Problem 086

Least M such that solutions for shortest cuboid routes ≤ M exceed one million.

Answer1818
Output1818
StatusPASS
Native helperno
Runtime70 ms
Peak memory1104 KB
Time complexityO(n) (estimated)
Space complexityO(1) (estimated)

Performance comparison

MetricOur solutionBest known
Time complexityO(n)O(n)
Space complexityO(1)O(n)
ApproachFlow solutionEnumerate and count
VerdictOptimal

Flow source

# Project Euler 086
# Least M such that solutions for shortest cuboid routes ≤ M exceed one million.

function is_square(n: i64) -> bool {
    if n <= 0 { return false }
    let mut r: i64 = 1
    # integer sqrt via Newton
    let mut x: i64 = n
    while x > 1 {
        let y: i64 = (x + n / x) / 2
        if y >= x { break }
        x = y
    }
    r = x
    while r * r > n {
        r = r - 1
    }
    while (r + 1) * (r + 1) <= n {
        r = r + 1
    }
    return r * r == n
}

function count_for_m(m: i64) -> i64 {
    # for cuboid a≤b≤c=m, shortest is min path on faces:
    # sqrt((a+b)^2 + c^2) is shortest when a,b ≤ c (the usual case for a≤b≤c)
    # number of a,b with 1≤a≤b≤m and (a+b)^2+m^2 square
    let mut cnt: i64 = 0
    let mut s: i64 = 2
    while s <= 2 * m {
        if is_square(s * s + m * m) {
            # a+b = s, 1≤a≤b≤m, a+b=s ⇒ max(1,s-m) ≤ a ≤ min(s/2, s-1) and b=s-a≤m
            let mut lo: i64 = s - m
            if lo < 1 { lo = 1 }
            let mut hi: i64 = s / 2
            if hi > m { hi = m }
            # also b = s-a ≥ a and b ≤ m already in bounds
            if hi >= lo {
                cnt = cnt + (hi - lo + 1)
            }
        }
        s = s + 1
    }
    return cnt
}

function main() -> i32 {
    let mut total: i64 = 0
    let mut m: i64 = 0
    while total <= 1000000 {
        m = m + 1
        total = total + count_for_m(m)
    }
    printf("%lld\n", m)
    return 0
}

Generated C

#include <stdint.h>
#include <stdbool.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>

/* Flow runtime helpers */
typedef struct flow_temp_node { struct flow_temp_node* next; } flow_temp_node;
static flow_temp_node* flow_temp_head = NULL;
static int flow_temp_atexit_set = 0;
__attribute__((unused)) static void flow_temp_free_all(void) {
    while (flow_temp_head) {
        flow_temp_node* n = flow_temp_head;
        flow_temp_head = n->next;
        free(n);
    }
}
__attribute__((unused)) static void* flow_temp_alloc(size_t nbytes) {
    flow_temp_node* node = (flow_temp_node*)malloc(sizeof(flow_temp_node) + nbytes);
    if (!node) return NULL;
    node->next = flow_temp_head;
    flow_temp_head = node;
    if (!flow_temp_atexit_set) {
        flow_temp_atexit_set = 1;
        atexit(flow_temp_free_all);
    }
    return (void*)(node + 1);
}
#ifndef FLOW_DIAG
#define FLOW_DIAG(msg) fprintf(stderr, "%s", (msg))
#endif
#ifndef FLOW_LOG
#define FLOW_LOG(fmt, ...) printf(fmt, __VA_ARGS__)
#endif
#ifndef FLOW_LOG_EMPTY
#define FLOW_LOG_EMPTY(fmt) printf(fmt)
#endif
static char* flow_strcat(const char* a, const char* b) {
    size_t la = strlen(a ? a : ""), lb = strlen(b ? b : "");
    char* r = (char*)flow_temp_alloc(la + lb + 1);
    if (!r) return NULL;
    if (la) memcpy(r, a, la);
    if (lb) memcpy(r + la, b, lb);
    r[la + lb] = '\0';
    return r;
}

#define __flow_in_arr(arr, val) __extension__ ({ \
    int _found = 0; \
    size_t _n = sizeof(arr)/sizeof((arr)[0]); \
    for (size_t _i = 0; _i < _n; _i++) { \
        if ((arr)[_i] == (val)) { _found = 1; break; } \
    } _found; })

/* Unified fault handler (MISRA #279) — override with -DFLOW_FAULT_HANDLER=fn */
#ifndef FLOW_FAULT_HANDLER
__attribute__((unused)) static inline void flow_fault_handler(const char* msg) {
    fprintf(stderr, "flow: %s\n", msg ? msg : "fault");
    abort();
#if defined(__GNUC__) || defined(__clang__)
    __builtin_unreachable();
#endif
}
#else
#define flow_fault_handler FLOW_FAULT_HANDLER
#endif
#define flow_div_by_zero_handler() flow_fault_handler("division by zero")
#define flow_shift_ub_handler() flow_fault_handler("invalid shift (amount out of range or left-shift of negative)")

#ifndef FLOW_CHECKED_DIV
#define FLOW_CHECKED_DIV(L, R) (((R) != 0) ? ((L) / (R)) : (flow_div_by_zero_handler(), (L) * 0))
#endif
#ifndef FLOW_CHECKED_MOD
#define FLOW_CHECKED_MOD(L, R) (((R) != 0) ? ((L) % (R)) : (flow_div_by_zero_handler(), (L) * 0))
#endif
#ifndef FLOW_CHECKED_SHL
#define FLOW_CHECKED_SHL(L, R) ((((R) >= 0) && ((unsigned long long)(R) < (sizeof(L) * 8ull)) && ((L) >= 0)) ? ((L) << (R)) : (flow_shift_ub_handler(), (L) * 0))
#endif
#ifndef FLOW_CHECKED_SHR
#define FLOW_CHECKED_SHR(L, R) ((((R) >= 0) && ((unsigned long long)(R) < (sizeof(L) * 8ull))) ? ((L) >> (R)) : (flow_shift_ub_handler(), (L) * 0))
#endif

#include <math.h>

void* _ui_state = NULL;

static inline float i32_to_f32(int32_t v) { return (float)v; }

/* Host stub for @gpu kernels (device codegen replaces this). */
static inline int32_t gpu_thread_id(void) { return 0; }

bool is_square_i64(int64_t n);
int64_t count_for_m_i64(int64_t m);
int32_t main(void);

bool is_square_i64(int64_t n) {
    if (n <= 0) {
        return 0;
    }
    int64_t r = 1;
    int64_t x = n;
    while (x > 1) {
        int64_t y = FLOW_CHECKED_DIV(((x + FLOW_CHECKED_DIV((n), (x)))), (2));
        if (y >= x) {
            break;
        }
        x = y;
    }
    r = x;
    while ((r * r) > n) {
        r = (r - 1);
    }
    while (((r + 1) * (r + 1)) <= n) {
        r = (r + 1);
    }
    return (r * r) == n;
}

int64_t count_for_m_i64(int64_t m) {
    int64_t cnt = 0;
    int64_t s = 2;
    while (s <= (2 * m)) {
        if (is_square_i64(((s * s) + (m * m)))) {
            int64_t lo = (s - m);
            if (lo < 1) {
                lo = 1;
            }
            int64_t hi = FLOW_CHECKED_DIV((s), (2));
            if (hi > m) {
                hi = m;
            }
            if (hi >= lo) {
                cnt = (cnt + ((hi - lo) + 1));
            }
        }
        s = (s + 1);
    }
    return cnt;
}

int32_t main(void) {
    int64_t total = 0;
    int64_t m = 0;
    while (total <= 1000000) {
        m = (m + 1);
        total = (total + count_for_m_i64(m));
    }
    printf("%lld\n", m);
    return 0;
}

Generated MLIR

module {
  llvm.func @printf(!llvm.ptr, ...) -> i32
  llvm.mlir.global internal constant @str_0("%lld\n\00") {addr_space = 0 : i32} : !llvm.array<6 x i8>
  func.func @is_square(%arg0: i64) -> i1 {
    %0 = arith.constant 0 : i32
    %2 = arith.extsi %0 : i32 to i64
    %1 = arith.cmpi sle, %arg0, %2 : i64
    cf.cond_br %1, ^bb0, ^bb1
    ^bb0:
      %3 = arith.constant 0 : i1
      func.return %3 : i1
    ^bb1:
      cf.br ^bb2
    ^bb2:
    %4 = arith.constant 1 : i32
    %5 = arith.extsi %4 : i32 to i64
    %6 = llvm.mlir.constant(1 : i64) : i64
    %7 = llvm.alloca %6 x i64 : (i64) -> !llvm.ptr
    llvm.store %5, %7 : i64, !llvm.ptr
    %8 = llvm.mlir.constant(1 : i64) : i64
    %9 = llvm.alloca %8 x i64 : (i64) -> !llvm.ptr
    llvm.store %arg0, %9 : i64, !llvm.ptr
    cf.br ^bb3
    ^bb3:
    %10 = llvm.load %9 : !llvm.ptr -> i64
    %11 = arith.constant 1 : i32
    %13 = arith.extsi %11 : i32 to i64
    %12 = arith.cmpi sgt, %10, %13 : i64
    cf.cond_br %12, ^bb4, ^bb5
    ^bb4:
      %14 = llvm.load %9 : !llvm.ptr -> i64
      %15 = llvm.load %9 : !llvm.ptr -> i64
      %16 = arith.divsi %arg0, %15 : i64
      %17 = arith.addi %14, %16 : i64
      %18 = arith.constant 2 : i32
      %20 = arith.extsi %18 : i32 to i64
      %19 = arith.divsi %17, %20 : i64
      %21 = llvm.load %9 : !llvm.ptr -> i64
      %22 = arith.cmpi sge, %19, %21 : i64
      cf.cond_br %22, ^bb6, ^bb7
      ^bb6:
        cf.br ^bb5
      ^bb7:
        cf.br ^bb8
      ^bb8:
      llvm.store %19, %9 : i64, !llvm.ptr
      cf.br ^bb3
    ^bb5:
    %23 = llvm.load %9 : !llvm.ptr -> i64
    llvm.store %23, %7 : i64, !llvm.ptr
    cf.br ^bb9
    ^bb9:
    %24 = llvm.load %7 : !llvm.ptr -> i64
    %25 = llvm.load %7 : !llvm.ptr -> i64
    %26 = arith.muli %24, %25 : i64
    %27 = arith.cmpi sgt, %26, %arg0 : i64
    cf.cond_br %27, ^bb10, ^bb11
    ^bb10:
      %28 = llvm.load %7 : !llvm.ptr -> i64
      %29 = arith.constant 1 : i32
      %31 = arith.extsi %29 : i32 to i64
      %30 = arith.subi %28, %31 : i64
      llvm.store %30, %7 : i64, !llvm.ptr
      cf.br ^bb9
    ^bb11:
    cf.br ^bb12
    ^bb12:
    %32 = llvm.load %7 : !llvm.ptr -> i64
    %33 = arith.constant 1 : i32
    %35 = arith.extsi %33 : i32 to i64
    %34 = arith.addi %32, %35 : i64
    %36 = llvm.load %7 : !llvm.ptr -> i64
    %37 = arith.constant 1 : i32
    %39 = arith.extsi %37 : i32 to i64
    %38 = arith.addi %36, %39 : i64
    %40 = arith.muli %34, %38 : i64
    %41 = arith.cmpi sle, %40, %arg0 : i64
    cf.cond_br %41, ^bb13, ^bb14
    ^bb13:
      %42 = llvm.load %7 : !llvm.ptr -> i64
      %43 = arith.constant 1 : i32
      %45 = arith.extsi %43 : i32 to i64
      %44 = arith.addi %42, %45 : i64
      llvm.store %44, %7 : i64, !llvm.ptr
      cf.br ^bb12
    ^bb14:
    %46 = llvm.load %7 : !llvm.ptr -> i64
    %47 = llvm.load %7 : !llvm.ptr -> i64
    %48 = arith.muli %46, %47 : i64
    %49 = arith.cmpi eq, %48, %arg0 : i64
    func.return %49 : i1
  }
  func.func @count_for_m(%arg0: i64) -> i64 {
    %50 = arith.constant 0 : i32
    %51 = arith.extsi %50 : i32 to i64
    %52 = llvm.mlir.constant(1 : i64) : i64
    %53 = llvm.alloca %52 x i64 : (i64) -> !llvm.ptr
    llvm.store %51, %53 : i64, !llvm.ptr
    %54 = arith.constant 2 : i32
    %55 = arith.extsi %54 : i32 to i64
    %56 = llvm.mlir.constant(1 : i64) : i64
    %57 = llvm.alloca %56 x i64 : (i64) -> !llvm.ptr
    llvm.store %55, %57 : i64, !llvm.ptr
    cf.br ^bb15
    ^bb15:
    %58 = llvm.load %57 : !llvm.ptr -> i64
    %59 = arith.constant 2 : i32
    %61 = arith.extsi %59 : i32 to i64
    %60 = arith.muli %61, %arg0 : i64
    %62 = arith.cmpi sle, %58, %60 : i64
    cf.cond_br %62, ^bb16, ^bb17
    ^bb16:
      %64 = llvm.load %57 : !llvm.ptr -> i64
      %65 = llvm.load %57 : !llvm.ptr -> i64
      %66 = arith.muli %64, %65 : i64
      %67 = arith.muli %arg0, %arg0 : i64
      %68 = arith.addi %66, %67 : i64
      %63 = func.call @is_square(%68) : (i64) -> i1
      cf.cond_br %63, ^bb18, ^bb19
      ^bb18:
        %69 = llvm.load %57 : !llvm.ptr -> i64
        %70 = arith.subi %69, %arg0 : i64
        %71 = llvm.mlir.constant(1 : i64) : i64
        %72 = llvm.alloca %71 x i64 : (i64) -> !llvm.ptr
        llvm.store %70, %72 : i64, !llvm.ptr
        %73 = llvm.load %72 : !llvm.ptr -> i64
        %74 = arith.constant 1 : i32
        %76 = arith.extsi %74 : i32 to i64
        %75 = arith.cmpi slt, %73, %76 : i64
        cf.cond_br %75, ^bb21, ^bb22
        ^bb21:
          %77 = arith.constant 1 : i32
          %78 = arith.extsi %77 : i32 to i64
          llvm.store %78, %72 : i64, !llvm.ptr
          cf.br ^bb23
        ^bb22:
          cf.br ^bb23
        ^bb23:
        %79 = llvm.load %57 : !llvm.ptr -> i64
        %80 = arith.constant 2 : i32
        %82 = arith.extsi %80 : i32 to i64
        %81 = arith.divsi %79, %82 : i64
        %83 = llvm.mlir.constant(1 : i64) : i64
        %84 = llvm.alloca %83 x i64 : (i64) -> !llvm.ptr
        llvm.store %81, %84 : i64, !llvm.ptr
        %85 = llvm.load %84 : !llvm.ptr -> i64
        %86 = arith.cmpi sgt, %85, %arg0 : i64
        cf.cond_br %86, ^bb24, ^bb25
        ^bb24:
          llvm.store %arg0, %84 : i64, !llvm.ptr
          cf.br ^bb26
        ^bb25:
          cf.br ^bb26
        ^bb26:
        %87 = llvm.load %84 : !llvm.ptr -> i64
        %88 = llvm.load %72 : !llvm.ptr -> i64
        %89 = arith.cmpi sge, %87, %88 : i64
        cf.cond_br %89, ^bb27, ^bb28
        ^bb27:
          %90 = llvm.load %53 : !llvm.ptr -> i64
          %91 = llvm.load %84 : !llvm.ptr -> i64
          %92 = llvm.load %72 : !llvm.ptr -> i64
          %93 = arith.subi %91, %92 : i64
          %94 = arith.constant 1 : i32
          %96 = arith.extsi %94 : i32 to i64
          %95 = arith.addi %93, %96 : i64
          %97 = arith.addi %90, %95 : i64
          llvm.store %97, %53 : i64, !llvm.ptr
          cf.br ^bb29
        ^bb28:
          cf.br ^bb29
        ^bb29:
        cf.br ^bb20
      ^bb19:
        cf.br ^bb20
      ^bb20:
      %98 = llvm.load %57 : !llvm.ptr -> i64
      %99 = arith.constant 1 : i32
      %101 = arith.extsi %99 : i32 to i64
      %100 = arith.addi %98, %101 : i64
      llvm.store %100, %57 : i64, !llvm.ptr
      cf.br ^bb15
    ^bb17:
    %102 = llvm.load %53 : !llvm.ptr -> i64
    func.return %102 : i64
  }
  func.func @main() -> i32 {
    %103 = arith.constant 0 : i32
    %104 = arith.extsi %103 : i32 to i64
    %105 = llvm.mlir.constant(1 : i64) : i64
    %106 = llvm.alloca %105 x i64 : (i64) -> !llvm.ptr
    llvm.store %104, %106 : i64, !llvm.ptr
    %107 = arith.constant 0 : i32
    %108 = arith.extsi %107 : i32 to i64
    %109 = llvm.mlir.constant(1 : i64) : i64
    %110 = llvm.alloca %109 x i64 : (i64) -> !llvm.ptr
    llvm.store %108, %110 : i64, !llvm.ptr
    cf.br ^bb30
    ^bb30:
    %111 = llvm.load %106 : !llvm.ptr -> i64
    %112 = arith.constant 1000000 : i32
    %114 = arith.extsi %112 : i32 to i64
    %113 = arith.cmpi sle, %111, %114 : i64
    cf.cond_br %113, ^bb31, ^bb32
    ^bb31:
      %115 = llvm.load %110 : !llvm.ptr -> i64
      %116 = arith.constant 1 : i32
      %118 = arith.extsi %116 : i32 to i64
      %117 = arith.addi %115, %118 : i64
      llvm.store %117, %110 : i64, !llvm.ptr
      %119 = llvm.load %106 : !llvm.ptr -> i64
      %121 = llvm.load %110 : !llvm.ptr -> i64
      %120 = func.call @count_for_m(%121) : (i64) -> i64
      %122 = arith.addi %119, %120 : i64
      llvm.store %122, %106 : i64, !llvm.ptr
      cf.br ^bb30
    ^bb32:
    %123 = llvm.mlir.addressof @str_0 : !llvm.ptr
    %124 = llvm.load %110 : !llvm.ptr -> i64
    %125 = llvm.call @printf(%123, %124) vararg(!llvm.func<i32 (ptr, ...)>) : (!llvm.ptr, i64) -> i32
    %126 = arith.constant 0 : i32
    func.return %126 : i32
  }
}