Problem 021

Sum of amicable numbers under 10000.

Answer31626
Output31626
StatusPASS
Native helperno
Runtime0 ms
Peak memory1184 KB
Time complexityO(n^2) (estimated)
Space complexityO(n) (estimated)

Performance comparison

MetricOur solutionBest known
Time complexityO(n^2)O(n log log n)
Space complexityO(n)O(n)
ApproachFlow solutionSieve-based sum of proper divisors
VerdictSuboptimal

Flow source

# Project Euler 021
# Sum of amicable numbers under 10000.

extern {
    function calloc(n: i64, size: i64) -> ptr<void>
    function free(p: ptr<void>) -> void
}

function main() -> i32 {
    let limit: i64 = 10000
    let sumdiv: ptr<i64> = calloc(limit, 8)
    if sumdiv == null {
        return 1
    }

    for i in 1..limit {
        let mut j: i64 = 2 * i
        while j < limit {
            sumdiv[j] = sumdiv[j] + i
            j = j + i
        }
    }

    let mut total: i64 = 0
    for i in 2..limit {
        let b: i64 = sumdiv[i]
        if b > i && b < limit && sumdiv[b] == i {
            total = total + i + b
        } elif b >= limit {
            # need sum of divisors of b if b is outside table
            let mut sb: i64 = 1
            let mut d: i64 = 2
            while d * d <= b {
                if b % d == 0 {
                    sb = sb + d
                    if d * d != b {
                        sb = sb + b / d
                    }
                }
                d = d + 1
            }
            if sb == i {
                total = total + i
            }
        }
    }

    printf("%lld\n", total)
    free(sumdiv)
    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; }

int32_t main(void);



int32_t main(void) {
    int64_t limit = 10000;
    int64_t* sumdiv = (int64_t*)(calloc(limit, 8));
    if (sumdiv == NULL) {
        return 1;
    }
    int32_t __flow_step_1 = 1;
    for (int32_t i = 1; (1 <= limit) ? i < limit : i > limit; i += (1 <= limit) ? 1 : -1) {
        int64_t j = (2 * i);
        while (j < limit) {
            sumdiv[j] = (sumdiv[j] + i);
            j = (j + i);
        }
    }
    int64_t total = 0;
    int32_t __flow_step_2 = 1;
    for (int32_t i = 2; (2 <= limit) ? i < limit : i > limit; i += (2 <= limit) ? 1 : -1) {
        int64_t b = sumdiv[i];
        if (((b > i && b < limit) && sumdiv[b] == i)) {
            total = ((total + i) + b);
        } else if (b >= limit) {
            int64_t sb = 1;
            int64_t d = 2;
            while ((d * d) <= b) {
                if (FLOW_CHECKED_MOD((b), (d)) == 0) {
                    sb = (sb + d);
                    if ((d * d) != b) {
                        sb = (sb + FLOW_CHECKED_DIV((b), (d)));
                    }
                }
                d = (d + 1);
            }
            if (sb == i) {
                total = (total + i);
            }
        }
    }
    printf("%lld\n", total);
    free(sumdiv);
    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 private @calloc(i64, i64) -> !llvm.ptr
  func.func private @free(!llvm.ptr) -> ()
  func.func @main() -> i32 {
    %0 = arith.constant 10000 : i32
    %1 = arith.extsi %0 : i32 to i64
    %3 = arith.constant 8 : i32
    %4 = arith.extsi %3 : i32 to i64
    %2 = func.call @calloc(%1, %4) : (i64, i64) -> !llvm.ptr
    %5 = llvm.mlir.zero : !llvm.ptr
    %6 = llvm.icmp "eq" %2, %5 : !llvm.ptr
    cf.cond_br %6, ^bb0, ^bb1
    ^bb0:
      %7 = arith.constant 1 : i32
      func.return %7 : i32
    ^bb1:
      cf.br ^bb2
    ^bb2:
    %8 = arith.constant 1 : i32
    %9 = arith.index_cast %8 : i32 to index
    %10 = arith.index_cast %1 : i32 to index
    %12 = arith.constant 1 : index
    %13 = arith.constant -1 : index
    %14 = arith.cmpi sle, %9, %10 : index
    %11 = arith.select %14, %12, %13 : index
    cf.br ^bb3(%9 : index)
    ^bb3(%15: index):
    %16 = arith.cmpi slt, %15, %10 : index
    %17 = arith.cmpi sgt, %15, %10 : index
    %18 = arith.select %14, %16, %17 : i1
    cf.cond_br %18, ^bb4(%15 : index), ^bb5(%15 : index)
    ^bb4(%19: index):
      %20 = arith.constant 2 : i32
      %22 = arith.index_cast %19 : index to i32
      %21 = arith.muli %20, %22 : i32
      %23 = arith.extsi %21 : i32 to i64
      %24 = llvm.mlir.constant(1 : i64) : i64
      %25 = llvm.alloca %24 x i64 : (i64) -> !llvm.ptr
      llvm.store %23, %25 : i64, !llvm.ptr
      cf.br ^bb6
      ^bb6:
      %26 = llvm.load %25 : !llvm.ptr -> i64
      %27 = arith.cmpi slt, %26, %1 : i64
      cf.cond_br %27, ^bb7, ^bb8
      ^bb7:
        %29 = llvm.load %25 : !llvm.ptr -> i64
        %30 = llvm.getelementptr %2[%29] : (!llvm.ptr, i64) -> !llvm.ptr, i64
        %28 = llvm.load %30 : !llvm.ptr -> i64
        %32 = arith.trunci %28 : i64 to i32
        %33 = arith.index_cast %19 : index to i32
        %31 = arith.addi %32, %33 : i32
        %34 = llvm.load %25 : !llvm.ptr -> i64
        %35 = arith.extsi %31 : i32 to i64
        %36 = llvm.getelementptr %2[%34] : (!llvm.ptr, i64) -> !llvm.ptr, i64
        llvm.store %35, %36 : i64, !llvm.ptr
        %37 = llvm.load %25 : !llvm.ptr -> i64
        %39 = arith.trunci %37 : i64 to i32
        %40 = arith.index_cast %19 : index to i32
        %38 = arith.addi %39, %40 : i32
        %41 = arith.extsi %38 : i32 to i64
        llvm.store %41, %25 : i64, !llvm.ptr
        cf.br ^bb6
      ^bb8:
      %42 = arith.addi %19, %11 : index
      cf.br ^bb3(%42 : index)
    ^bb5(%43: index):
    %44 = arith.constant 0 : i32
    %45 = arith.extsi %44 : i32 to i64
    %46 = llvm.mlir.constant(1 : i64) : i64
    %47 = llvm.alloca %46 x i64 : (i64) -> !llvm.ptr
    llvm.store %45, %47 : i64, !llvm.ptr
    %48 = arith.constant 2 : i32
    %49 = arith.index_cast %48 : i32 to index
    %50 = arith.index_cast %1 : i32 to index
    %52 = arith.constant 1 : index
    %53 = arith.constant -1 : index
    %54 = arith.cmpi sle, %49, %50 : index
    %51 = arith.select %54, %52, %53 : index
    cf.br ^bb9(%49 : index)
    ^bb9(%55: index):
    %56 = arith.cmpi slt, %55, %50 : index
    %57 = arith.cmpi sgt, %55, %50 : index
    %58 = arith.select %54, %56, %57 : i1
    cf.cond_br %58, ^bb10(%55 : index), ^bb11(%55 : index)
    ^bb10(%59: index):
      %61 = arith.index_cast %59 : index to i64
      %62 = llvm.getelementptr %2[%61] : (!llvm.ptr, i64) -> !llvm.ptr, i64
      %60 = llvm.load %62 : !llvm.ptr -> i64
      %64 = arith.trunci %60 : i64 to i32
      %65 = arith.index_cast %59 : index to i32
      %63 = arith.cmpi sgt, %64, %65 : i32
      %66 = scf.if %63 -> (i1) {
        %67 = arith.cmpi slt, %60, %1 : i64
        scf.yield %67 : i1
      } else {
        %68 = arith.constant false
        scf.yield %68 : i1
      }
      %69 = scf.if %66 -> (i1) {
        %71 = llvm.getelementptr %2[%60] : (!llvm.ptr, i64) -> !llvm.ptr, i64
        %70 = llvm.load %71 : !llvm.ptr -> i64
        %73 = arith.trunci %70 : i64 to i32
        %74 = arith.index_cast %59 : index to i32
        %72 = arith.cmpi eq, %73, %74 : i32
        scf.yield %72 : i1
      } else {
        %75 = arith.constant false
        scf.yield %75 : i1
      }
      cf.cond_br %69, ^bb12, ^bb13
      ^bb12:
        %76 = llvm.load %47 : !llvm.ptr -> i64
        %78 = arith.trunci %76 : i64 to i32
        %79 = arith.index_cast %59 : index to i32
        %77 = arith.addi %78, %79 : i32
        %81 = arith.extsi %77 : i32 to i64
        %80 = arith.addi %81, %60 : i64
        llvm.store %80, %47 : i64, !llvm.ptr
        cf.br ^bb14
      ^bb13:
        %82 = arith.cmpi sge, %60, %1 : i64
        cf.cond_br %82, ^bb15, ^bb14
      ^bb15:
        %83 = arith.constant 1 : i32
        %84 = arith.extsi %83 : i32 to i64
        %85 = llvm.mlir.constant(1 : i64) : i64
        %86 = llvm.alloca %85 x i64 : (i64) -> !llvm.ptr
        llvm.store %84, %86 : i64, !llvm.ptr
        %87 = arith.constant 2 : i32
        %88 = arith.extsi %87 : i32 to i64
        %89 = llvm.mlir.constant(1 : i64) : i64
        %90 = llvm.alloca %89 x i64 : (i64) -> !llvm.ptr
        llvm.store %88, %90 : i64, !llvm.ptr
        cf.br ^bb16
        ^bb16:
        %91 = llvm.load %90 : !llvm.ptr -> i64
        %92 = llvm.load %90 : !llvm.ptr -> i64
        %93 = arith.muli %91, %92 : i64
        %94 = arith.cmpi sle, %93, %60 : i64
        cf.cond_br %94, ^bb17, ^bb18
        ^bb17:
          %95 = llvm.load %90 : !llvm.ptr -> i64
          %96 = arith.remsi %60, %95 : i64
          %97 = arith.constant 0 : i32
          %99 = arith.extsi %97 : i32 to i64
          %98 = arith.cmpi eq, %96, %99 : i64
          cf.cond_br %98, ^bb19, ^bb20
          ^bb19:
            %100 = llvm.load %86 : !llvm.ptr -> i64
            %101 = llvm.load %90 : !llvm.ptr -> i64
            %102 = arith.addi %100, %101 : i64
            llvm.store %102, %86 : i64, !llvm.ptr
            %103 = llvm.load %90 : !llvm.ptr -> i64
            %104 = llvm.load %90 : !llvm.ptr -> i64
            %105 = arith.muli %103, %104 : i64
            %106 = arith.cmpi ne, %105, %60 : i64
            cf.cond_br %106, ^bb22, ^bb23
            ^bb22:
              %107 = llvm.load %86 : !llvm.ptr -> i64
              %108 = llvm.load %90 : !llvm.ptr -> i64
              %109 = arith.divsi %60, %108 : i64
              %110 = arith.addi %107, %109 : i64
              llvm.store %110, %86 : i64, !llvm.ptr
              cf.br ^bb24
            ^bb23:
              cf.br ^bb24
            ^bb24:
            cf.br ^bb21
          ^bb20:
            cf.br ^bb21
          ^bb21:
          %111 = llvm.load %90 : !llvm.ptr -> i64
          %112 = arith.constant 1 : i32
          %114 = arith.extsi %112 : i32 to i64
          %113 = arith.addi %111, %114 : i64
          llvm.store %113, %90 : i64, !llvm.ptr
          cf.br ^bb16
        ^bb18:
        %115 = llvm.load %86 : !llvm.ptr -> i64
        %117 = arith.trunci %115 : i64 to i32
        %118 = arith.index_cast %59 : index to i32
        %116 = arith.cmpi eq, %117, %118 : i32
        cf.cond_br %116, ^bb25, ^bb26
        ^bb25:
          %119 = llvm.load %47 : !llvm.ptr -> i64
          %121 = arith.trunci %119 : i64 to i32
          %122 = arith.index_cast %59 : index to i32
          %120 = arith.addi %121, %122 : i32
          %123 = arith.extsi %120 : i32 to i64
          llvm.store %123, %47 : i64, !llvm.ptr
          cf.br ^bb27
        ^bb26:
          cf.br ^bb27
        ^bb27:
        cf.br ^bb14
      ^bb14:
      %124 = arith.addi %59, %51 : index
      cf.br ^bb9(%124 : index)
    ^bb11(%125: index):
    %126 = llvm.mlir.addressof @str_0 : !llvm.ptr
    %127 = llvm.load %47 : !llvm.ptr -> i64
    %128 = llvm.call @printf(%126, %127) vararg(!llvm.func<i32 (ptr, ...)>) : (!llvm.ptr, i64) -> i32
    func.call @free(%2) : (!llvm.ptr) -> ()
    %130 = arith.constant 0 : i32
    func.return %130 : i32
  }
}