Problem 733

Ascending sequences: S(10^6) mod 1e9+7 for a_i = 153^i mod 10000019. Method: rank the distinct values with a direct-address table, then use six Fenwick trees over ranks holding the count and term-sum of ascending subsequences of length 1..3 that end strictly below the current value. Each new term extends those to lengths 2..4; length-4 sums accumulate into the answer. Verified against S(6) and S(100) from the statement.

View problem on Project Euler

Performance comparison

Flow source

# Project Euler 733
# Ascending sequences: S(10^6) mod 1e9+7 for a_i = 153^i mod 10000019.
# Method: rank the distinct values with a direct-address table, then use
# six Fenwick trees over ranks holding the count and term-sum of ascending
# subsequences of length 1..3 that end strictly below the current value.
# Each new term extends those to lengths 2..4; length-4 sums accumulate
# into the answer. Verified against S(6) and S(100) from the statement.

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

function bit_query(t: ptr<i64>, i0: i64) -> i64 {
    let mut i: i64 = i0
    let mut r: i64 = 0
    while i > 0 {
        r = r + t[i]
        i = i - (i & (0 - i))
    }
    return r % 1000000007
}

function bit_update(t: ptr<i64>, i0: i64, d: i64, size: i64) -> i64 {
    let mut i: i64 = i0
    while i <= size {
        t[i] = (t[i] + d) % 1000000007
        i = i + (i & (0 - i))
    }
    return 0
}

function main() -> i32 {
    let p: i64 = 10000019
    let m: i64 = 1000000007
    let n: i64 = 1000000

    let a: ptr<i64> = calloc(n, 8)
    let rank: ptr<i32> = calloc(p, 4)
    let c1: ptr<i64> = calloc(n + 1, 8)
    let c2: ptr<i64> = calloc(n + 1, 8)
    let c3: ptr<i64> = calloc(n + 1, 8)
    let s1: ptr<i64> = calloc(n + 1, 8)
    let s2: ptr<i64> = calloc(n + 1, 8)
    let s3: ptr<i64> = calloc(n + 1, 8)

    let mut x: i64 = 1
    for i in 0..n {
        x = (x * 153) % p
        a[i] = x
        rank[x] = 1
    }

    let mut r: i64 = 0
    for v in 0..p {
        if rank[v] != 0 {
            r = r + 1
            rank[v] = (r as i32)
        }
    }

    let mut total: i64 = 0
    for i in 0..n {
        let v: i64 = a[i]
        let idx: i64 = (rank[v] as i64)
        let q: i64 = idx - 1
        let qc1: i64 = bit_query(c1, q)
        let qc2: i64 = bit_query(c2, q)
        let qc3: i64 = bit_query(c3, q)
        let qs1: i64 = bit_query(s1, q)
        let qs2: i64 = bit_query(s2, q)
        let qs3: i64 = bit_query(s3, q)
        let sum2: i64 = (qs1 + v * qc1) % m
        let sum3: i64 = (qs2 + v * qc2) % m
        let sum4: i64 = (qs3 + v * qc3) % m
        total = (total + sum4) % m
        bit_update(c1, idx, 1, n)
        bit_update(c2, idx, qc1, n)
        bit_update(c3, idx, qc2, n)
        bit_update(s1, idx, v, n)
        bit_update(s2, idx, sum2, n)
        bit_update(s3, idx, sum3, n)
    }

    free(a)
    free(rank)
    free(c1)
    free(c2)
    free(c3)
    free(s1)
    free(s2)
    free(s3)

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

int64_t bit_query_ptr_i64_i64(int64_t* t, int64_t i0);
int64_t bit_update_ptr_i64_i64_i64_i64(int64_t* t, int64_t i0, int64_t d, int64_t size);
int32_t main(void);



int64_t bit_query_ptr_i64_i64(int64_t* t, int64_t i0) {
    int64_t i = i0;
    int64_t r = 0;
    while (i > 0) {
        r = (r + t[i]);
        i = (i - (i & (0 - i)));
    }
    return FLOW_CHECKED_MOD((r), (1000000007));
}

int64_t bit_update_ptr_i64_i64_i64_i64(int64_t* t, int64_t i0, int64_t d, int64_t size) {
    int64_t i = i0;
    while (i <= size) {
        t[i] = FLOW_CHECKED_MOD(((t[i] + d)), (1000000007));
        i = (i + (i & (0 - i)));
    }
    return 0;
}

int32_t main(void) {
    int64_t p = 10000019;
    int64_t m = 1000000007;
    int64_t n = 1000000;
    int64_t* a = (int64_t*)(calloc(n, 8));
    int32_t* rank = (int32_t*)(calloc(p, 4));
    int64_t* c1 = (int64_t*)(calloc((n + 1), 8));
    int64_t* c2 = (int64_t*)(calloc((n + 1), 8));
    int64_t* c3 = (int64_t*)(calloc((n + 1), 8));
    int64_t* s1 = (int64_t*)(calloc((n + 1), 8));
    int64_t* s2 = (int64_t*)(calloc((n + 1), 8));
    int64_t* s3 = (int64_t*)(calloc((n + 1), 8));
    int64_t x = 1;
    int32_t __flow_step_1 = 1;
    for (int32_t i = 0; (0 <= n) ? i < n : i > n; i += (0 <= n) ? 1 : -1) {
        x = FLOW_CHECKED_MOD(((x * 153)), (p));
        a[i] = x;
        rank[x] = 1;
    }
    int64_t r = 0;
    int32_t __flow_step_2 = 1;
    for (int32_t v = 0; (0 <= p) ? v < p : v > p; v += (0 <= p) ? 1 : -1) {
        if (rank[v] != 0) {
            r = (r + 1);
            rank[v] = ((int32_t)(r));
        }
    }
    int64_t total = 0;
    int32_t __flow_step_3 = 1;
    for (int32_t i = 0; (0 <= n) ? i < n : i > n; i += (0 <= n) ? 1 : -1) {
        int64_t v = a[i];
        int64_t idx = ((int64_t)(rank[v]));
        int64_t q = (idx - 1);
        int64_t qc1 = bit_query_ptr_i64_i64(c1, q);
        int64_t qc2 = bit_query_ptr_i64_i64(c2, q);
        int64_t qc3 = bit_query_ptr_i64_i64(c3, q);
        int64_t qs1 = bit_query_ptr_i64_i64(s1, q);
        int64_t qs2 = bit_query_ptr_i64_i64(s2, q);
        int64_t qs3 = bit_query_ptr_i64_i64(s3, q);
        int64_t sum2 = FLOW_CHECKED_MOD(((qs1 + (v * qc1))), (m));
        int64_t sum3 = FLOW_CHECKED_MOD(((qs2 + (v * qc2))), (m));
        int64_t sum4 = FLOW_CHECKED_MOD(((qs3 + (v * qc3))), (m));
        total = FLOW_CHECKED_MOD(((total + sum4)), (m));
        bit_update_ptr_i64_i64_i64_i64(c1, idx, 1, n);
        bit_update_ptr_i64_i64_i64_i64(c2, idx, qc1, n);
        bit_update_ptr_i64_i64_i64_i64(c3, idx, qc2, n);
        bit_update_ptr_i64_i64_i64_i64(s1, idx, v, n);
        bit_update_ptr_i64_i64_i64_i64(s2, idx, sum2, n);
        bit_update_ptr_i64_i64_i64_i64(s3, idx, sum3, n);
    }
    free(a);
    free(rank);
    free(c1);
    free(c2);
    free(c3);
    free(s1);
    free(s2);
    free(s3);
    printf("%lld\n", total);
    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 @bit_query(%arg0: !llvm.ptr, %arg1: i64) -> i64 {
    %0 = llvm.mlir.constant(1 : i64) : i64
    %1 = llvm.alloca %0 x i64 : (i64) -> !llvm.ptr
    llvm.store %arg1, %1 : i64, !llvm.ptr
    %2 = arith.constant 0 : i32
    %3 = arith.extsi %2 : i32 to i64
    %4 = llvm.mlir.constant(1 : i64) : i64
    %5 = llvm.alloca %4 x i64 : (i64) -> !llvm.ptr
    llvm.store %3, %5 : i64, !llvm.ptr
    cf.br ^bb0
    ^bb0:
    %6 = llvm.load %1 : !llvm.ptr -> i64
    %7 = arith.constant 0 : i32
    %9 = arith.extsi %7 : i32 to i64
    %8 = arith.cmpi sgt, %6, %9 : i64
    cf.cond_br %8, ^bb1, ^bb2
    ^bb1:
      %10 = llvm.load %5 : !llvm.ptr -> i64
      %12 = llvm.load %1 : !llvm.ptr -> i64
      %13 = llvm.getelementptr %arg0[%12] : (!llvm.ptr, i64) -> !llvm.ptr, i64
      %11 = llvm.load %13 : !llvm.ptr -> i64
      %14 = arith.addi %10, %11 : i64
      llvm.store %14, %5 : i64, !llvm.ptr
      %15 = llvm.load %1 : !llvm.ptr -> i64
      %16 = llvm.load %1 : !llvm.ptr -> i64
      %17 = arith.constant 0 : i32
      %18 = llvm.load %1 : !llvm.ptr -> i64
      %20 = arith.extsi %17 : i32 to i64
      %19 = arith.subi %20, %18 : i64
      %21 = arith.andi %16, %19 : i64
      %22 = arith.subi %15, %21 : i64
      llvm.store %22, %1 : i64, !llvm.ptr
      cf.br ^bb0
    ^bb2:
    %23 = llvm.load %5 : !llvm.ptr -> i64
    %24 = arith.constant 1000000007 : i32
    %26 = arith.extsi %24 : i32 to i64
    %25 = arith.remsi %23, %26 : i64
    func.return %25 : i64
  }
  func.func @bit_update(%arg0: !llvm.ptr, %arg1: i64, %arg2: i64, %arg3: i64) -> i64 {
    %27 = llvm.mlir.constant(1 : i64) : i64
    %28 = llvm.alloca %27 x i64 : (i64) -> !llvm.ptr
    llvm.store %arg1, %28 : i64, !llvm.ptr
    cf.br ^bb3
    ^bb3:
    %29 = llvm.load %28 : !llvm.ptr -> i64
    %30 = arith.cmpi sle, %29, %arg3 : i64
    cf.cond_br %30, ^bb4, ^bb5
    ^bb4:
      %32 = llvm.load %28 : !llvm.ptr -> i64
      %33 = llvm.getelementptr %arg0[%32] : (!llvm.ptr, i64) -> !llvm.ptr, i64
      %31 = llvm.load %33 : !llvm.ptr -> i64
      %34 = arith.addi %31, %arg2 : i64
      %35 = arith.constant 1000000007 : i32
      %37 = arith.extsi %35 : i32 to i64
      %36 = arith.remsi %34, %37 : i64
      %38 = llvm.load %28 : !llvm.ptr -> i64
      %39 = llvm.getelementptr %arg0[%38] : (!llvm.ptr, i64) -> !llvm.ptr, i64
      llvm.store %36, %39 : i64, !llvm.ptr
      %40 = llvm.load %28 : !llvm.ptr -> i64
      %41 = llvm.load %28 : !llvm.ptr -> i64
      %42 = arith.constant 0 : i32
      %43 = llvm.load %28 : !llvm.ptr -> i64
      %45 = arith.extsi %42 : i32 to i64
      %44 = arith.subi %45, %43 : i64
      %46 = arith.andi %41, %44 : i64
      %47 = arith.addi %40, %46 : i64
      llvm.store %47, %28 : i64, !llvm.ptr
      cf.br ^bb3
    ^bb5:
    %48 = arith.constant 0 : i32
    %49 = arith.extsi %48 : i32 to i64
    func.return %49 : i64
  }
  func.func @main() -> i32 {
    %50 = arith.constant 10000019 : i32
    %51 = arith.extsi %50 : i32 to i64
    %52 = arith.constant 1000000007 : i32
    %53 = arith.extsi %52 : i32 to i64
    %54 = arith.constant 1000000 : i32
    %55 = arith.extsi %54 : i32 to i64
    %57 = arith.constant 8 : i32
    %58 = arith.extsi %57 : i32 to i64
    %56 = func.call @calloc(%55, %58) : (i64, i64) -> !llvm.ptr
    %60 = arith.constant 4 : i32
    %61 = arith.extsi %60 : i32 to i64
    %59 = func.call @calloc(%51, %61) : (i64, i64) -> !llvm.ptr
    %63 = arith.constant 1 : i32
    %65 = arith.extsi %63 : i32 to i64
    %64 = arith.addi %55, %65 : i64
    %66 = arith.constant 8 : i32
    %67 = arith.extsi %66 : i32 to i64
    %62 = func.call @calloc(%64, %67) : (i64, i64) -> !llvm.ptr
    %69 = arith.constant 1 : i32
    %71 = arith.extsi %69 : i32 to i64
    %70 = arith.addi %55, %71 : i64
    %72 = arith.constant 8 : i32
    %73 = arith.extsi %72 : i32 to i64
    %68 = func.call @calloc(%70, %73) : (i64, i64) -> !llvm.ptr
    %75 = arith.constant 1 : i32
    %77 = arith.extsi %75 : i32 to i64
    %76 = arith.addi %55, %77 : i64
    %78 = arith.constant 8 : i32
    %79 = arith.extsi %78 : i32 to i64
    %74 = func.call @calloc(%76, %79) : (i64, i64) -> !llvm.ptr
    %81 = arith.constant 1 : i32
    %83 = arith.extsi %81 : i32 to i64
    %82 = arith.addi %55, %83 : i64
    %84 = arith.constant 8 : i32
    %85 = arith.extsi %84 : i32 to i64
    %80 = func.call @calloc(%82, %85) : (i64, i64) -> !llvm.ptr
    %87 = arith.constant 1 : i32
    %89 = arith.extsi %87 : i32 to i64
    %88 = arith.addi %55, %89 : i64
    %90 = arith.constant 8 : i32
    %91 = arith.extsi %90 : i32 to i64
    %86 = func.call @calloc(%88, %91) : (i64, i64) -> !llvm.ptr
    %93 = arith.constant 1 : i32
    %95 = arith.extsi %93 : i32 to i64
    %94 = arith.addi %55, %95 : i64
    %96 = arith.constant 8 : i32
    %97 = arith.extsi %96 : i32 to i64
    %92 = func.call @calloc(%94, %97) : (i64, i64) -> !llvm.ptr
    %98 = arith.constant 1 : i32
    %99 = arith.extsi %98 : i32 to i64
    %100 = llvm.mlir.constant(1 : i64) : i64
    %101 = llvm.alloca %100 x i64 : (i64) -> !llvm.ptr
    llvm.store %99, %101 : i64, !llvm.ptr
    %102 = arith.constant 0 : i32
    %103 = arith.index_cast %102 : i32 to index
    %104 = arith.index_cast %55 : i32 to index
    %106 = arith.constant 1 : index
    %107 = arith.constant -1 : index
    %108 = arith.cmpi sle, %103, %104 : index
    %105 = arith.select %108, %106, %107 : index
    cf.br ^bb6(%103 : index)
    ^bb6(%109: index):
    %110 = arith.cmpi slt, %109, %104 : index
    %111 = arith.cmpi sgt, %109, %104 : index
    %112 = arith.select %108, %110, %111 : i1
    cf.cond_br %112, ^bb7(%109 : index), ^bb8(%109 : index)
    ^bb7(%113: index):
      %114 = llvm.load %101 : !llvm.ptr -> i64
      %115 = arith.constant 153 : i32
      %117 = arith.extsi %115 : i32 to i64
      %116 = arith.muli %114, %117 : i64
      %118 = arith.remsi %116, %51 : i64
      llvm.store %118, %101 : i64, !llvm.ptr
      %119 = llvm.load %101 : !llvm.ptr -> i64
      %120 = arith.index_cast %113 : index to i64
      %121 = llvm.getelementptr %56[%120] : (!llvm.ptr, i64) -> !llvm.ptr, i64
      llvm.store %119, %121 : i64, !llvm.ptr
      %122 = arith.constant 1 : i32
      %123 = llvm.load %101 : !llvm.ptr -> i64
      %124 = llvm.getelementptr %59[%123] : (!llvm.ptr, i64) -> !llvm.ptr, i32
      llvm.store %122, %124 : i32, !llvm.ptr
      %125 = arith.addi %113, %105 : index
      cf.br ^bb6(%125 : index)
    ^bb8(%126: index):
    %127 = arith.constant 0 : i32
    %128 = arith.extsi %127 : i32 to i64
    %129 = llvm.mlir.constant(1 : i64) : i64
    %130 = llvm.alloca %129 x i64 : (i64) -> !llvm.ptr
    llvm.store %128, %130 : i64, !llvm.ptr
    %131 = arith.constant 0 : i32
    %132 = arith.index_cast %131 : i32 to index
    %133 = arith.index_cast %51 : i32 to index
    %135 = arith.constant 1 : index
    %136 = arith.constant -1 : index
    %137 = arith.cmpi sle, %132, %133 : index
    %134 = arith.select %137, %135, %136 : index
    cf.br ^bb9(%132 : index)
    ^bb9(%138: index):
    %139 = arith.cmpi slt, %138, %133 : index
    %140 = arith.cmpi sgt, %138, %133 : index
    %141 = arith.select %137, %139, %140 : i1
    cf.cond_br %141, ^bb10(%138 : index), ^bb11(%138 : index)
    ^bb10(%142: index):
      %144 = arith.index_cast %142 : index to i64
      %145 = llvm.getelementptr %59[%144] : (!llvm.ptr, i64) -> !llvm.ptr, i32
      %143 = llvm.load %145 : !llvm.ptr -> i32
      %146 = arith.constant 0 : i32
      %147 = arith.cmpi ne, %143, %146 : i32
      cf.cond_br %147, ^bb12, ^bb13
      ^bb12:
        %148 = llvm.load %130 : !llvm.ptr -> i64
        %149 = arith.constant 1 : i32
        %151 = arith.extsi %149 : i32 to i64
        %150 = arith.addi %148, %151 : i64
        llvm.store %150, %130 : i64, !llvm.ptr
        %152 = llvm.load %130 : !llvm.ptr -> i64
        %153 = arith.trunci %152 : i64 to i32
        %154 = arith.index_cast %142 : index to i64
        %155 = llvm.getelementptr %59[%154] : (!llvm.ptr, i64) -> !llvm.ptr, i32
        llvm.store %153, %155 : i32, !llvm.ptr
        cf.br ^bb14
      ^bb13:
        cf.br ^bb14
      ^bb14:
      %156 = arith.addi %142, %134 : index
      cf.br ^bb9(%156 : index)
    ^bb11(%157: index):
    %158 = arith.constant 0 : i32
    %159 = arith.extsi %158 : i32 to i64
    %160 = llvm.mlir.constant(1 : i64) : i64
    %161 = llvm.alloca %160 x i64 : (i64) -> !llvm.ptr
    llvm.store %159, %161 : i64, !llvm.ptr
    %162 = arith.constant 0 : i32
    %163 = arith.index_cast %162 : i32 to index
    %164 = arith.index_cast %55 : i32 to index
    %166 = arith.constant 1 : index
    %167 = arith.constant -1 : index
    %168 = arith.cmpi sle, %163, %164 : index
    %165 = arith.select %168, %166, %167 : index
    cf.br ^bb15(%163 : index)
    ^bb15(%169: index):
    %170 = arith.cmpi slt, %169, %164 : index
    %171 = arith.cmpi sgt, %169, %164 : index
    %172 = arith.select %168, %170, %171 : i1
    cf.cond_br %172, ^bb16(%169 : index), ^bb17(%169 : index)
    ^bb16(%173: index):
      %175 = arith.index_cast %173 : index to i64
      %176 = llvm.getelementptr %56[%175] : (!llvm.ptr, i64) -> !llvm.ptr, i64
      %174 = llvm.load %176 : !llvm.ptr -> i64
      %178 = llvm.getelementptr %59[%174] : (!llvm.ptr, i64) -> !llvm.ptr, i32
      %177 = llvm.load %178 : !llvm.ptr -> i32
      %179 = arith.extsi %177 : i32 to i64
      %180 = arith.constant 1 : i32
      %182 = arith.extsi %180 : i32 to i64
      %181 = arith.subi %179, %182 : i64
      %183 = func.call @bit_query(%62, %181) : (!llvm.ptr, i64) -> i64
      %184 = func.call @bit_query(%68, %181) : (!llvm.ptr, i64) -> i64
      %185 = func.call @bit_query(%74, %181) : (!llvm.ptr, i64) -> i64
      %186 = func.call @bit_query(%80, %181) : (!llvm.ptr, i64) -> i64
      %187 = func.call @bit_query(%86, %181) : (!llvm.ptr, i64) -> i64
      %188 = func.call @bit_query(%92, %181) : (!llvm.ptr, i64) -> i64
      %189 = arith.muli %174, %183 : i64
      %190 = arith.addi %186, %189 : i64
      %191 = arith.remsi %190, %53 : i64
      %192 = arith.muli %174, %184 : i64
      %193 = arith.addi %187, %192 : i64
      %194 = arith.remsi %193, %53 : i64
      %195 = arith.muli %174, %185 : i64
      %196 = arith.addi %188, %195 : i64
      %197 = arith.remsi %196, %53 : i64
      %198 = llvm.load %161 : !llvm.ptr -> i64
      %199 = arith.addi %198, %197 : i64
      %200 = arith.remsi %199, %53 : i64
      llvm.store %200, %161 : i64, !llvm.ptr
      %202 = arith.constant 1 : i32
      %203 = arith.extsi %202 : i32 to i64
      %201 = func.call @bit_update(%62, %179, %203, %55) : (!llvm.ptr, i64, i64, i64) -> i64
      %204 = func.call @bit_update(%68, %179, %183, %55) : (!llvm.ptr, i64, i64, i64) -> i64
      %205 = func.call @bit_update(%74, %179, %184, %55) : (!llvm.ptr, i64, i64, i64) -> i64
      %206 = func.call @bit_update(%80, %179, %174, %55) : (!llvm.ptr, i64, i64, i64) -> i64
      %207 = func.call @bit_update(%86, %179, %191, %55) : (!llvm.ptr, i64, i64, i64) -> i64
      %208 = func.call @bit_update(%92, %179, %194, %55) : (!llvm.ptr, i64, i64, i64) -> i64
      %209 = arith.addi %173, %165 : index
      cf.br ^bb15(%209 : index)
    ^bb17(%210: index):
    func.call @free(%56) : (!llvm.ptr) -> ()
    func.call @free(%59) : (!llvm.ptr) -> ()
    func.call @free(%62) : (!llvm.ptr) -> ()
    func.call @free(%68) : (!llvm.ptr) -> ()
    func.call @free(%74) : (!llvm.ptr) -> ()
    func.call @free(%80) : (!llvm.ptr) -> ()
    func.call @free(%86) : (!llvm.ptr) -> ()
    func.call @free(%92) : (!llvm.ptr) -> ()
    %219 = llvm.mlir.addressof @str_0 : !llvm.ptr
    %220 = llvm.load %161 : !llvm.ptr -> i64
    %221 = llvm.call @printf(%219, %220) vararg(!llvm.func<i32 (ptr, ...)>) : (!llvm.ptr, i64) -> i32
    %222 = arith.constant 0 : i32
    func.return %222 : i32
  }
}