#include "case.h"

#include "ccv_case.h"

#include "ccv_nnc_case.h"

#include <ccv.h>

#include <nnc/ccv_nnc.h>

#include <nnc/ccv_nnc_easy.h>

#include "3rdparty/dsfmt/dSFMT.h"

TEST_SETUP()

{

  ccv_nnc_init();

}

TEST_CASE("random uniform distribution")

{

  ccv_nnc_symbolic_graph_t* symbolic_graph = ccv_nnc_symbolic_graph_new();

  const ccv_nnc_tensor_symbol_t x = ccv_nnc_tensor_symbol_new(symbolic_graph, CPU_TENSOR_NHWC(32F, 100000), "x");

  ccv_nnc_graph_exec_symbol_new(symbolic_graph, CMD_RANDOM_UNIFORM_FORWARD(-8, 4), TENSOR_SYMBOL_LIST(), TENSOR_SYMBOL_LIST(x), "random uniform");

  ccv_nnc_graph_exec_symbol_autogen(symbolic_graph, 0, 0, CCV_NNC_AUTOGEN_ALL_EXECS | CCV_NNC_AUTOGEN_SOURCES_AND_DESTINATIONS);

  SYMBOLIC_GRAPH_GEN(symbolic_graph, CCV_NNC_LONG_DOT_GRAPH);

  ccv_nnc_graph_t* graph = 0;

  ccv_nnc_tensor_arena_t* tensor_arena = 0;

  ccv_nnc_graph_exec_arena_t* graph_exec_arena = 0;

  ccv_nnc_symbolic_graph_compile(symbolic_graph, ccv_nnc_default_compile_params, 0, 0, 0, 0, SYMBOLIC_GRAPH_SOURCES(symbolic_graph), SYMBOLIC_GRAPH_DESTINATIONS(symbolic_graph), &graph, &tensor_arena, &graph_exec_arena);

  GRAPH_GEN(graph, CCV_NNC_LONG_DOT_GRAPH);

  ccv_nnc_graph_run(graph, 0, TRAVERSE_FULL, 0, 0);

  ccv_nnc_tensor_t* const x_tensor = ccv_nnc_tensor_from_symbol(tensor_arena, x);

  int i;

  int h[4 + 8] = {};

  for (i = 0; i < 100000; 
i++100k
)

  {

    REQUIRE(x_tensor->data.f32[i] > -8 - 1e-5, "it must be bigger than lower bound");

    REQUIRE(x_tensor->data.f32[i] < 4 + 1e-5, "and smaller than upper bound");

    int b = (int)roundf(x_tensor->data.f32[i] - 0.5) + 8;

    b = ccv_max(ccv_min(b, 11), 0);

    ++h[b];

  }

  const int count = (int)roundf(100000. / (4 + 8));

  for (i = 0; i < 12; 
i++12
)

    { REQUIRE(h[i] >= count - 1000 && h[i] <= count + 1000, "uniform distribution"); }

  ccv_nnc_graph_free(graph);

  ccv_nnc_tensor_arena_free(tensor_arena);

  ccv_nnc_graph_exec_arena_free(graph_exec_arena);

  ccv_nnc_symbolic_graph_free(symbolic_graph);

}

TEST_CASE("random normal distribution")

{

  ccv_nnc_symbolic_graph_t* symbolic_graph = ccv_nnc_symbolic_graph_new();

  const ccv_nnc_tensor_symbol_t x = ccv_nnc_tensor_symbol_new(symbolic_graph, CPU_TENSOR_NHWC(32F, 100001), "x");

  ccv_nnc_graph_exec_symbol_new(symbolic_graph, CMD_RANDOM_NORMAL_FORWARD(2, 1), TENSOR_SYMBOL_LIST(), TENSOR_SYMBOL_LIST(x), "random uniform");

  ccv_nnc_graph_exec_symbol_autogen(symbolic_graph, 0, 0, CCV_NNC_AUTOGEN_ALL_EXECS | CCV_NNC_AUTOGEN_SOURCES_AND_DESTINATIONS);

  SYMBOLIC_GRAPH_GEN(symbolic_graph, CCV_NNC_LONG_DOT_GRAPH);

  ccv_nnc_graph_t* graph = 0;

  ccv_nnc_tensor_arena_t* tensor_arena = 0;

  ccv_nnc_graph_exec_arena_t* graph_exec_arena = 0;

  ccv_nnc_symbolic_graph_compile(symbolic_graph, ccv_nnc_default_compile_params, 0, 0, 0, 0, SYMBOLIC_GRAPH_SOURCES(symbolic_graph), SYMBOLIC_GRAPH_DESTINATIONS(symbolic_graph), &graph, &tensor_arena, &graph_exec_arena);

  GRAPH_GEN(graph, CCV_NNC_LONG_DOT_GRAPH);

  ccv_nnc_graph_run(graph, 0, TRAVERSE_FULL, 0, 0);

  ccv_nnc_tensor_t* const x_tensor = ccv_nnc_tensor_from_symbol(tensor_arena, x);

  int i;

  double mean = 0;

  for (i = 0; i < 100001; 
i++100k
)

    mean += x_tensor->data.f32[i];

  mean = mean / 100001.0;

  double std = 0;

  for (i = 0; i < 100001; 
i++100k
)

    std += (x_tensor->data.f32[i] - mean) * (x_tensor->data.f32[i] - mean);

  std = sqrt(std / 100001.0);

  REQUIRE_EQ_WITH_TOLERANCE(std, 2, 2e-2, "std should be 2");

  REQUIRE_EQ_WITH_TOLERANCE(mean, 1, 2e-2, "mean should be 1");

  ccv_nnc_graph_free(graph);

  ccv_nnc_tensor_arena_free(tensor_arena);

  ccv_nnc_graph_exec_arena_free(graph_exec_arena);

  ccv_nnc_symbolic_graph_free(symbolic_graph);

}

#include "case_main.h"