From b8886a8d490d72fba1ea3c705a1afa952851b1be Mon Sep 17 00:00:00 2001 From: zhangbo Date: Sat, 24 Feb 2024 21:45:24 +0800 Subject: [PATCH] =?UTF-8?q?=E5=A4=A7=E6=95=B4=E6=95=B0=E4=B8=8D=E7=94=A8GM?= =?UTF-8?q?P=E8=BF=90=E7=AE=97=EF=BC=8C=E7=9B=B4=E6=8E=A5=E7=94=A8uint64?= =?UTF-8?q?=5Ft=EF=BC=8C=E5=9B=A0=E4=B8=BA=E5=B7=B2=E7=BB=8F=E7=9F=A5?= =?UTF-8?q?=E9=81=93=E4=B8=8D=E4=BC=9A=E8=B6=85=E8=BF=8764-bit=E4=BA=86?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- README.md | 4 +- ReciprocalSearch3.cpp | 160 ++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 163 insertions(+), 1 deletion(-) create mode 100644 ReciprocalSearch3.cpp diff --git a/README.md b/README.md index 30a4a01..2b6a8a5 100644 --- a/README.md +++ b/README.md @@ -10,4 +10,6 @@ 整个代码也可以不用double来运算，而是用GMP的自然数库来运算。 -用GMP算出来的是69014个解，详见txt文件。5800 X3D CPU上算了1507秒。用的是ReciprocalSearch2.cpp这个。 +ReciprocalSearch2.cpp这个用GMP算出来的是69014个解，详见txt文件。5800 X3D CPU上算了1241秒。 + +ReciprocalSearch3.cpp这个改用uint64_t进行大整数运算，也可以在30秒以内完成。 diff --git a/ReciprocalSearch3.cpp b/ReciprocalSearch3.cpp new file mode 100644 index 0000000..89d2bf2 --- /dev/null +++ b/ReciprocalSearch3.cpp @@ -0,0 +1,160 @@ +#include +#include + +using bigint = uint64_t; + +__forceinline bigint ceil(bigint d, bigint n) +{ + bigint q = n / d; + if (q >= 100) + return 100; + + if (n % d) + ++q; + + return q; +} + +__forceinline void subdiv(bigint n1, bigint d1, bigint n2, bigint d2, int k, bigint &n3, bigint &d3, + bigint &n4, bigint &d4) +{ + d3 = d1 * d2; + n3 = n1 * d2 - n2 * d1; + + n4 = n3; + d4 = d3 * k; +} + +int main() +{ + const auto start = std::chrono::steady_clock::now(); + + int total = 0; + int valid = 0; + + for (bigint i1 = 2; i1 <= 10; i1++) + { + bigint n20, d20, n2, d2; + subdiv(1, 1, 1, i1, 9, n20, d20, n2, d2); + const bigint u2 = (std::min)((bigint)92, ceil(n2, d2)); + for (bigint i2 = i1 + 1; i2 <= u2; i2++) + { + const bigint s2 = i1 * i2; + if (s2 <= (i2 + i1)) + continue; + + bigint n30, d30, n3, d3; + subdiv(n20, d20, 1, i2, 8, n30, d30, n3, d3); + const bigint u3 = (std::min)((bigint)93, ceil(n3, d3)); + for (bigint i3 = i2 + 1; i3 <= u3; i3++) + { + const bigint s3 = s2 * i3; + if (s3 <= (i2 * i3 + i1 * i3 + s2)) + continue; + + bigint n40, d40, n4, d4; + subdiv(n30, d30, 1, i3, 7, n40, d40, n4, d4); + const bigint u4 = (std::min)((bigint)94, ceil(n4, d4)); + for (bigint i4 = i3 + 1; i4 <= u4; i4++) + { + const bigint s4 = s3 * i4; + if (s4 <= (i2 * i3 * i4 + i1 * i3 * i4 + i1 * i2 * i4 + s3)) + continue; + + bigint n50, d50, n5, d5; + subdiv(n40, d40, 1, i4, 6, n50, d50, n5, d5); + const bigint u5 = (std::min)((bigint)95, ceil(n5, d5)); + for (bigint i5 = i4 + 1; i5 <= u5; i5++) + { + const bigint s5 = s4 * i5; + if (s5 <= (i2 * i3 * i4 * i5 + i1 * i3 * i4 * i5 + i1 * i2 * i4 * i5 + + i1 * i2 * i3 * i5 + s4)) + continue; + + bigint n60, d60, n6, d6; + subdiv(n50, d50, 1, i5, 5, n60, d60, n6, d6); + const bigint u6 = (std::min)((bigint)96, ceil(n6, d6)); + for (bigint i6 = i5 + 1; i6 <= u6; i6++) + { + const bigint s6 = s5 * i6; + if (s6 <= + ((i2 * i3 * i4 * i5 * i6 + i1 * i3 * i4 * i5 * i6 + i1 * i2 * i4 * i5 * i6 + + i1 * i2 * i3 * i5 * i6 + i1 * i2 * i3 * i4 * i6 + s5))) + continue; + + bigint n70, d70, n7, d7; + subdiv(n60, d60, 1, i6, 4, n70, d70, n7, d7); + const bigint u7 = (std::min)((bigint)97, ceil(n7, d7)); + for (bigint i7 = i6 + 1; i7 <= u7; i7++) + { + const bigint s7 = s6 * i7; + if (s7 <= ((i2 * i3 * i4 * i5 * i6 * i7 + i1 * i3 * i4 * i5 * i6 * i7 + + i1 * i2 * i4 * i5 * i6 * i7 + i1 * i2 * i3 * i5 * i6 * i7 + + i1 * i2 * i3 * i4 * i6 * i7 + i1 * i2 * i3 * i4 * i5 * i7 + s6))) + continue; + + bigint n80, d80, n8, d8; + subdiv(n70, d70, 1, i7, 3, n80, d80, n8, d8); + const bigint u8 = (std::min)((bigint)98, ceil(n8, d8)); + for (bigint i8 = i7 + 1; i8 <= u8; i8++) + { + const bigint s8 = s7 * i8; + if (s8 <= + (i2 * i3 * i4 * i5 * i6 * i7 * i8 + i1 * i3 * i4 * i5 * i6 * i7 * i8 + + i1 * i2 * i4 * i5 * i6 * i7 * i8 + i1 * i2 * i3 * i5 * i6 * i7 * i8 + + i1 * i2 * i3 * i4 * i6 * i7 * i8 + i1 * i2 * i3 * i4 * i5 * i7 * i8 + + i1 * i2 * i3 * i4 * i5 * i6 * i8 + s7)) + continue; + + bigint n90, d90, n9, d9; + subdiv(n80, d80, 1, i8, 2, n90, d90, n9, d9); + const bigint u9 = (std::min)((bigint)99, ceil(n9, d9)); + for (bigint i9 = i8 + 1; i9 <= u9; i9++) + { + const bigint s9 = s8 * i9; + if (s9 <= (i2 * i3 * i4 * i5 * i6 * i7 * i8 * i9 + + i1 * i3 * i4 * i5 * i6 * i7 * i8 * i9 + + i1 * i2 * i4 * i5 * i6 * i7 * i8 * i9 + + i1 * i2 * i3 * i5 * i6 * i7 * i8 * i9 + + i1 * i2 * i3 * i4 * i6 * i7 * i8 * i9 + + i1 * i2 * i3 * i4 * i5 * i7 * i8 * i9 + + i1 * i2 * i3 * i4 * i5 * i6 * i8 * i9 + + i1 * i2 * i3 * i4 * i5 * i6 * i7 * i9 + s8)) + continue; + + bigint n100, d100, n10, d10; + subdiv(n90, d90, 1, i9, 1, n100, d100, n10, d10); + const bigint u10 = (std::min)((bigint)100, ceil(n10, d10)); + for (bigint i10 = i9 + 1; i10 <= u10; i10++) + { + const bigint s10 = s9 * i10; + if (s10 == (i2 * i3 * i4 * i5 * i6 * i7 * i8 * i9 * i10 + + i1 * i3 * i4 * i5 * i6 * i7 * i8 * i9 * i10 + + i1 * i2 * i4 * i5 * i6 * i7 * i8 * i9 * i10 + + i1 * i2 * i3 * i5 * i6 * i7 * i8 * i9 * i10 + + i1 * i2 * i3 * i4 * i6 * i7 * i8 * i9 * i10 + + i1 * i2 * i3 * i4 * i5 * i7 * i8 * i9 * i10 + + i1 * i2 * i3 * i4 * i5 * i6 * i8 * i9 * i10 + + i1 * i2 * i3 * i4 * i5 * i6 * i7 * i9 * i10 + + i1 * i2 * i3 * i4 * i5 * i6 * i7 * i8 * i10 + s9)) + { + ++valid; + + printf("%5d: %lld %lld %lld %lld %lld %lld %lld %lld %lld %lld\n", valid, i1, i2, + i3, i4, i5, i6, i7, i8, i9, i10); + } + } + } + } + } + } + } + } + } + } + } + + const auto end = std::chrono::steady_clock::now(); + printf("%.3f seconds\n", + std::chrono::duration_cast(end - start).count() / 1000.0); +} \ No newline at end of file