 10,822 calls compared byte for byte. CAS 036 moves 3, each listed with its before (tests/golden/moved/cas-036.txt). A check judges each added row apart from casim: at a letter's value it excludes, the line has no value at the empty sum. CAS 037 moves none.
- Frame and symbol gates: no frame over 1 KiB, no
powlinked, and every function's frame against the recorded table. No frame moved from 1.4.0, on the host or the S3. CAS 036's three new functions take at most 96 B on the S3 (td_at0), and CAS 037's five at most 224 B (td_kq). - Value checks beside the corpus judge Taylor polynomials, limits, integrals and sums against independent constructions. New in 1.5.0: the sum judges read a function's domain row (
Z>0for LN(Z)) as the typed domain's. - The docs: every
casimrow of docs/language.md and the units' example files is run (2,602 checks, each again with the series route compiled out), and every backticked answer in docs/language.md's prose must be in one of them or in the golden corpus. - CI on every push:
make checkunder gcc and clang,make test, CMake under gcc and clang, WebAssembly, and the ESP-IDF component for the ESP32-S3. - Mutants (scripts/mutants.py): 543 deliberate faults, and at this tag's tree
make testcatches 537 (work/v1.5.0-mutants.txt, which names what masks each survivor). Five survived at 1.4.0 too: cancel-rewind-keeps-cache, which differs from the build on no known input, and four older checks beside CAS 035's typed walk. Searched again at this tree over the 1.4.0 search's 2,220,409 lines, each of the four moves exactly the lines it moved at 1.4.0. div-letter-off moves the same 16 sum lines by rows the kept rows imply and ser-ln-no-value the same limit line on the series-only build; the other two move none. The sixth, factor026-letter-off, takes off CAS 026's row for a letter whose fraction power a term from LO divides by. CAS 037's row on a power with K in its exponent is the same row, soK×SQRT(Z)^(2×K-1)×Y^Kfrom 0 to ∞, which caught it at 1.4.0, keepsZ≠0with the mutant, and no searched line separates it from the build. One of 1.4.0's, ratio025-negpow-row-off, survived at CAS 036's tree: CAS 036 moved work/033-examples.txt:55, its catcher, to the line the mutant prints. CAS 037 put the row back, and it is caught again. - A check that no sum line has a value where the typed sum has none (CAS 037, scripts/wrong-values.py): 596 sums, each added term by term from the typed summand. Through CAS 036's build it fails 253 of them.
- Hidden sets kept by the project's reviewer, independent of this repository, rerun at each unit's acceptance. At CAS 036's: its own, 293 rows at three arenas, with the 12 failing runs ruled (above), and the earlier units'. At CAS 037's: its own, 189 rows at three arenas, the earlier units' and a comparison of every line against CAS 036's build. That comparison found two right answers CAS 037's first build took past the cap (a row said twice, in another spelling), fixed before acceptance.
- On the ESP32-S3, every one of the bench's 1,412 cases is byte-identical to the host (at aaf5ace, whose src is this tag's but for the version text):
- with a 16,384 B arena, the size the calculator runs it with;
- worst stack 11,360 B read with interrupts masked, plus a 384 B interrupt allowance (two frames, or the most measured over the masked reading where that is larger), 11,744 of a 12 KiB task;
- nesting deeper than 44 levels is refused (NO_ROOM);
- about 265 KB of Casimir's code at -O2 (265,194 B), 5,142 B under the 270,336 B line.
Not yet, after 1.5.0
These print a line without the note it could have, print a row the answer doesn't need, have no va...
Casimir 1.4.0
Casimir 1.4.0: the wrong answers found after 1.3.0 are fixed. sum, defint and limit read the input's domain as typed, so none of them prints a value where the input has none, and a row whose other letters sit inside a function with a value everywhere now says the letter's row. 1.4.0's code is CAS 034's and CAS 035's, accepted at 4c0523b and a6836d0, with the version text changed. src/casim.h's interface is 1.0's, but answers' bytes change, which is why this is 1.4.0 and not 1.3.1: a moved answer has the same value wherever both lines have one, and the rows and refusals it gains exclude only points where the input has no value, with one named cost (below).
What's new since 1.3.0
- The input's domain, as typed (CAS 035, the fix). canon's domain rule lets it define a result where the input has none (u÷u is 1, EXP(LN(u)) is u, SQRT(u)^2 is u). A walk over the typed input now lists each part's condition: a divisor not 0, LN's and LOG's argument above 0, a root's not below 0, ASIN's and ACOS's argument within ±1, ACOSH's at least 1, ATANH's within ±1, TAN's cosine not 0. Each op reads the list:
- a sum prints a row for each condition in the other letters, unless a row already says it or the answer shows the hole itself:
(Y÷Y)^Kfrom 1 to N isN; Y≠0(1.3.0:N), and(Z×EXP(LN(Y))+1)^Kfrom 0 to ∞ is-1÷Y÷Z; Y>0; ABS(Y×Z+1)<1. Summed term by term, the input is read with each K put in, so an exponent in K counts:Y^K×Y^(-K)from -1 to 1 is3; Y≠0; - defint is IMPROPER where the typed integrand has no real value on the interval (
SQRT(X)^2over -2..-1, 1.3.0:-3÷2), and BAD_ARG where another letter's value decides it (X×Y÷Yover 0..1, 1.3.0:1÷2); - limit is UNDEFINED on a side with no real value near the point:
EXP(LN(X))at -1 (1.3.0:-1),SQRT(X)at -1 (1.3.0:SQRT(-1)), andSQRT(X)at 0 two-sided, while from above it is still0. A condition in another letter is CAS 023's row:X×Y÷Yat 1 is1; Y≠0; - limit, a divisor 0 for every X at a letter's value:
SIN(K×X)÷SIN(K×X)at 0 is1; K≠0(1.3.0:1), since at K = 0 it has no value anywhere. Where casim can't show that no other letter's value does the same, the limit is UNDECIDED; - the closed form's cancelled divisor is a typed divisor too:
K×(X×Y)^K÷Yfrom 1 to ∞ gainsY≠0.
- a sum prints a row for each condition in the other letters, unless a row already says it or the answer shows the hole itself:
- A term-by-term sum's rows count toward the 191 bytes. Where they take the line past it, the closed form answers instead:
(ACOS(Z)^2×Y)^K÷ACOS(Z)from 1 to 12 gainsACOS(Z)≠0(the input has no value at Z = 1, where 1.3.0's line is 0), and its 12 terms with the row take 198 bytes, so it prints-Y×ACOS(Z)÷(Y×ACOS(Z)^2-1)+Y^13×ACOS(Z)^25÷(Y×ACOS(Z)^2-1); ACOS(Z)≠0; Y×ACOS(Z)^2≠1. - A row whose other letters are inside a function with a value everywhere says the letter's row (CAS 034). CAS 032 let a kept row say
L≠0only where its other letters were in a sum, a product or a whole power. ABS, SIN, COS, EXP, ALOG, SINH, COSH, TANH, ATAN and ASINH of those now count as well:(1+ABS(Y)×Z)^Kfrom 0 to N printsZ×ABS(Y)+1≠1without 1.3.0'sZ≠0, and(1+Z×Y)^Kfrom 0 to ∞ is-1÷Y÷Z; ABS(Y×Z+1)<1(1.3.0:-1÷Y÷Z; Y≠0; Z≠0; ABS(Y×Z+1)<1). - A sum to N below LO-1 (documented, no code change). A closed form to N holds from N = LO-1, the empty sum, up. Below that it is the formula carried on, which agreed with the subtracting convention wherever both have a value in every case measured (work/sum-below-lo.tsv).
- The named cost. To fit the flash budget, limit tries a letter's value only for a letter or its power. A divisor's k that is linear in a letter gets its row untried, so
(SIN((Y-1)×X)+1)÷(SIN((Y-1)×X)+1)at 0 prints1; Y≠1, though at Y = 1 the input is 1 everywhere. The row is never wrong, only stricter than it needs to be.(SIN((Y-1)×X)+Y)÷(SIN((Y-1)×X)+Y)does the same. - What moves: CAS 034 moves 934 of a family's 2,784 calls, 3 documented examples and none of the golden corpus. CAS 035 moves 37 of the golden corpus's 10,822 lines and their 37 under →DEC (34 to a refusal, 3 gaining a row), and 26 documented examples, each by a row. Of the earlier units' hidden sets, CAS 034 moves 116 runs in CAS 032's set and 2 in CAS 033's, each by its rule; at CAS 035's acceptance the only lines its judge failed were the ACOS sum's closed form (above), which the reviewer accepted by hand.
- Measured: on the ESP32-S3 at 16,384 B, the 13 new bench rows take at most 7,272 B of arena and 5,808 B of stack (a sum of a 40-deep nest of
1÷(, 59 ms), and the worst stack is unchanged. Against 1.3.0's code with CAS 034, CAS 035 changes no bench case's status; 51 cases need more arena (at most 1,976 B more, a 43-deep nest's limit at 4,864 B) and 14 more stack (at most 464 B more).
The guarantee is unchanged: an exact answer, or a status that says why not. Never an approximation presented as an answer.
The known wrong answers
None. Those found after 1.3.0 (canon's domain rule reaching sum, defint and limit, and limit reading SQRT outside its real domain) print the refusals and rows above.
How it's tested (at this tag)
make test: 20 programs with 118,449 checks. They read every answer back through an independent evaluator (the calculator firmware's core) and compare values.- The golden corpus, 10,822 calls compared byte for byte: CAS 034 moves none, and CAS 035 moves 37, each listed with its before (tests/golden/moved/cas-035.txt). A check judges each moved line apart from casim: the typed input, valued by an independent evaluator, has no value somewhere the line no longer gives one.
- Frame and symbol gates: no frame over 1 KiB, no
powlinked, and every function's frame against the recorded table. No frame moved from 1.3.0, on the host or the S3. CAS 035's 29 new functions take at most 384 B on the S3 (td_defint). - Value checks beside the corpus judge Taylor polynomials, limits, integrals and sums against independent constructions. New in 1.4.0: sums whose letter's row a row holding a total function may say (CAS 034), judged by the rule and by its sentence; and sums whose typed term has no value at the low bound, judged by a strict evaluator that gives no value to a divisor 0 to 40 digits or to LN(0).
- CI on every push:
make checkunder gcc and clang,make test, CMake under gcc and clang, WebAssembly, and the ESP-IDF component for the ESP32-S3. - Mutants (scripts/mutants.py): 519 deliberate faults, and at this tag's tree
make testcatches 514 (work/v1.4.0-mutants.txt). At the version commit it caught 509: nine faults in older checks, each caught at CAS 034's tree, survived because CAS 035's typed rows print the same line. A search of generated inputs, each mutant against the build, found the lines where five of those checks alone decide, and 6090b8c pins them (rows only). Five survive, each named in work/v1.4.0-mutants.txt with what masks it: cancel-rewind-keeps-cache, as at 1.3.0; cond-site2-lhopital and ser-root-of-negative, which no searched line separates from the build; div-letter-off, which drops only rows the kept rows imply; and ser-ln-no-value, which differs from the library build on no line. With the series route forced alone it differs on one, where 1.4.0 refuses a line the mutant gets right. Some pins hold what 1.3.0 printed, and no ruling stands behind them:K^Xat 0.5 isSQRT(K); K>0. At K = 0 the typed text has a value (0^X is 0 for X>0, and the limit is 0, which is SQRT(0)), soK≥0would be exact; the stricter row is lawful at that edge.- The three defint pins keep IMPROPER where the integrand also holds a constant with no value (
X^(-2)×ACOS(3)over 0..1). UNDEFINED would be just as lawful, and arguably the plainer word. defint's interval check is redundant to the output except for this choice of word.
- Hidden sets kept by the project's reviewer, independent of this repository, rerun at each unit's acceptance. At CAS 034's: its own, 744 runs, none failing; at CAS 035's: its own, 1,074 rows at 8,800, 16,384 and 65,536 B, none failing; and the earlier units' sets.
- On the ESP32-S3, every one of the bench's 1,412 cases is byte-identical to the host:
- with a 16,384 B arena, the size the calculator runs it with;
- worst stack 11,360 B read with interrupts masked, plus a 384 B interrupt allowance (two frames, or the most measured over the masked reading where that is larger), within a 12 KiB task;
- nesting deeper than 44 levels is refused (NO_ROOM);
- about 262 KB of Casimir's code at -O2 (262,036 B, 13,674 more than 1.3.0's), 108 B under the 262,144 B budget.
Not yet, after 1.4.0
These print a line without the note it could have, print a row the answer doesn't need, have no value where the sum has one, refuse, or print a constant as written. None of them is a wrong value.
- The typed domain, stricter than it needs to be (CAS 035):
- limit's letter conditions at the edge:
SQRT(Y)^2×Xat 1 printsY; Y>0, though Y = 0 has a value, andASIN(Y)×Xat 1 printsASIN(Y); ABS(Y)<1, though |Y| = 1 has one; SQRT(X^2×Y)^2at 0 gives0; Y>0, though Y = 0 has a value;- a divisor's k linear in a letter gets its row untried (the named cost, above).
- limit's letter conditions at the edge:
- The typed domain, refused where it could be answered (CAS 035):
- an X×Y-form limit at ±∞ is UNDECIDED: its sign is Y's, and no row is given (
SQRT(X×Y)÷SQRT(X×Y)at ∞); - a divisor 0 for every X at a letter's value that casim can't rule out is UNDECIDED:
(Y×SIN(X)-SIN(Y×X))÷(Y×SIN(X)-SIN(Y×X))at 0, whose divisor is 0 for every X at Y = ±1; - rows count toward the 191 bytes, so some lines with them are TOO_LONG:
K×(1+Z×F(u))^Kfrom 1 to ∞ with u Y+1, Y-2 or W-1, and(K-1)×(F(Z)×Y)^K÷F(Z)from 1 to N for ACOSH, ASINH and ATANH, and LN from 2.
- an X×Y-form limit at ±∞ is UNDECIDED: its sign is Y's, and no row is given (
- Older checks beside the typed rows (found by 1.4.0's mutant search; 1.3.0 prints the same):
- which refusal a constant with no value gets beside a pole on the interval: `X^(-2...
Casimir 1.3.0
Casimir 1.3.0: the known wrong answer found after 1.2.0 is fixed, and a sum's line no longer prints a letter's row that another row already says. Summed from 0 to ∞, (Z÷Y+1)^K printed -Y÷Z; Z≠0; ABS(Z÷Y+1)<1 in every release from 1.0.0 to 1.2.0. That reads 0 at Y = 0, where every term from K = 1 has no value, so the sum has none, and no row excluded the point. It now prints -Y÷Z; Y≠0; Z≠0; ABS(Z÷Y+1)<1. 1.3.0's code is CAS 032's and CAS 033's, accepted at fde2bc3 and 63b9215, with the version text changed. src/casim.h's interface is 1.0's, but answers' bytes change, which is why this is 1.3.0 and not 1.2.1, as for 1.1.0 and 1.2.0: a moved answer has the same value wherever both lines have one, and the rows it gains exclude only points where the sum has no value.
What's new since 1.2.0
- A divisor inside the ratio, to ∞ (CAS 033, the fix). Where the ratio holds a letter under a division inside a sum, every term from K = 1 has no value where the divisor is 0, and the closed form can cancel the divisor:
(Z÷Y+1)^Kand(Z÷Y^2+1)^Kfrom 0 to ∞ getY≠0, the letter's row, where the ratio itself has no value at the letter's 0;(Z÷(Y-1)+1)^KgetsY-1≠0, and(Z÷SIN(Y)+1)^KgetsSIN(Y)≠0: the divisor as the summand reads it, before the closed form;- where the answer itself has no value with the divisor at 0, it shows the hole, and no row is added:
K×(Z÷(Y-1)+1)^Kfrom 1 to ∞ prints its 188 B line as 1.2.0 did; - a divisor that is never 0 (EXP, COSH, 2^Y) gets none. To N nothing moves: each answer keeps the divisor, so it has no value where the sum has none.
- A letter's row that another row says, left out (CAS 032). A letter's row
L≠0goes where a kept row has a value at L = 0 and is false there, whatever the other letters are:COS(Z)^Kfrom 1 to N is-COS(Z)÷(COS(Z)-1)+COS(Z)^(N+1)÷(COS(Z)-1); COS(Z)≠1, since COS(0) is 1. 1.2.0 printedZ≠0beside it;X^(-K)from 0 to ∞ is-1÷(1÷X-1); ABS(X)>1,(Y÷SIN(Z))^Kfrom 1 to N printsSIN(Z)≠0; Y÷SIN(Z)≠1, and(K+1)×((Z-1)×Y)^Kfrom -1 to ∞ prints-Y+Y×Z≠0; ABS(-Y+Y×Z)<1, each without 1.2.0's letter row;EXP(X)^Kfrom 0 to N printsEXP(X)≠1alone, thoughX≠0names the same point more plainly. That reverses a choice made in 1.0's design, the plainer row first, and is the change's named cost.
- How: a row says
L≠0only where every other letter in it is in a sum, a product or a whole power not below 0, so that it is false at L = 0 whatever they are:Z÷Y+1≠1has no value at Y = 0 and says nothing, andZ×ABS(Y)+1≠1keepsZ≠0(below). A row that gives way at the cap says it only on a line that keeps that row; where it gives way, the letter's row comes back, and the line is 1.2.0's. - The bytes it frees make room at the cap:
K×SIN(Z)^(K-1)from -1 to N prints…; SIN(Z)≠0; SIN(Z)≠1in 189 B, where 1.2.0'sSIN(Z)≠0gave way (1.2.0's Not yet), andK×COSH(Z)^(K-1)from 1 to N, TOO_LONG in 1.2.0, fits in 187 B. - What moves: CAS 032 moves 66 of the 1,280 calls of the 640 function-factor probe sums at two arenas, and 290 of a family's 5,280; 18 of the documented examples; 3 lines of the golden corpus, all
EXP(X)^Kfrom 0 to N. CAS 033 moves only its own class on those sets (12 calls), one example, and nothing in the golden corpus. Of the earlier units' hidden sets, CAS 032 moves 298 runs and CAS 033 20, all in CAS 032's set, each by its rule; the rest are 1.2.0's byte for byte. - Measured: on the ESP32-S3 at 16,384 B, the ten new bench rows take at most 5,840 B of arena and 2,672 B of stack.
The guarantee is unchanged: an exact answer, or a status that says why not. Never an approximation presented as an answer.
The known wrong answers
None. 1.0.0 to 1.2.0's, (Z÷Y+1)^K from 0 to ∞ reading 0 at Y = 0 with no row, prints its row now (above).
Found after this release, on 26 Oct 2026. canon may drop a part of the input that has no value at some points, since its domain rule lets it define a result where the input has none (u÷u is 1, u-u and 0×u are 0, u^0 is 1, EXP(LN(u)) and ALOG(LOG(u)) are u, SQRT(u)^2 is u). sum, defint and limit then print a value where the input as typed has none:
- Summed from 1 to N,
(Y÷Y)^KprintsN. At Y = 0 every term is 0÷0, so the sum has no value, whileNhas one. Summed from 0 to ∞,(Z×LN(Y)÷LN(Y))^Kprints-1÷(Z-1); ABS(Z)<1, which has a value at Y = 1, where every term holds LN(1)÷LN(1), and at every Y ≤ 0, where LN(Y) has none. - Summed from 0 to ∞,
(Z×EXP(LN(Y))+1)^Kprints-1÷Y÷Z; Y≠0; Z≠0; ABS(Y×Z+1)<1. At Y = -1 and Z = 1÷2 every term holds LN(-1), which has no real value, so the sum has none, while the answer is 2 there and each row holds. With ALOG(LOG(Y)), SQRT(Y)^2 or SQRT(Y)×SQRT(Y) in place of EXP(LN(Y)) it prints the same line. defint "SQRT(X)^2" X -2 -1prints-3÷2, though SQRT(X) has no real value on the interval (SQRT(X)over it is IMPROPER).SQRT(X)×SQRT(X)and(X^(1÷4))^4print the same, and0×SQRT(X)prints0.limit "EXP(LN(X))" X -1 ''prints-1, though LN(X) has no real value near -1.- With another letter, a 0÷0 reaches limit and defint as well:
defint "X×Y÷Y" X 0 1prints1÷2andlimit "X×Y÷Y" X 1 ''prints1, each with a value at Y = 0, where the input has none. - Where the sum's closed form cancels a divisor of the summand: summed from 1 to ∞,
K×(X×Y)^K÷Yprints-X÷(X×Y-1)-X^2×Y÷2÷(X×Y-X^2×Y^2÷2-1÷2); ABS(X×Y)<1. At Y = 0 the term at K = 1 is 0÷0, so the sum has no value, while the answer is X. From 1.0.1,K×(SIN(Z)×Y)^K÷SIN(Z)does the same where SIN(Z) is 0, and(K-1)×(TAN(Z)×Y)^K÷TAN(Z)where TAN(Z) is 0 (1.0.0 printed both with no value there).
limit also reads a function outside its real domain as typed, with no canon involved: limit "SQRT(X)" X -1 '' prints SQRT(-1), and from below 0 it prints 0, though SQRT(X) has no real value near -1 or below 0.
simplify promises no more than the domain rule (docs/canonical-form.md), and expand, collect, subst, derive and integrate answer through canon under the same rule. taylor and the polynomial ops read what was typed: taylor "EXP(LN(X))" X -1 2 prints UNDEFINED, and solve, factor, cancel, partfrac and pdiv refuse these forms (NOT_POLYNOMIAL). 1.0.0 to 1.3.0 print every line above. The project's reviewer found it while testing a later unit, and a fix is being made (CAS 035).
How it's tested (at this tag)
make test: 20 programs with 118,439 checks. They read every answer back through an independent evaluator (the calculator firmware's core) and compare values.- The golden corpus, 10,822 calls compared byte for byte: CAS 032 moves 3, each its line less
X≠0, and CAS 033 none. - Frame and symbol gates: no frame over 1 KiB, no
powlinked, and every function's frame against the recorded table. No frame moved from 1.2.0, on the host or the S3. The new functions take 128 B (sm_said), 80 (sm_false0), 48 (sm_whole), 112 (sm_rowd_), 8 (sm_rowd), 80 (sm_holeb) and 80 (sm_rowfd) on the host, and 64, 64, 48, 64, 32, 48 and 48 B on the S3. - Value checks beside the corpus judge Taylor polynomials, limits, integrals and sums against independent constructions: power series over fractions, numeric derivatives at 90 digits, and sums added term by term at 60 digits. New in 1.3.0: sums whose letter's row another row may say, predicted by construction and judged both ways (a row the rule drops but the line keeps, and a drop the rule doesn't allow); and sums holding a divisor inside the ratio, judged at each of its exact zeros, where a value with no row excluding it is a failure.
- CI on every push:
make checkunder gcc and clang,make test, CMake under gcc and clang, WebAssembly, and the ESP-IDF component for the ESP32-S3. - 477 planted mutants, each a fault the checks must catch: 475 are caught at this tag's tree (work/v1.3.0-mutants.txt), the 13 new on CAS 032's and CAS 033's rows among them. Two are named: cancel-rewind-keeps-cache, and ratio025-negpow-mono-row, whose fault CAS 032 made invisible. No input is known where either differs from the build.
- Hidden sets kept by the project's reviewer, independent of this repository, rerun at each unit's acceptance. At CAS 032's: its own, 1,456 runs at four arenas, judged by the rule (what it drops, what it keeps, and the give-way at the cap); at CAS 033's: its own, 484 runs at two arenas, judged at each divisor's exact zeros; and the earlier units' sets.
- On the ESP32-S3, every one of the bench's 1,394 cases is byte-identical to the host:
- with a 16,384 B arena, the size the calculator runs it with;
- worst stack 11,360 B read with interrupts masked, plus a 384 B interrupt allowance (two frames, or the most measured over the masked reading where that is larger), within a 12 KiB task;
- nesting deeper than 44 levels is refused (NO_ROOM);
- about 248 KB of Casimir's code at -O2 (248,362 B, 2,027 more than 1.2.0's).
Not yet, after 1.3.0
These print a line without the note it could have, print a row the answer doesn't need, have no value where the sum has one, refuse, or print a constant as written. None of them is a wrong value.
- A quotient that canon cancels, or whose divisor falls in K, is still refused.
X^(2×K)÷X^Kis NO_CLOSED_FORM: canon reads it asX^K, which has a value at X = 0, where its terms are 0÷0. So isX^K÷Y^(-K): its divisor has no value at Y = 0, while canon's(X×Y)^Khas one. Answering them needs a row that(X÷Y)^K-style twins don't print (X≠0,Y≠0). - The
÷spelling reaches the rows of the×one. A quotient prints the answer its canon spelling did in 1.1.0, rows for nothing and missing values included. At a letter's 0, a quotient's terms whose power of the letter is 0 or less have values (0^0 = 1), and there the letter's row excludes the sum's value.X^K÷Y^Kfrom 0 to N prints-1÷(X÷Y-1)+(X÷Y)^(N+1)÷(X÷Y-1); Y≠0; X÷Y≠1, though at Y = 0 and N = 0 the sum is 1, asX^K×Y^(-K)printed in 1...
Casimir 1.2.0
Casimir 1.2.0: sum's closed form loses the factor each term shares with its denominator, so an answer that was 0÷0 at a letter's 0 or a factor's zero now has the sum's value there. (K+1)×SIN(Z)^(K+1)×Y^K from -1 to ∞ printed Y×SIN(Z)^2÷(Y×SIN(Z)-2×Y^2×SIN(Z)^2+Y^3×SIN(Z)^3); Y≠0; Z≠0; ABS(Y×SIN(Z))<1 in 1.1.2, which is 0÷0 at Z = π, where every term is 0. It now prints -SIN(Z)÷2÷(Y×SIN(Z)-Y^2×SIN(Z)^2÷2-1÷2); Y≠0; ABS(Y×SIN(Z))<1, which is 0 there. 1.2.0's code is CAS 031's, accepted at 3bc0b11, with the version text changed. src/casim.h's interface is 1.0's, but answers' bytes change, which is why this is 1.2.0 and not 1.1.3, as for 1.1.0: a moved answer has the same value wherever both lines have one.
What's new since 1.1.2
- The common factor, cancelled (CAS 031). K from lo to hi as said:
(K+1)×X^(K+1)×Y^Kfrom -1 to ∞ is-X÷2÷(X×Y-X^2×Y^2÷2-1÷2); Y≠0; ABS(X×Y)<1, which is 0 at X = 0, as the sum is. 1.1.2 printedX≠0there, a row for nothing;(K-1)×(SQRT(2)×X×Y)^K÷Xfrom 1 to ∞ is-X×Y^2÷(X×Y×SQRT(2)-X^2×Y^2-1÷2); ABS(X×Y×SQRT(2))<1, andK×(EXP(Z)×X)^K÷Xfrom 1 to ∞ is-X×EXP(Z)^2÷2÷(X×EXP(Z)-X^2×EXP(Z)^2÷2-1÷2)-EXP(Z)÷(X×EXP(Z)-1); ABS(X×EXP(Z))<1, each without 1.1.2'sX≠0;K×(SQRT(5)÷3×Y×SIN(Z))^K÷SIN(Z)from 1 to ∞ is-Y×SQRT(5)÷(Y×SIN(Z)×SQRT(5)-3)-5×Y^2×SIN(Z)÷6÷(Y×SIN(Z)×SQRT(5)-5×Y^2×SIN(Z)^2÷6-3÷2); ABS(Y×SIN(Z)×SQRT(5)÷3)<1, which is Y×SQRT(5)÷3 at Z = π, as the sum is. 1.1.2 had no value there;K×EXP(Z)^(K-1)from 1 to N, TOO_LONG in 1.1.2, fits:-1÷(EXP(Z)-1)-N×EXP(Z)^(N+1)÷(EXP(Z)-EXP(Z)^2)-EXP(Z)÷2÷(EXP(Z)-EXP(Z)^2÷2-1÷2)-EXP(Z)^(N+1)÷(EXP(Z)-EXP(Z)^2)+EXP(Z)^(N+1)÷2÷(EXP(Z)-EXP(Z)^2÷2-1÷2); Z≠0; EXP(Z)≠1.
- How: each term n×P^(-1) of the closed form, P a sum, loses the monomial n and P share: for each base n holds at a positive whole power, its least whole power over P's terms. The rows are made after, from the new answer, by the rules as they were. So a letter's row that only the 0÷0 called for goes, and where the answer now has a value at a factor's zero and a term divides by the factor, the factor's row stands:
(1+K)×(SIN(Z)×X×Y)^Kfrom -1 to ∞ printsSIN(Z)≠0in place of 1.1.2'sZ≠0, and(1+K)×(ABS(Z)×X×Y)^KprintsABS(Z)≠0, which excludes the same point. - It gives way. Where the cancel runs out of room, or where the cancelled line with its rows would pass 191 B, the line is 1.1.2's. Canon's spelling of a cancelled form can be the longer:
(K+1)×(Z-1)^K×Y^Kfrom -1 to ∞ prints its 188 B line as 1.1.2 did, never TOO_LONG for the cancel. - What moves: of the 640 function-factor sums the earlier units probed, 31 reword at both arenas, every one shorter or brought under the cap; of 2,640 more, 70 calls. 12 of the documented examples move. The golden corpus moves nowhere. Of the earlier units' hidden sets, 128 runs move (CAS 024's 14, 025's 50, 026's 14, 027's 30, 029's 4, 030's 18), each to a line with the same value wherever both have one; the rest are 1.1.2's byte for byte.
- Measured: on the ESP32-S3 at 16,384 B, the five new bench rows take at most 4,992 B of arena and 2,960 B of stack.
The guarantee is unchanged: an exact answer, or a status that says why not. Never an approximation presented as an answer.
The known wrong answers
None.
Found after this release, on 25 Oct 2026. Summed from 0 to ∞, (Z÷Y+1)^K prints -Y÷Z; Z≠0; ABS(Z÷Y+1)<1. Every term from K = 1 has no value at Y = 0, so neither has the sum, while the answer does: it reads 0 there, and no row excludes the point (a row with no value there doesn't exclude it). (Z÷(Y-1)+1)^K at Y = 1, (Z÷Y^2+1)^K at Y = 0, and (Z÷SIN(Y)+1)^K wherever SIN(Y) is 0 do the same to ∞. Summed to N, none of them has a value where the sum has none. The project's reviewer found it while testing a later unit. The fix is CAS 033, in the release after 1.2.0, which adds the rows.
Found after this release, on 26 Oct 2026. canon may drop a part of the input that has no value at some points, since its domain rule lets it define a result where the input has none (u÷u is 1, u-u and 0×u are 0, u^0 is 1, EXP(LN(u)) and ALOG(LOG(u)) are u, SQRT(u)^2 is u). sum, defint and limit then print a value where the input as typed has none:
- Summed from 1 to N,
(Y÷Y)^KprintsN. At Y = 0 every term is 0÷0, so the sum has no value, whileNhas one. Summed from 0 to ∞,(Z×LN(Y)÷LN(Y))^Kprints-1÷(Z-1); ABS(Z)<1, which has a value at Y = 1, where every term holds LN(1)÷LN(1), and at every Y ≤ 0, where LN(Y) has none. - Summed from 0 to ∞,
(Z×EXP(LN(Y))+1)^Kprints-1÷Y÷Z; Y≠0; Z≠0; ABS(Y×Z+1)<1. At Y = -1 and Z = 1÷2 every term holds LN(-1), which has no real value, so the sum has none, while the answer is 2 there and each row holds. With ALOG(LOG(Y)), SQRT(Y)^2 or SQRT(Y)×SQRT(Y) in place of EXP(LN(Y)) it prints the same line. defint "SQRT(X)^2" X -2 -1prints-3÷2, though SQRT(X) has no real value on the interval (SQRT(X)over it is IMPROPER).SQRT(X)×SQRT(X)and(X^(1÷4))^4print the same, and0×SQRT(X)prints0.limit "EXP(LN(X))" X -1 ''prints-1, though LN(X) has no real value near -1.- With another letter, a 0÷0 reaches limit and defint as well:
defint "X×Y÷Y" X 0 1prints1÷2andlimit "X×Y÷Y" X 1 ''prints1, each with a value at Y = 0, where the input has none. - Where the sum's closed form cancels a divisor of the summand: summed from 1 to ∞,
K×(X×Y)^K÷Yprints-X÷(X×Y-1)-X^2×Y÷2÷(X×Y-X^2×Y^2÷2-1÷2); ABS(X×Y)<1. At Y = 0 the term at K = 1 is 0÷0, so the sum has no value, while the answer is X. From 1.0.1,K×(SIN(Z)×Y)^K÷SIN(Z)does the same where SIN(Z) is 0, and(K-1)×(TAN(Z)×Y)^K÷TAN(Z)where TAN(Z) is 0 (1.0.0 printed both with no value there).
limit also reads a function outside its real domain as typed, with no canon involved: limit "SQRT(X)" X -1 '' prints SQRT(-1), and from below 0 it prints 0, though SQRT(X) has no real value near -1 or below 0.
simplify promises no more than the domain rule (docs/canonical-form.md), and expand, collect, subst, derive and integrate answer through canon under the same rule. taylor and the polynomial ops read what was typed: taylor "EXP(LN(X))" X -1 2 prints UNDEFINED, and solve, factor, cancel, partfrac and pdiv refuse these forms (NOT_POLYNOMIAL). 1.0.0 to 1.3.0 print every line above. The project's reviewer found it while testing a later unit, and a fix is being made (CAS 035).
How it's tested (at this tag)
make test: 20 programs with 118,439 checks. They read every answer back through an independent evaluator (the calculator firmware's core) and compare values.- The golden corpus, 10,822 calls compared byte for byte: CAS 031 moves none.
- Frame and symbol gates: no frame over 1 KiB, no
powlinked, and every function's frame against the recorded table. No frame moved from 1.1.2. The new functions take 240 B (sm_cancel), 304 B (sm_fits) and 96 B (sm_pw) on the host, and 176, 176 and 64 B on the S3. - Value checks beside the corpus judge Taylor polynomials, limits, integrals and sums against independent constructions: power series over fractions, numeric derivatives at 90 digits, and sums added term by term at 60 digits. New in 1.2.0: sums of the shape CAS 031 fixes (to ∞ from a negative lo, the term at lo 0), 60 a run (SIN, COS, TAN and SIN(2×Z) in each angle mode) and 45 held, judged at the factor's zeros, where no value is a failure.
- CI on every push:
make checkunder gcc and clang,make test, CMake under gcc and clang, WebAssembly, and the ESP-IDF component for the ESP32-S3. - 464 planted mutants, each a fault the checks must catch: 463 are caught at this tag's tree (work/v1.2.0-mutants.txt), seven of the eight new on CAS 031's cancel among them. The eighth, cancel-rewind-keeps-cache, is named: no input is known where it differs from the build.
- Hidden sets kept by the project's reviewer, independent of this repository, rerun at each unit's acceptance. At CAS 031's: its own, 794 runs at two arenas, judged by value at the letters' and the factors' zeros, by the rows, and by the give-way at the cap; and the earlier units' sets.
- On the ESP32-S3, every one of the bench's 1,384 cases is byte-identical to the host:
- with a 16,384 B arena, the size the calculator runs it with;
- worst stack 11,360 B read with interrupts masked, plus a 400 B interrupt allowance (two frames, or the most measured over the masked reading where that is larger), within a 12 KiB task;
- nesting deeper than 44 levels is refused (NO_ROOM);
- about 246 KB of Casimir's code at -O2 (246,335 B, 3,113 more than 1.1.2's).
Not yet, after 1.2.0
These print a line without the note it could have, print a row the answer doesn't need, have no value where the sum has one, refuse, or print a constant as written. None of them is a wrong value.
- A quotient that canon cancels, or whose divisor falls in K, is still refused.
X^(2×K)÷X^Kis NO_CLOSED_FORM: canon reads it asX^K, which has a value at X = 0, where its terms are 0÷0. So isX^K÷Y^(-K): its divisor has no value at Y = 0, while canon's(X×Y)^Khas one. Answering them needs a row that(X÷Y)^K-style twins don't print (X≠0,Y≠0). - The
÷spelling reaches the rows of the×one. A quotient prints the answer its canon spelling did in 1.1.0, rows for nothing and missing values included. At a letter's 0, a quotient's terms whose power of the letter is 0 or less have values (0^0 = 1), and there the letter's row excludes the sum's value.X^K÷Y^Kfrom 0 to N prints-1÷(X÷Y-1)+(X÷Y)^(N+1)÷(X÷Y-1); Y≠0; X÷Y≠1, though at Y = 0 and N = 0 the sum is 1, asX^K×Y^(-K)printed in 1.1.0.X^(K-1)÷Y^(K+2)from 1 to N printsX≠0, though at X = 0 the sum is 1÷Y^3 for every N ≥ 1. Where the factor is one of CAS 030's functions of a letter, its row excludes the point now:Y^K÷SIN(Z)^Kfrom 0 to N printsSIN(Z)≠0, though at Z = π and N = 0 the sum is 1 (rows for nothing, below). They...
Casimir 1.1.2
Casimir 1.1.2: sum's answer to N excludes a function factor's zero by its row, as its answer for a letter does. (Y÷SIN(Z))^K from 1 to N had no value at Z = π in 1.1.1, for N = 0, where the sum is 0, and no row said so. It now prints -Y÷(Y-SIN(Z))+(Y÷SIN(Z))^(N+1)÷(Y÷SIN(Z)-1); Z≠0; SIN(Z)≠0; Y÷SIN(Z)≠1, as its twin (Y÷X)^K prints X≠0. 1.1.2's code is CAS 030's, accepted at 8462e77, with the version text changed. src/casim.h's interface is 1.0's, and no answer's value changes: a line gains one row, or stays as 1.1.1 printed it, byte for byte.
What's new since 1.1.1
- A function factor's row to N (CAS 030). Each line is 1.1.1's plus one row, the factor's own, K from 1 to N unless said:
TAN(Z)^(2×K-1)×Y^Kfrom 0 to N is1÷(TAN(Z)-Y×TAN(Z)^3)-(Y×TAN(Z)^2)^(N+1)÷(TAN(Z)-Y×TAN(Z)^3); Z≠0; TAN(Z)≠0; Y×TAN(Z)^2≠1;(Y÷LOG(Z))^Kis-Y÷(Y-LOG(Z))+(Y÷LOG(Z))^(N+1)÷(Y÷LOG(Z)-1); Z≠0; LOG(Z)≠0; Y÷LOG(Z)≠1;(Y÷ACOS(Z))^Kfrom 0 to N is-1÷(Y÷ACOS(Z)-1)+(Y÷ACOS(Z))^(N+1)÷(Y÷ACOS(Z)-1); ACOS(Z)≠0; Y÷ACOS(Z)≠1;- in GRAD,
(Y÷COS(Z))^Kis-Y÷(Y-COS(Z))+(Y÷COS(Z))^(N+1)÷(Y÷COS(Z)-1); COS(Z)≠0; Y÷COS(Z)≠1. The row is the factor's own text in RAD, DEG and GRAD alike, since the test puts the factor to 0, not Z to a zero; GRAD's zero of COS is 100.
- How: a factor of the ratio gets its row where a term from lo divides by it (CAS 026), unless the answer itself has no value at the factor's zero. To N the sum has a value there at the empty sum, or at an N before the first term that divides, so that point was missing. Now, to N (an upper bound neither ∞ nor a whole number), SIN, COS, TAN, LN, LOG, ACOS and ACOSH of a letter keep their row there too, as a letter's row stands where the answer has no value at 0. ASIN, ATAN, SINH, TANH, ASINH and ATANH are 0 only where their letter is, and the letter's own row (
Z≠0) already excludes that. - It gives way at the cap. Where the line with the row would pass 191 B, the line is printed without it, as 1.1.0's note for solve gives way, never TOO_LONG for the row (Not yet, below).
- What stays as it was, byte for byte: every other answer 1.1.1 gave, and every value. The golden corpus moves nowhere. Of the earlier units' hidden sets, 36 runs gain the row (CAS 027's 20, CAS 029's 16), and the rest are 1.1.1's byte for byte.
- Measured: on the ESP32-S3 at 16,384 B, the five new bench rows take at most 4,312 B of arena and 2,608 B of stack.
The guarantee is unchanged: an exact answer, or a status that says why not. Never an approximation presented as an answer.
The known wrong answers
None.
Found after this release, on 25 Oct 2026. Summed from 0 to ∞, (Z÷Y+1)^K prints -Y÷Z; Z≠0; ABS(Z÷Y+1)<1. Every term from K = 1 has no value at Y = 0, so neither has the sum, while the answer does: it reads 0 there, and no row excludes the point (a row with no value there doesn't exclude it). (Z÷(Y-1)+1)^K at Y = 1, (Z÷Y^2+1)^K at Y = 0, and (Z÷SIN(Y)+1)^K wherever SIN(Y) is 0 do the same to ∞. Summed to N, none of them has a value where the sum has none. The project's reviewer found it while testing a later unit. The fix is CAS 033, in the release after 1.2.0, which adds the rows.
Found after this release, on 26 Oct 2026. canon may drop a part of the input that has no value at some points, since its domain rule lets it define a result where the input has none (u÷u is 1, u-u and 0×u are 0, u^0 is 1, EXP(LN(u)) and ALOG(LOG(u)) are u, SQRT(u)^2 is u). sum, defint and limit then print a value where the input as typed has none:
- Summed from 1 to N,
(Y÷Y)^KprintsN. At Y = 0 every term is 0÷0, so the sum has no value, whileNhas one. Summed from 0 to ∞,(Z×LN(Y)÷LN(Y))^Kprints-1÷(Z-1); ABS(Z)<1, which has a value at Y = 1, where every term holds LN(1)÷LN(1), and at every Y ≤ 0, where LN(Y) has none. - Summed from 0 to ∞,
(Z×EXP(LN(Y))+1)^Kprints-1÷Y÷Z; Y≠0; Z≠0; ABS(Y×Z+1)<1. At Y = -1 and Z = 1÷2 every term holds LN(-1), which has no real value, so the sum has none, while the answer is 2 there and each row holds. With ALOG(LOG(Y)), SQRT(Y)^2 or SQRT(Y)×SQRT(Y) in place of EXP(LN(Y)) it prints the same line. defint "SQRT(X)^2" X -2 -1prints-3÷2, though SQRT(X) has no real value on the interval (SQRT(X)over it is IMPROPER).SQRT(X)×SQRT(X)and(X^(1÷4))^4print the same, and0×SQRT(X)prints0.limit "EXP(LN(X))" X -1 ''prints-1, though LN(X) has no real value near -1.- With another letter, a 0÷0 reaches limit and defint as well:
defint "X×Y÷Y" X 0 1prints1÷2andlimit "X×Y÷Y" X 1 ''prints1, each with a value at Y = 0, where the input has none. - Where the sum's closed form cancels a divisor of the summand: summed from 1 to ∞,
K×(X×Y)^K÷Yprints-X÷(X×Y-1)-X^2×Y÷2÷(X×Y-X^2×Y^2÷2-1÷2); ABS(X×Y)<1. At Y = 0 the term at K = 1 is 0÷0, so the sum has no value, while the answer is X. From 1.0.1,K×(SIN(Z)×Y)^K÷SIN(Z)does the same where SIN(Z) is 0, and(K-1)×(TAN(Z)×Y)^K÷TAN(Z)where TAN(Z) is 0 (1.0.0 printed both with no value there).
limit also reads a function outside its real domain as typed, with no canon involved: limit "SQRT(X)" X -1 '' prints SQRT(-1), and from below 0 it prints 0, though SQRT(X) has no real value near -1 or below 0.
simplify promises no more than the domain rule (docs/canonical-form.md), and expand, collect, subst, derive and integrate answer through canon under the same rule. taylor and the polynomial ops read what was typed: taylor "EXP(LN(X))" X -1 2 prints UNDEFINED, and solve, factor, cancel, partfrac and pdiv refuse these forms (NOT_POLYNOMIAL). 1.0.0 to 1.3.0 print every line above. The project's reviewer found it while testing a later unit, and a fix is being made (CAS 035).
How it's tested (at this tag)
make test: 20 programs with 118,439 checks. They read every answer back through an independent evaluator (the calculator firmware's core) and compare values.- The golden corpus, 10,822 calls compared byte for byte: CAS 030 moves none.
- Frame and symbol gates: no frame over 1 KiB, no
powlinked, and every function's frame against the recorded table. No frame moved from 1.1.1. The new function, sm_rput, takes 80 B on the host and 48 B on the S3. - Value checks beside the corpus judge Taylor polynomials, limits, integrals and sums against independent constructions: power series over fractions, numeric derivatives at 90 digits, and sums added term by term at 60 digits. New in 1.1.2: a function factor's family in each angle mode (80 a run, and 18 held), judged at the mode's own zeros, where a missing value of the shape CAS 030 rows is a failure unless its row would pass the cap; and TAN and the inverse functions read in the mode.
- CI on every push:
make checkunder gcc and clang,make test, CMake under gcc and clang, WebAssembly, and the ESP-IDF component for the ESP32-S3. - 456 planted mutants, each a fault the checks must catch: all 456 are caught at this tag's tree (work/v1.1.2-mutants.txt), CAS 030's six among them.
- Hidden sets kept by the project's reviewer, independent of this repository, rerun at each unit's acceptance. At CAS 030's: its own, 1,040 runs in RAD, DEG and GRAD, judged by the byte shape (1.1.1's line plus exactly one row, the factor's own, within 191 B), at the mode's own zeros, and by value; and the earlier units' sets.
- On the ESP32-S3, every one of the bench's 1,379 cases is byte-identical to the host:
- with a 16,384 B arena, the size the calculator runs it with;
- worst stack 11,360 B read with interrupts masked, plus a 384 B interrupt allowance (two frames, or the most measured over the masked reading where that is larger), within a 12 KiB task;
- nesting deeper than 44 levels is refused (NO_ROOM);
- about 243 KB of Casimir's code at -O2 (243,222 B, 257 more than 1.1.1's).
Not yet, after 1.1.2
These print a line without the note it could have, print a row the answer doesn't need, have no value where the sum has one, refuse, or print a constant as written. None of them is a wrong value.
-
A quotient that canon cancels, or whose divisor falls in K, is still refused.
X^(2×K)÷X^Kis NO_CLOSED_FORM: canon reads it asX^K, which has a value at X = 0, where its terms are 0÷0. So isX^K÷Y^(-K): its divisor has no value at Y = 0, while canon's(X×Y)^Khas one. Answering them needs a row that(X÷Y)^K-style twins don't print (X≠0,Y≠0). -
The
÷spelling reaches the rows of the×one. A quotient prints the answer its canon spelling did in 1.1.0, rows for nothing and missing values included. At a letter's 0, a quotient's terms whose power of the letter is 0 or less have values (0^0 = 1), and there the letter's row excludes the sum's value.X^K÷Y^Kfrom 0 to N prints-1÷(X÷Y-1)+(X÷Y)^(N+1)÷(X÷Y-1); Y≠0; X÷Y≠1, though at Y = 0 and N = 0 the sum is 1, asX^K×Y^(-K)printed in 1.1.0.X^(K-1)÷Y^(K+2)from 1 to N printsX≠0, though at X = 0 the sum is 1÷Y^3 for every N ≥ 1. Where the factor is one of CAS 030's functions of a letter, its row excludes the point now:Y^K÷SIN(Z)^Kfrom 0 to N printsSIN(Z)≠0, though at Z = π and N = 0 the sum is 1 (rows for nothing, below). They're the class below, where the shift that clears a rational ratio's row is refused past a letter at a negative power. -
A note that doesn't fit. Where the line with the note would pass solve's 191 B, solve prints the line as 1.0.1 did, with no note, never TOO_LONG for it and with no root dropped.
(7×X-2)×(3×X^2-9×X-7)×(8×X^2+3×X-2)×(8×X-3)×(X^2-3×X+8)×(8×X^2+6×X-1)×(1×X^2+9×X+3)prints its ten real roots in 184 B asSOLVED: …, and, 2 CPLXwould make it 192 B. With--complexthe same input is TOO_LONG. In a family of degree-12 polynomials built to reach the cap, 59 of 1,416 noted answers give way at one seed and 69 of 1,403 at another; none in the golden corpus does. -
The complex roots of a factor of degree 3 or more aren't counted, since solve doesn't fi...
Casimir 1.1.1
Casimir 1.1.1: sum reads a quotient of powers in K as one ratio. X^K÷Y^K from 1 to N was NO_CLOSED_FORM in 1.1.0. It now prints -X÷(X-Y)+(X÷Y)^(N+1)÷(X÷Y-1); Y≠0; X÷Y≠1, the answer 1.1.0 already gave for X^K×Y^(-K), which is how canon reads the quotient, and for (X÷Y)^K. 1.1.1's code is CAS 029's, accepted at 33b265d, with the version text changed. src/casim.h's interface is 1.0's, and no answer 1.1.0 gave changes. Only sums it refused move: most now answer, and a constant whose sign casim can't decide gives UNDECIDED in place of NO_CLOSED_FORM.
What's new since 1.1.0
- A quotient of powers in K, read as one ratio (CAS 029). Each answer is, byte for byte, 1.1.0's answer for the summand as simplify writes it, K from 1 to N unless said:
2^K÷X^Kis2÷(X-2)+(2÷X)^(N+1)÷2÷(1÷X-1÷2); X≠2, asX^(-K)×2^Kwas;X^(K+1)÷Y^KisX×(X÷Y)^(N+1)÷(X÷Y-1)-X^2÷(X-Y); Y≠0; X÷Y≠1;K×X^K÷Y^Kfrom 0 to ∞ is-X÷2÷(X-Y÷2-X^2÷2÷Y); Y≠0; ABS(X÷Y)<1;(X^K+1)÷Y^Kfrom 0 to N is-1÷(1÷Y-1)-1÷(X÷Y-1)+(1÷Y)^(N+1)÷(1÷Y-1)+(X÷Y)^(N+1)÷(X÷Y-1); Y≠0; Y≠1; X÷Y≠1;SIN(Z)^K÷Y^KisSIN(Z)÷(Y-SIN(Z))+(SIN(Z)÷Y)^(N+1)÷(SIN(Z)÷Y-1); Y≠0; SIN(Z)÷Y≠1.
- How: a typed divisor's zeros must survive canon's reading, or the sum refuses (CAS 017). A divisor's power of b, with b free of K, now counts as kept where canon's power of b falls in K, since the ratio then holds b at a negative power: where b = 0 the answer has no value, or its row
b≠0excludes it. - A constant's power in a divisor is read too.
1÷(-2)^Kis-2×(-1÷2)^(N+1)÷3-1÷3, and1÷(-10^(-9))^Kis-(-1000000000)^(N+1)÷1000000001-1000000000÷1000000001, where 1.1.0 refused both. A constant whose sign casim can't decide gives UNDECIDED, as for any ratio (CAS 023):1÷(SIN(1)^2+COS(1)^2-1)^K, whose constant is exactly 0. - What stays refused:
X^(2×K)÷X^K,Y^K÷Y^KandX^K÷Y^(-K)are NO_CLOSED_FORM, as in 1.1.0 (Not yet, below). - What stays as it was, byte for byte: every answer 1.1.0 gave. The golden corpus moves nowhere, and the hidden sets of earlier units are 1.1.0's byte for byte. The one exception is 52 runs of CAS 022's, each a constant's power in a divisor that 1.1.0 refused (above): 49 now say UNDECIDED, 1 NO_ROOM, and 2 answer, exactly.
- Measured: on the ESP32-S3 at 16,384 B, the five new bench rows take at most 3,848 B of arena and 2,592 B of stack.
The guarantee is unchanged: an exact answer, or a status that says why not. Never an approximation presented as an answer.
The known wrong answers
None.
Found after this release, on 25 Oct 2026. Summed from 0 to ∞, (Z÷Y+1)^K prints -Y÷Z; Z≠0; ABS(Z÷Y+1)<1. Every term from K = 1 has no value at Y = 0, so neither has the sum, while the answer does: it reads 0 there, and no row excludes the point (a row with no value there doesn't exclude it). (Z÷(Y-1)+1)^K at Y = 1, (Z÷Y^2+1)^K at Y = 0, and (Z÷SIN(Y)+1)^K wherever SIN(Y) is 0 do the same to ∞. Summed to N, none of them has a value where the sum has none. The project's reviewer found it while testing a later unit. The fix is CAS 033, in the release after 1.2.0, which adds the rows.
Found after this release, on 26 Oct 2026. canon may drop a part of the input that has no value at some points, since its domain rule lets it define a result where the input has none (u÷u is 1, u-u and 0×u are 0, u^0 is 1, EXP(LN(u)) and ALOG(LOG(u)) are u, SQRT(u)^2 is u). sum, defint and limit then print a value where the input as typed has none:
- Summed from 1 to N,
(Y÷Y)^KprintsN. At Y = 0 every term is 0÷0, so the sum has no value, whileNhas one. Summed from 0 to ∞,(Z×LN(Y)÷LN(Y))^Kprints-1÷(Z-1); ABS(Z)<1, which has a value at Y = 1, where every term holds LN(1)÷LN(1), and at every Y ≤ 0, where LN(Y) has none. - Summed from 0 to ∞,
(Z×EXP(LN(Y))+1)^Kprints-1÷Y÷Z; Y≠0; Z≠0; ABS(Y×Z+1)<1. At Y = -1 and Z = 1÷2 every term holds LN(-1), which has no real value, so the sum has none, while the answer is 2 there and each row holds. With ALOG(LOG(Y)), SQRT(Y)^2 or SQRT(Y)×SQRT(Y) in place of EXP(LN(Y)) it prints the same line. defint "SQRT(X)^2" X -2 -1prints-3÷2, though SQRT(X) has no real value on the interval (SQRT(X)over it is IMPROPER).SQRT(X)×SQRT(X)and(X^(1÷4))^4print the same, and0×SQRT(X)prints0.limit "EXP(LN(X))" X -1 ''prints-1, though LN(X) has no real value near -1.- With another letter, a 0÷0 reaches limit and defint as well:
defint "X×Y÷Y" X 0 1prints1÷2andlimit "X×Y÷Y" X 1 ''prints1, each with a value at Y = 0, where the input has none. - Where the sum's closed form cancels a divisor of the summand: summed from 1 to ∞,
K×(X×Y)^K÷Yprints-X÷(X×Y-1)-X^2×Y÷2÷(X×Y-X^2×Y^2÷2-1÷2); ABS(X×Y)<1. At Y = 0 the term at K = 1 is 0÷0, so the sum has no value, while the answer is X. From 1.0.1,K×(SIN(Z)×Y)^K÷SIN(Z)does the same where SIN(Z) is 0, and(K-1)×(TAN(Z)×Y)^K÷TAN(Z)where TAN(Z) is 0 (1.0.0 printed both with no value there).
limit also reads a function outside its real domain as typed, with no canon involved: limit "SQRT(X)" X -1 '' prints SQRT(-1), and from below 0 it prints 0, though SQRT(X) has no real value near -1 or below 0.
simplify promises no more than the domain rule (docs/canonical-form.md), and expand, collect, subst, derive and integrate answer through canon under the same rule. taylor and the polynomial ops read what was typed: taylor "EXP(LN(X))" X -1 2 prints UNDEFINED, and solve, factor, cancel, partfrac and pdiv refuse these forms (NOT_POLYNOMIAL). 1.0.0 to 1.3.0 print every line above. The project's reviewer found it while testing a later unit, and a fix is being made (CAS 035).
How it's tested (at this tag)
make test: 20 programs with 118,439 checks. They read every answer back through an independent evaluator (the calculator firmware's core) and compare values.- The golden corpus, 10,822 calls compared byte for byte: CAS 029 moves none.
- Frame and symbol gates: no frame over 1 KiB, no
powlinked, and every function's frame against the recorded table. No frame moved from 1.1.0. The two new functions, sm_fall and sm_held, take 80 B and 48 B on the host, and 64 B and 32 B on the S3. - Value checks beside the corpus judge Taylor polynomials, limits, integrals and sums against independent constructions: power series over fractions, numeric derivatives at 90 digits, and sums added term by term at 60 digits. New in 1.1.1: seeded quotients (80 a run, and 14 held), each checked byte for byte against the answer for simplify's spelling, and by value against its typed terms wherever it has a value, with each letter at 0 in turn; one in eight is a shape that must stay refused.
- CI on every push:
make checkunder gcc and clang,make test, CMake under gcc and clang, WebAssembly, and the ESP-IDF component for the ESP32-S3. - 450 planted mutants, each a fault the checks must catch: all 450 are caught at this tag's tree (work/v1.1.1-mutants.txt). One, sum-positive-base-hole, survived the first run, at the version commit's tree. CAS 029's new rule keeps
1÷2^Kby its fall, the case that had caught it, so no check reached the older rule where it alone keeps a sum, a positive constant whose power rises in K (1÷2^(-K)). Three rows now reach it. The tag is the version commit plus those rows; the code is the version commit's. - Hidden sets kept by the project's reviewer, independent of this repository, rerun at each unit's acceptance. At CAS 029's: its own, 700 runs, judged against the typed sum, the ratio written as one power, and 1.1.0's answer for simplify's spelling; CAS 028's; CAS 005's 3,000 equations; CAS 020's to 027's; and CAS 014's, 015's and 018's.
- On the ESP32-S3, every one of the bench's 1,374 cases is byte-identical to the host:
- with a 16,384 B arena, the size the calculator runs it with;
- worst stack 11,360 B read with interrupts masked, plus a 384 B interrupt allowance (two frames, or the most measured over the masked reading where that is larger; another run of this code's base measured 400), within a 12 KiB task;
- nesting deeper than 44 levels is refused (NO_ROOM);
- about 243 KB of Casimir's code at -O2 (242,965 B, 433 more than 1.1.0's).
Not yet, after 1.1.1
These print a line without the note it could have, print a row the answer doesn't need, have no value where the sum has one, refuse, or print a constant as written. None of them is a wrong value.
-
A quotient that canon cancels, or whose divisor falls in K, is still refused.
X^(2×K)÷X^Kis NO_CLOSED_FORM: canon reads it asX^K, which has a value at X = 0, where its terms are 0÷0. So isX^K÷Y^(-K): its divisor has no value at Y = 0, while canon's(X×Y)^Khas one. Answering them needs a row that(X÷Y)^K-style twins don't print (X≠0,Y≠0). -
The
÷spelling reaches the rows of the×one. A quotient prints the answer its canon spelling did in 1.1.0, rows for nothing and missing values included. At a letter's 0, a quotient's terms whose power of the letter is 0 or less have values (0^0 = 1), and there the letter's row excludes the sum's value.X^K÷Y^Kfrom 0 to N prints-1÷(X÷Y-1)+(X÷Y)^(N+1)÷(X÷Y-1); Y≠0; X÷Y≠1, though at Y = 0 and N = 0 the sum is 1, asX^K×Y^(-K)printed in 1.1.0.X^(K-1)÷Y^(K+2)from 1 to N printsX≠0, though at X = 0 the sum is 1÷Y^3 for every N ≥ 1. Where the factor is a function, the answer has no value there:Y^K÷SIN(Z)^Kfrom 0 to N at Z = π and N = 0, where the sum is 1. They're the class below, where the shift that clears a rational ratio's row is refused past a letter at a negative power. -
A note that doesn't fit. Where the line with the note would pass solve's 191 B, solve prints the line as 1.0.1 did, with no note, never TOO_LONG for it and with no root dropped.
(7×X-2)×(3×X^2-9×X-7)×(8×X^2+3×X-2)×(8×X-3)×(X^2-3×X+8)×(8×X^2+6×X-1)×(1×X^2+9×X+3)prints its ten real roots in 184 B asSOLVED: …, and, 2 CPLXw...
Casimir 1.1.0
Casimir 1.1.0: solve says when it leaves complex roots out. With the complex bit off, and always in model 33, a SOLVED or PARTIAL answer now counts the roots the bit would print, after its status: (X-2)×(X^2+1) is SOLVED, 2 CPLX: X=2, where 1.0.1 printed SOLVED: X=2. 1.1.0's code is CAS 028's, accepted at 8a9b010, with the version text changed. src/casim.h's interface is 1.0's, but those answers' bytes change, which is why this is 1.1.0 and not 1.0.2: a caller that compares solve's status to SOLVED exactly should read its first six bytes, or split the line at its first : .
What's new since 1.0.1
- Roots left out (CAS 028). Without CASIM_COMPLEX (
--complex), and in model 33, which reads no complex number, SOLVED's and PARTIAL's status gains, n CPLX. n is the roots the bit would print, counted with their multiplicities.(X^2+1)^2×(X-3)isSOLVED, 4 CPLX: X=3.X^6-1isSOLVED, 4 CPLX: X=-1; X=1.(X^2+1)×(X^3-2×X-5)isPARTIAL, DEGREE 3 LEFT, 2 CPLX.- In model 33,
(X-2)×(X^2+1)isSOLVED, 2 CPLX: X=2, with the bit or without it. - Show steps ends on solve's line:
X^4+X^2=0ends onANSWER: SOLVED, 2 CPLX: X=0 (MULT 2). - The note goes before the line's first
:, so a reader that splits there, as the calculator's report does, keeps the roots as they were. The widest status isPARTIAL, DEGREE 10 LEFT, 2 CPLX, 31 characters ((X^2+1)×(X^10+X+1)): the degree left and n together are at most 12.
- n counts only the pairs factoring finds. solve finds the linear and quadratic factors over the rationals and leaves a factor of degree 3 or more whole. That factor adds nothing to n:
X^4+1is stillPARTIAL, DEGREE 4 LEFT, as the bit prints none of its roots either. - What stays as it was, byte for byte:
- with the bit, outside model 33:
(X-2)×(X^2+1)with--complexisSOLVED: X=2; X=-0i1; X=0i1, as at 1.0.1; - NO REAL ROOTS, which says it already:
X^4+4isNO REAL ROOTS; - every other operation, and every solve answer with no root left out.
- with the bit, outside model 33:
- At solve's 191 B, the note gives way. A line that fits only without the note is printed as at 1.0.1 (Not yet, below).
- Measured: 84 of the golden corpus's lines gain the note, in the corpus and again in its decimal twin (44 in model 35, 17 in STU, 23 in model 33), and nothing else moves. On the ESP32-S3 at 16,384 B, the five new bench rows take at most 4,360 B of arena and 1,936 B of stack.
The guarantee is unchanged: an exact answer, or a status that says why not. Never an approximation presented as an answer.
The known wrong answers
None.
Found after this release, on 25 Oct 2026. Summed from 0 to ∞, (Z÷Y+1)^K prints -Y÷Z; Z≠0; ABS(Z÷Y+1)<1. Every term from K = 1 has no value at Y = 0, so neither has the sum, while the answer does: it reads 0 there, and no row excludes the point (a row with no value there doesn't exclude it). (Z÷(Y-1)+1)^K at Y = 1, (Z÷Y^2+1)^K at Y = 0, and (Z÷SIN(Y)+1)^K wherever SIN(Y) is 0 do the same to ∞. Summed to N, none of them has a value where the sum has none. The project's reviewer found it while testing a later unit. The fix is CAS 033, in the release after 1.2.0, which adds the rows.
Found after this release, on 26 Oct 2026. canon may drop a part of the input that has no value at some points, since its domain rule lets it define a result where the input has none (u÷u is 1, u-u and 0×u are 0, u^0 is 1, EXP(LN(u)) and ALOG(LOG(u)) are u, SQRT(u)^2 is u). sum, defint and limit then print a value where the input as typed has none:
- Summed from 1 to N,
(Y÷Y)^KprintsN. At Y = 0 every term is 0÷0, so the sum has no value, whileNhas one. Summed from 0 to ∞,(Z×LN(Y)÷LN(Y))^Kprints-1÷(Z-1); ABS(Z)<1, which has a value at Y = 1, where every term holds LN(1)÷LN(1), and at every Y ≤ 0, where LN(Y) has none. - Summed from 0 to ∞,
(Z×EXP(LN(Y))+1)^Kprints-1÷Y÷Z; Y≠0; Z≠0; ABS(Y×Z+1)<1. At Y = -1 and Z = 1÷2 every term holds LN(-1), which has no real value, so the sum has none, while the answer is 2 there and each row holds. With ALOG(LOG(Y)), SQRT(Y)^2 or SQRT(Y)×SQRT(Y) in place of EXP(LN(Y)) it prints the same line. defint "SQRT(X)^2" X -2 -1prints-3÷2, though SQRT(X) has no real value on the interval (SQRT(X)over it is IMPROPER).SQRT(X)×SQRT(X)and(X^(1÷4))^4print the same, and0×SQRT(X)prints0.limit "EXP(LN(X))" X -1 ''prints-1, though LN(X) has no real value near -1.- With another letter, a 0÷0 reaches limit and defint as well:
defint "X×Y÷Y" X 0 1prints1÷2andlimit "X×Y÷Y" X 1 ''prints1, each with a value at Y = 0, where the input has none. - Where the sum's closed form cancels a divisor of the summand: summed from 1 to ∞,
K×(X×Y)^K÷Yprints-X÷(X×Y-1)-X^2×Y÷2÷(X×Y-X^2×Y^2÷2-1÷2); ABS(X×Y)<1. At Y = 0 the term at K = 1 is 0÷0, so the sum has no value, while the answer is X. From 1.0.1,K×(SIN(Z)×Y)^K÷SIN(Z)does the same where SIN(Z) is 0, and(K-1)×(TAN(Z)×Y)^K÷TAN(Z)where TAN(Z) is 0 (1.0.0 printed both with no value there).
limit also reads a function outside its real domain as typed, with no canon involved: limit "SQRT(X)" X -1 '' prints SQRT(-1), and from below 0 it prints 0, though SQRT(X) has no real value near -1 or below 0.
simplify promises no more than the domain rule (docs/canonical-form.md), and expand, collect, subst, derive and integrate answer through canon under the same rule. taylor and the polynomial ops read what was typed: taylor "EXP(LN(X))" X -1 2 prints UNDEFINED, and solve, factor, cancel, partfrac and pdiv refuse these forms (NOT_POLYNOMIAL). 1.0.0 to 1.3.0 print every line above. The project's reviewer found it while testing a later unit, and a fix is being made (CAS 035).
How it's tested (at this tag)
make test: 20 programs with 118,439 checks. They read every answer back through an independent evaluator (the calculator firmware's core) and compare values.- The golden corpus, 10,822 calls compared byte for byte. The 84 lines CAS 028 moves are listed with their before (tests/golden/moved/cas-028.txt), and a check puts each before back and gets 1.0.1's files to the byte.
- Frame and symbol gates: no frame over 1 KiB, no
powlinked, and every function's frame against the recorded table. No frame moved from 1.0.1; the one new function, put_left_out, takes 80 B on the host and 48 B on the S3. - Value checks beside the corpus judge Taylor polynomials, limits, integrals and sums against independent constructions: power series over fractions, numeric derivatives at 90 digits, and sums added term by term at 60 digits. New in 1.1.0: solve's n is judged on 2,199 seeded polynomials built from known factors, against their construction, and on 2,190 of them against the same input's answer with the bit, which must add exactly n roots.
- CI on every push:
make checkunder gcc and clang,make test, CMake under gcc and clang, WebAssembly, and the ESP-IDF component for the ESP32-S3. - 445 planted mutants, each caught by the tests. All 445 were run at this tag's tree and caught, 445 of 445 (work/v1.1.0-mutants.txt): 1.0.1's 435 and the ten new on CAS 028's code.
- Hidden sets kept by the project's reviewer, independent of this repository, rerun at each unit's acceptance. At CAS 028's: its own; CAS 005's 3,000 equations, solve and steps, with the bit off; CAS 020's; CAS 021's to 027's; and CAS 014's, 015's and 018's.
- On the ESP32-S3, every one of the bench's 1,369 cases is byte-identical to the host:
- with a 16,384 B arena, the size the calculator runs it with;
- worst stack 11,360 B read with interrupts masked, plus a 384 B interrupt allowance (two frames, or the most measured over the masked reading where that is larger; another run of this code measured 400), within a 12 KiB task;
- nesting deeper than 44 levels is refused (NO_ROOM);
- about 243 KB of Casimir's code at -O2 (242,532 B, 679 more than 1.0.1's).
Not yet, after 1.1.0
These print a line without the note it could have, print a row the answer doesn't need, have no value where the sum has one, refuse, or print a constant as written. None of them is a wrong value.
-
A note that doesn't fit. Where the line with the note would pass solve's 191 B, solve prints the line as 1.0.1 did, with no note, never TOO_LONG for it and with no root dropped.
(7×X-2)×(3×X^2-9×X-7)×(8×X^2+3×X-2)×(8×X-3)×(X^2-3×X+8)×(8×X^2+6×X-1)×(1×X^2+9×X+3)prints its ten real roots in 184 B asSOLVED: …, and, 2 CPLXwould make it 192 B. With--complexthe same input is TOO_LONG. In a family of degree-12 polynomials built to reach the cap, 59 of 1,416 noted answers give way at one seed and 69 of 1,403 at another; none in the golden corpus does. -
The complex roots of a factor of degree 3 or more aren't counted, since solve doesn't find them, with the bit or without it:
X^4+1isPARTIAL, DEGREE 4 LEFT. -
Rows for nothing. Each is excluded where the sum has a value, while the answer has none there:
- a sum to ∞ from a negative bound:
(K+1)×X^(K+1)×Y^Kfrom -1 to ∞ printsX^2×Y÷(X×Y-2×X^2×Y^2+X^3×Y^3); X≠0; Y≠0; ABS(X×Y)<1, though the sum is 0 at X = 0; (X÷Y)^K÷Xfrom 1 to N printsX≠0, though the sum has a value at X = 0;- past an irrational constant, a factor that is never 0 or a letter at a negative power, the shift that clears a rational ratio's row is refused.
(K-1)×(SQRT(2)×X×Y)^K÷Xfrom 1 to ∞ keepsX≠0, while its rational twin(K-1)×(X×Y)^K÷Xdoesn't, andK×(EXP(Z)×X)^K÷Xfrom 1 to ∞ keepsX≠0. (Past a letter in a function the shift is made now, above.)
- a sum to ∞ from a negative bound:
-
A redundant row. The letter's row can be printed beside a row that already excludes the letter's 0.
K×COS(Z)^(K-1)from 1 to N printsZ≠0besideCOS(Z)≠1, and COS(0) is 1. It excludes nothing the other row doesn't. -
A missing value. Each has no value at a point where the sum has one, and no row excludes that point:
- at the empty sum, at a letter's 0...
Casimir 1.0.1
Casimir 1.0.1: a fast follow to 1.0.0. In most of the cases where 1.0.0's closed form for a sum had no value at a zero of a function factor of its ratio, 1.0.1's has the sum's value there. 1.0.0 printed no wrong value in those cases: its answers had no value at those points, and its notes named them. 1.0.1's code is CAS 027's, accepted at cfdb18f, with the version text changed. src/casim.h's interface is 1.0.0's.
What's new since 1.0.0
-
The shift past a function factor (CAS 027). A sum q(K)×B^K whose B holds a function of a letter, SIN(Z), COS(Z), TAN(Z), LN(Z), SQRT(Z), ABS(Z) or Z^(1÷3), is shifted past it as past a letter. Where the shift clears the factor, the answer no longer divides by it.
K×(SIN(Z)×Y)^K÷SIN(Z)from 1 to ∞ is-Y÷(Y×SIN(Z)-1)-Y^2×SIN(Z)÷2÷(Y×SIN(Z)-Y^2×SIN(Z)^2÷2-1÷2); ABS(Y×SIN(Z))<1. That is Y at Z = π, as the sum is. 1.0.0 printed a form that divides by SIN(Z), with no value at π, andZ≠0.- Summed to N it is the same:
SIN(Z)^(2×K-1)×Y^Kfrom 1 to N isY×SIN(Z)×(Y×SIN(Z)^2)^N÷(Y×SIN(Z)^2-1)-Y×SIN(Z)÷(Y×SIN(Z)^2-1); Y×SIN(Z)^2≠1, which is 0 at Z = π for every N, as the sum is. - At the letter's own 0, from a negative lo:
(SQRT(Z)×Y)^K×SQRT(Z)from -1 to N is1÷(Y-Y^2×SQRT(Z))-(Y×SQRT(Z))^(N+2)÷(Y-Y^2×SQRT(Z)); Y≠0; Y×SQRT(Z)≠1, which is 0 at Z = 0 and N = -2, the empty sum. 1.0.0's form had no value there.
-
What comes with it: rows for nothing dropped, and sums that were TOO_LONG answered.
K×(SIN(Z)×Y)^K÷Yfrom 1 to ∞ is-Y×SIN(Z)^2÷2÷(Y×SIN(Z)-Y^2×SIN(Z)^2÷2-1÷2)-SIN(Z)÷(Y×SIN(Z)-1); ABS(Y×SIN(Z))<1, with noY≠0: the sum is SIN(Z) at Y = 0, and so is the answer.K×SIN(Z)^(K-1)from 1 to N was TOO_LONG. It's now-1÷(SIN(Z)-1)+N×SIN(Z)^N÷(SIN(Z)-1)-SIN(Z)÷2÷(SIN(Z)-SIN(Z)^2÷2-1÷2)+SIN(Z)^N÷(SIN(Z)-1)+SIN(Z)^(N+1)÷2÷(SIN(Z)-SIN(Z)^2÷2-1÷2); SIN(Z)≠1.
-
What stays as it was, byte for byte:
- Where the shift wouldn't clear the factor, the answer is 1.0.0's. From 0, the first term of
K×SIN(Z)^(K-1)is 0×SIN(Z)^(-1), with no value where SIN(Z) is 0, soK×SIN(Z)^(K-1)from 0 to ∞ is still-1÷2÷(SIN(Z)-SIN(Z)^2÷2-1÷2); SIN(Z)≠0; ABS(SIN(Z))<1. - A factor that is never 0 (EXP, COSH, 2^Z) and a constant (SQRT(2)) still stop the shift.
K×(EXP(Z)×X)^K÷Xfrom 1 to ∞ keepsX≠0(below).
- Where the shift wouldn't clear the factor, the answer is 1.0.0's. From 0, the first term of
-
Measured over CAS 026's two probe sets, 640 sums: seven factors in eight shapes from -1 to 2, and eight factors at powers 2 and 3 from 0 and 1, each to N and to ∞. The sums with a missing value at a factor's zero, 65 at 1.0.0, are 24. 105 sums print something new:
- 41 gain a value;
- 42 drop a row for nothing;
- 14 that were TOO_LONG answer;
- 8 change form only.
At 9,758 points, at the factors' zeros and off them, each answer's value is the sum's.
The guarantee is unchanged: an exact answer, or a status that says why not. Never an approximation presented as an answer.
The known wrong answers
None.
Found after this release, on 25 Oct 2026. Summed from 0 to ∞, (Z÷Y+1)^K prints -Y÷Z; Z≠0; ABS(Z÷Y+1)<1. Every term from K = 1 has no value at Y = 0, so neither has the sum, while the answer does: it reads 0 there, and no row excludes the point (a row with no value there doesn't exclude it). (Z÷(Y-1)+1)^K at Y = 1, (Z÷Y^2+1)^K at Y = 0, and (Z÷SIN(Y)+1)^K wherever SIN(Y) is 0 do the same to ∞. Summed to N, none of them has a value where the sum has none. The project's reviewer found it while testing a later unit. The fix is CAS 033, in the release after 1.2.0, which adds the rows.
Found after this release, on 26 Oct 2026. canon may drop a part of the input that has no value at some points, since its domain rule lets it define a result where the input has none (u÷u is 1, u-u and 0×u are 0, u^0 is 1, EXP(LN(u)) and ALOG(LOG(u)) are u, SQRT(u)^2 is u). sum, defint and limit then print a value where the input as typed has none:
- Summed from 1 to N,
(Y÷Y)^KprintsN. At Y = 0 every term is 0÷0, so the sum has no value, whileNhas one. Summed from 0 to ∞,(Z×LN(Y)÷LN(Y))^Kprints-1÷(Z-1); ABS(Z)<1, which has a value at Y = 1, where every term holds LN(1)÷LN(1), and at every Y ≤ 0, where LN(Y) has none. - Summed from 0 to ∞,
(Z×EXP(LN(Y))+1)^Kprints-1÷Y÷Z; Y≠0; Z≠0; ABS(Y×Z+1)<1. At Y = -1 and Z = 1÷2 every term holds LN(-1), which has no real value, so the sum has none, while the answer is 2 there and each row holds. With ALOG(LOG(Y)), SQRT(Y)^2 or SQRT(Y)×SQRT(Y) in place of EXP(LN(Y)) it prints the same line. defint "SQRT(X)^2" X -2 -1prints-3÷2, though SQRT(X) has no real value on the interval (SQRT(X)over it is IMPROPER).SQRT(X)×SQRT(X)and(X^(1÷4))^4print the same, and0×SQRT(X)prints0.limit "EXP(LN(X))" X -1 ''prints-1, though LN(X) has no real value near -1.- With another letter, a 0÷0 reaches limit and defint as well:
defint "X×Y÷Y" X 0 1prints1÷2andlimit "X×Y÷Y" X 1 ''prints1, each with a value at Y = 0, where the input has none. - Where the sum's closed form cancels a divisor of the summand: summed from 1 to ∞,
K×(X×Y)^K÷Yprints-X÷(X×Y-1)-X^2×Y÷2÷(X×Y-X^2×Y^2÷2-1÷2); ABS(X×Y)<1. At Y = 0 the term at K = 1 is 0÷0, so the sum has no value, while the answer is X. From 1.0.1,K×(SIN(Z)×Y)^K÷SIN(Z)does the same where SIN(Z) is 0, and(K-1)×(TAN(Z)×Y)^K÷TAN(Z)where TAN(Z) is 0 (1.0.0 printed both with no value there).
limit also reads a function outside its real domain as typed, with no canon involved: limit "SQRT(X)" X -1 '' prints SQRT(-1), and from below 0 it prints 0, though SQRT(X) has no real value near -1 or below 0.
simplify promises no more than the domain rule (docs/canonical-form.md), and expand, collect, subst, derive and integrate answer through canon under the same rule. taylor and the polynomial ops read what was typed: taylor "EXP(LN(X))" X -1 2 prints UNDEFINED, and solve, factor, cancel, partfrac and pdiv refuse these forms (NOT_POLYNOMIAL). 1.0.0 to 1.3.0 print every line above. The project's reviewer found it while testing a later unit, and a fix is being made (CAS 035).
How it's tested (at this tag)
make test: 20 programs with 118,421 checks. They read every answer back through an independent evaluator (the calculator firmware's core) and compare values.- The golden corpus, 10,822 calls compared byte for byte.
- Frame and symbol gates: no frame over 1 KiB, no
powlinked, and every function's frame against the recorded table. No frame moved from 1.0.0; the one new function, sm_atom, takes 144 B on the host and 96 B on the S3. - Value checks beside the corpus judge Taylor polynomials, limits, integrals and sums against independent constructions: power series over fractions, numeric derivatives at 90 digits, and sums added term by term at 60 digits. A sum's rows are judged at each letter's 0 and at each factor's own zeros (SIN at 0 and π, COS at π÷2, LN at 1), with the factor put to 0 exactly. New in 1.0.1: at each such zero where the sum has a value, the answer must have one, and it must be the sum's.
- CI on every push:
make checkunder gcc and clang,make test, CMake under gcc and clang, WebAssembly, and the ESP-IDF component for the ESP32-S3. - 435 planted mutants, each caught by the tests. All 435 were run at this tag's tree and caught, 435 of 435 (work/v1.0.1-mutants.txt): rc7's 428 and the seven new on CAS 027's code.
- Hidden sets kept by the project's reviewer, independent of this repository, rerun at each unit's acceptance. At CAS 027's: its own, and CAS 014's to 026's.
- On the ESP32-S3, every one of the bench's 1,364 cases is byte-identical to the host:
- with a 16,384 B arena, the size the calculator runs it with;
- worst stack 11,360 B read with interrupts masked, plus a 400 B interrupt allowance (the most measured over the masked reading), within a 12 KiB task;
- nesting deeper than 44 levels is refused (NO_ROOM);
- about 242 KB of Casimir's code at -O2 (241,853 B, 772 more than 1.0.0's).
Not yet, after 1.0.1
These print a row the answer doesn't need, have no value where the sum has one, refuse, or print a constant as written. None of them is a wrong value.
- Rows for nothing. Each is excluded where the sum has a value, while the answer has none there:
- a sum to ∞ from a negative bound:
(K+1)×X^(K+1)×Y^Kfrom -1 to ∞ printsX^2×Y÷(X×Y-2×X^2×Y^2+X^3×Y^3); X≠0; Y≠0; ABS(X×Y)<1, though the sum is 0 at X = 0; (X÷Y)^K÷Xfrom 1 to N printsX≠0, though the sum has a value at X = 0;- past an irrational constant, a factor that is never 0 or a letter at a negative power, the shift that clears a rational ratio's row is refused.
(K-1)×(SQRT(2)×X×Y)^K÷Xfrom 1 to ∞ keepsX≠0, while its rational twin(K-1)×(X×Y)^K÷Xdoesn't, andK×(EXP(Z)×X)^K÷Xfrom 1 to ∞ keepsX≠0. (Past a letter in a function the shift is made now, above.)
- a sum to ∞ from a negative bound:
- A redundant row. The letter's row can be printed beside a row that already excludes the letter's 0.
K×COS(Z)^(K-1)from 1 to N printsZ≠0besideCOS(Z)≠1, and COS(0) is 1. It excludes nothing the other row doesn't. - A missing value. Each has no value at a point where the sum has one, and no row excludes that point:
- summed to N from a negative lo, at a letter's 0 for N ≤ −2.
X^(K+2)×Y^(-2×K)from -2 to N printsX^2×(X÷Y^2)^(N+1)÷(X÷Y^2-1)-Y^4÷(X÷Y^2-1); Y≠0; X÷Y^2≠1. At X = 0 and N = -2 the sum is Y^4, while X^2×(X÷Y^2)^(-1) has no value; - at a zero of a function factor other than its letter's 0 (SIN and TAN at π, COS at π÷2, LN at 1), now only where the same sum with the factor read as a letter, its rational twin, prints that letter's row for nothing. Of the 640 probe sums, 1.0.0 had this at 62 and 1.0.1 has it at 24 (the first shape below); the other two shapes were found beyond those sets:
- to N, at an N before the first term that divides by the factor, where the sum has a value (the empty sum, or a sum of terms that don't divide) and the terms after it divide.
(Y÷SIN(Z))^Kfro...
- to N, at an N before the first term that divides by the factor, where the sum has a value (the empty sum, or a sum of terms that don't divide) and the terms after it divide.
- summed to N from a negative lo, at a letter's 0 for N ≤ −2.