Problem 750

Optimal Card Stacking: G(976). Positions generated by x=(x*3)%(N+1). DP over intervals [l,r] with cost = |pos[k]-pos[r]| for the split point k.

Answer160640
Output160640
StatusPASS
Native helperno
Runtime160 ms
Peak memory8576 KB
Time complexityO(n^3) (estimated)
Space complexityO(n) (estimated)

Performance comparison

MetricOur solutionBest known
Time complexityO(n^3)O(n)
Space complexityO(n)O(1)
ApproachFlow solutionDirect hand evaluation or enumeration
VerdictSuboptimal

Flow source

# Project Euler 750
# Optimal Card Stacking: G(976).
# Positions generated by x=(x*3)%(N+1). DP over intervals [l,r]
# with cost = |pos[k]-pos[r]| for the split point k.

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

const N: i64 = 976
const INF: i64 = 4000000000000000000

function iabs(x: i64) -> i64 {
    if x < 0 {
        return 0 - x
    }
    return x
}

function main() -> i32 {
    let mod: i64 = N + 1
    let pos: ptr<i64> = calloc(N + 1, 8)
    if pos == null {
        return 1
    }
    let mut x: i64 = 1
    let mut i: i64 = 1
    while i <= N {
        x = (x * 3) % mod
        if x == 0 || x > N || pos[x] != 0 {
            free(pos as ptr<void>)
            printf("%lld\n", -1)
            return 0
        }
        pos[x] = i
        i = i + 1
    }

    let dp: ptr<i64> = calloc((N + 1) * (N + 1), 8)
    if dp == null {
        free(pos as ptr<void>)
        return 1
    }

    let mut r: i64 = 2
    while r <= N {
        let mut l: i64 = r - 1
        while l >= 1 {
            let mut best: i64 = INF
            let mut k: i64 = l
            while k < r {
                let cost: i64 = dp[l * (N + 1) + k] + dp[(k + 1) * (N + 1) + r] + iabs(pos[k] - pos[r])
                if cost < best {
                    best = cost
                }
                k = k + 1
            }
            dp[l * (N + 1) + r] = best
            l = l - 1
        }
        r = r + 1
    }

    let ans: i64 = dp[1 * (N + 1) + N]
    free(dp as ptr<void>)
    free(pos as ptr<void>)
    printf("%lld\n", ans)
    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 iabs_i64(int64_t x);
int32_t main(void);

static const int64_t N = 976;
static const int64_t INF = 4000000000000000000;



int64_t iabs_i64(int64_t x) {
    if (x < 0) {
        return (0 - x);
    }
    return x;
}

int32_t main(void) {
    int64_t mod = (N + 1);
    int64_t* pos = (int64_t*)(calloc((N + 1), 8));
    if (pos == NULL) {
        return 1;
    }
    int64_t x = 1;
    int64_t i = 1;
    while (i <= N) {
        x = FLOW_CHECKED_MOD(((x * 3)), (mod));
        if (((x == 0 || x > N) || pos[x] != 0)) {
            free(((void*)(pos)));
            printf("%lld\n", (-1));
            return 0;
        }
        pos[x] = i;
        i = (i + 1);
    }
    int64_t* dp = (int64_t*)(calloc(((N + 1) * (N + 1)), 8));
    if (dp == NULL) {
        free(((void*)(pos)));
        return 1;
    }
    int64_t r = 2;
    while (r <= N) {
        int64_t l = (r - 1);
        while (l >= 1) {
            int64_t best = INF;
            int64_t k = l;
            while (k < r) {
                int64_t cost = ((dp[((l * (N + 1)) + k)] + dp[(((k + 1) * (N + 1)) + r)]) + iabs_i64((pos[k] - pos[r])));
                if (cost < best) {
                    best = cost;
                }
                k = (k + 1);
            }
            dp[((l * (N + 1)) + r)] = best;
            l = (l - 1);
        }
        r = (r + 1);
    }
    int64_t ans = dp[((1 * (N + 1)) + N)];
    free(((void*)(dp)));
    free(((void*)(pos)));
    printf("%lld\n", ans);
    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) -> ()
  // Constant: N
  llvm.mlir.global internal constant @N(976 : i64) : i64
  // Constant: INF
  llvm.mlir.global internal constant @INF(4000000000000000000 : i64) : i64
  func.func @iabs(%arg0: i64) -> i64 {
    %0 = arith.constant 0 : i32
    %2 = arith.extsi %0 : i32 to i64
    %1 = arith.cmpi slt, %arg0, %2 : i64
    cf.cond_br %1, ^bb0, ^bb1
    ^bb0:
      %3 = arith.constant 0 : i32
      %5 = arith.extsi %3 : i32 to i64
      %4 = arith.subi %5, %arg0 : i64
      func.return %4 : i64
    ^bb1:
      cf.br ^bb2
    ^bb2:
    func.return %arg0 : i64
  }
  func.func @main() -> i32 {
    %6 = llvm.mlir.addressof @N : !llvm.ptr
    %7 = llvm.load %6 : !llvm.ptr -> i64
    %8 = arith.constant 1 : i32
    %10 = arith.extsi %8 : i32 to i64
    %9 = arith.addi %7, %10 : i64
    %12 = llvm.mlir.addressof @N : !llvm.ptr
    %13 = llvm.load %12 : !llvm.ptr -> i64
    %14 = arith.constant 1 : i32
    %16 = arith.extsi %14 : i32 to i64
    %15 = arith.addi %13, %16 : i64
    %17 = arith.constant 8 : i32
    %18 = arith.extsi %17 : i32 to i64
    %11 = func.call @calloc(%15, %18) : (i64, i64) -> !llvm.ptr
    %19 = llvm.mlir.zero : !llvm.ptr
    %20 = llvm.icmp "eq" %11, %19 : !llvm.ptr
    cf.cond_br %20, ^bb3, ^bb4
    ^bb3:
      %21 = arith.constant 1 : i32
      func.return %21 : i32
    ^bb4:
      cf.br ^bb5
    ^bb5:
    %22 = arith.constant 1 : i32
    %23 = arith.extsi %22 : 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
    %26 = arith.constant 1 : i32
    %27 = arith.extsi %26 : i32 to i64
    %28 = llvm.mlir.constant(1 : i64) : i64
    %29 = llvm.alloca %28 x i64 : (i64) -> !llvm.ptr
    llvm.store %27, %29 : i64, !llvm.ptr
    cf.br ^bb6
    ^bb6:
    %30 = llvm.load %29 : !llvm.ptr -> i64
    %31 = llvm.mlir.addressof @N : !llvm.ptr
    %32 = llvm.load %31 : !llvm.ptr -> i64
    %33 = arith.cmpi sle, %30, %32 : i64
    cf.cond_br %33, ^bb7, ^bb8
    ^bb7:
      %34 = llvm.load %25 : !llvm.ptr -> i64
      %35 = arith.constant 3 : i32
      %37 = arith.extsi %35 : i32 to i64
      %36 = arith.muli %34, %37 : i64
      %38 = arith.remsi %36, %9 : i64
      llvm.store %38, %25 : i64, !llvm.ptr
      %39 = llvm.load %25 : !llvm.ptr -> i64
      %40 = arith.constant 0 : i32
      %42 = arith.extsi %40 : i32 to i64
      %41 = arith.cmpi eq, %39, %42 : i64
      %43 = scf.if %41 -> (i1) {
        %44 = arith.constant true
        scf.yield %44 : i1
      } else {
        %45 = llvm.load %25 : !llvm.ptr -> i64
        %46 = llvm.mlir.addressof @N : !llvm.ptr
        %47 = llvm.load %46 : !llvm.ptr -> i64
        %48 = arith.cmpi sgt, %45, %47 : i64
        scf.yield %48 : i1
      }
      %49 = scf.if %43 -> (i1) {
        %50 = arith.constant true
        scf.yield %50 : i1
      } else {
        %52 = llvm.load %25 : !llvm.ptr -> i64
        %53 = llvm.getelementptr %11[%52] : (!llvm.ptr, i64) -> !llvm.ptr, i64
        %51 = llvm.load %53 : !llvm.ptr -> i64
        %54 = arith.constant 0 : i32
        %56 = arith.extsi %54 : i32 to i64
        %55 = arith.cmpi ne, %51, %56 : i64
        scf.yield %55 : i1
      }
      cf.cond_br %49, ^bb9, ^bb10
      ^bb9:
        func.call @free(%11) : (!llvm.ptr) -> ()
        %58 = llvm.mlir.addressof @str_0 : !llvm.ptr
        %59 = arith.constant 1 : i32
        %61 = arith.constant 0 : i32
        %60 = arith.subi %61, %59 : i32
        %62 = llvm.call @printf(%58, %60) vararg(!llvm.func<i32 (ptr, ...)>) : (!llvm.ptr, i32) -> i32
        %63 = arith.constant 0 : i32
        func.return %63 : i32
      ^bb10:
        cf.br ^bb11
      ^bb11:
      %64 = llvm.load %29 : !llvm.ptr -> i64
      %65 = llvm.load %25 : !llvm.ptr -> i64
      %66 = llvm.getelementptr %11[%65] : (!llvm.ptr, i64) -> !llvm.ptr, i64
      llvm.store %64, %66 : i64, !llvm.ptr
      %67 = llvm.load %29 : !llvm.ptr -> i64
      %68 = arith.constant 1 : i32
      %70 = arith.extsi %68 : i32 to i64
      %69 = arith.addi %67, %70 : i64
      llvm.store %69, %29 : i64, !llvm.ptr
      cf.br ^bb6
    ^bb8:
    %72 = llvm.mlir.addressof @N : !llvm.ptr
    %73 = llvm.load %72 : !llvm.ptr -> i64
    %74 = arith.constant 1 : i32
    %76 = arith.extsi %74 : i32 to i64
    %75 = arith.addi %73, %76 : i64
    %77 = llvm.mlir.addressof @N : !llvm.ptr
    %78 = llvm.load %77 : !llvm.ptr -> i64
    %79 = arith.constant 1 : i32
    %81 = arith.extsi %79 : i32 to i64
    %80 = arith.addi %78, %81 : i64
    %82 = arith.muli %75, %80 : i64
    %83 = arith.constant 8 : i32
    %84 = arith.extsi %83 : i32 to i64
    %71 = func.call @calloc(%82, %84) : (i64, i64) -> !llvm.ptr
    %85 = llvm.mlir.zero : !llvm.ptr
    %86 = llvm.icmp "eq" %71, %85 : !llvm.ptr
    cf.cond_br %86, ^bb12, ^bb13
    ^bb12:
      func.call @free(%11) : (!llvm.ptr) -> ()
      %88 = arith.constant 1 : i32
      func.return %88 : i32
    ^bb13:
      cf.br ^bb14
    ^bb14:
    %89 = arith.constant 2 : i32
    %90 = arith.extsi %89 : i32 to i64
    %91 = llvm.mlir.constant(1 : i64) : i64
    %92 = llvm.alloca %91 x i64 : (i64) -> !llvm.ptr
    llvm.store %90, %92 : i64, !llvm.ptr
    cf.br ^bb15
    ^bb15:
    %93 = llvm.load %92 : !llvm.ptr -> i64
    %94 = llvm.mlir.addressof @N : !llvm.ptr
    %95 = llvm.load %94 : !llvm.ptr -> i64
    %96 = arith.cmpi sle, %93, %95 : i64
    cf.cond_br %96, ^bb16, ^bb17
    ^bb16:
      %97 = llvm.load %92 : !llvm.ptr -> i64
      %98 = arith.constant 1 : i32
      %100 = arith.extsi %98 : i32 to i64
      %99 = arith.subi %97, %100 : i64
      %101 = llvm.mlir.constant(1 : i64) : i64
      %102 = llvm.alloca %101 x i64 : (i64) -> !llvm.ptr
      llvm.store %99, %102 : i64, !llvm.ptr
      cf.br ^bb18
      ^bb18:
      %103 = llvm.load %102 : !llvm.ptr -> i64
      %104 = arith.constant 1 : i32
      %106 = arith.extsi %104 : i32 to i64
      %105 = arith.cmpi sge, %103, %106 : i64
      cf.cond_br %105, ^bb19, ^bb20
      ^bb19:
        %107 = llvm.mlir.addressof @INF : !llvm.ptr
        %108 = llvm.load %107 : !llvm.ptr -> i64
        %109 = llvm.mlir.constant(1 : i64) : i64
        %110 = llvm.alloca %109 x i64 : (i64) -> !llvm.ptr
        llvm.store %108, %110 : i64, !llvm.ptr
        %111 = llvm.load %102 : !llvm.ptr -> i64
        %112 = llvm.mlir.constant(1 : i64) : i64
        %113 = llvm.alloca %112 x i64 : (i64) -> !llvm.ptr
        llvm.store %111, %113 : i64, !llvm.ptr
        cf.br ^bb21
        ^bb21:
        %114 = llvm.load %113 : !llvm.ptr -> i64
        %115 = llvm.load %92 : !llvm.ptr -> i64
        %116 = arith.cmpi slt, %114, %115 : i64
        cf.cond_br %116, ^bb22, ^bb23
        ^bb22:
          %118 = llvm.load %102 : !llvm.ptr -> i64
          %119 = llvm.mlir.addressof @N : !llvm.ptr
          %120 = llvm.load %119 : !llvm.ptr -> i64
          %121 = arith.constant 1 : i32
          %123 = arith.extsi %121 : i32 to i64
          %122 = arith.addi %120, %123 : i64
          %124 = arith.muli %118, %122 : i64
          %125 = llvm.load %113 : !llvm.ptr -> i64
          %126 = arith.addi %124, %125 : i64
          %127 = llvm.getelementptr %71[%126] : (!llvm.ptr, i64) -> !llvm.ptr, i64
          %117 = llvm.load %127 : !llvm.ptr -> i64
          %129 = llvm.load %113 : !llvm.ptr -> i64
          %130 = arith.constant 1 : i32
          %132 = arith.extsi %130 : i32 to i64
          %131 = arith.addi %129, %132 : i64
          %133 = llvm.mlir.addressof @N : !llvm.ptr
          %134 = llvm.load %133 : !llvm.ptr -> i64
          %135 = arith.constant 1 : i32
          %137 = arith.extsi %135 : i32 to i64
          %136 = arith.addi %134, %137 : i64
          %138 = arith.muli %131, %136 : i64
          %139 = llvm.load %92 : !llvm.ptr -> i64
          %140 = arith.addi %138, %139 : i64
          %141 = llvm.getelementptr %71[%140] : (!llvm.ptr, i64) -> !llvm.ptr, i64
          %128 = llvm.load %141 : !llvm.ptr -> i64
          %142 = arith.addi %117, %128 : i64
          %145 = llvm.load %113 : !llvm.ptr -> i64
          %146 = llvm.getelementptr %11[%145] : (!llvm.ptr, i64) -> !llvm.ptr, i64
          %144 = llvm.load %146 : !llvm.ptr -> i64
          %148 = llvm.load %92 : !llvm.ptr -> i64
          %149 = llvm.getelementptr %11[%148] : (!llvm.ptr, i64) -> !llvm.ptr, i64
          %147 = llvm.load %149 : !llvm.ptr -> i64
          %150 = arith.subi %144, %147 : i64
          %143 = func.call @iabs(%150) : (i64) -> i64
          %151 = arith.addi %142, %143 : i64
          %152 = llvm.load %110 : !llvm.ptr -> i64
          %153 = arith.cmpi slt, %151, %152 : i64
          cf.cond_br %153, ^bb24, ^bb25
          ^bb24:
            llvm.store %151, %110 : i64, !llvm.ptr
            cf.br ^bb26
          ^bb25:
            cf.br ^bb26
          ^bb26:
          %154 = llvm.load %113 : !llvm.ptr -> i64
          %155 = arith.constant 1 : i32
          %157 = arith.extsi %155 : i32 to i64
          %156 = arith.addi %154, %157 : i64
          llvm.store %156, %113 : i64, !llvm.ptr
          cf.br ^bb21
        ^bb23:
        %158 = llvm.load %110 : !llvm.ptr -> i64
        %159 = llvm.load %102 : !llvm.ptr -> i64
        %160 = llvm.mlir.addressof @N : !llvm.ptr
        %161 = llvm.load %160 : !llvm.ptr -> i64
        %162 = arith.constant 1 : i32
        %164 = arith.extsi %162 : i32 to i64
        %163 = arith.addi %161, %164 : i64
        %165 = arith.muli %159, %163 : i64
        %166 = llvm.load %92 : !llvm.ptr -> i64
        %167 = arith.addi %165, %166 : i64
        %168 = llvm.getelementptr %71[%167] : (!llvm.ptr, i64) -> !llvm.ptr, i64
        llvm.store %158, %168 : i64, !llvm.ptr
        %169 = llvm.load %102 : !llvm.ptr -> i64
        %170 = arith.constant 1 : i32
        %172 = arith.extsi %170 : i32 to i64
        %171 = arith.subi %169, %172 : i64
        llvm.store %171, %102 : i64, !llvm.ptr
        cf.br ^bb18
      ^bb20:
      %173 = llvm.load %92 : !llvm.ptr -> i64
      %174 = arith.constant 1 : i32
      %176 = arith.extsi %174 : i32 to i64
      %175 = arith.addi %173, %176 : i64
      llvm.store %175, %92 : i64, !llvm.ptr
      cf.br ^bb15
    ^bb17:
    %178 = arith.constant 1 : i32
    %179 = llvm.mlir.addressof @N : !llvm.ptr
    %180 = llvm.load %179 : !llvm.ptr -> i64
    %181 = arith.constant 1 : i32
    %183 = arith.extsi %181 : i32 to i64
    %182 = arith.addi %180, %183 : i64
    %185 = arith.extsi %178 : i32 to i64
    %184 = arith.muli %185, %182 : i64
    %186 = llvm.mlir.addressof @N : !llvm.ptr
    %187 = llvm.load %186 : !llvm.ptr -> i64
    %188 = arith.addi %184, %187 : i64
    %189 = llvm.getelementptr %71[%188] : (!llvm.ptr, i64) -> !llvm.ptr, i64
    %177 = llvm.load %189 : !llvm.ptr -> i64
    func.call @free(%71) : (!llvm.ptr) -> ()
    func.call @free(%11) : (!llvm.ptr) -> ()
    %192 = llvm.mlir.addressof @str_0 : !llvm.ptr
    %193 = llvm.call @printf(%192, %177) vararg(!llvm.func<i32 (ptr, ...)>) : (!llvm.ptr, i64) -> i32
    %194 = arith.constant 0 : i32
    func.return %194 : i32
  }
}