Clarkes_model_DM_vcos.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "67140e6f",
   "metadata": {
    "id": "3d79849a"
   },
   "source": [
    "[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/azad-academy/denoising-diffusion-model/blob/main/diffusion_model_demo.ipynb)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "68255ba6",
   "metadata": {
    "id": "6671430e"
   },
   "source": [
    "## Learning Channel Distribution by Diffusion Models\n",
    " \n",
    "This notebook file is a framework to learn channels by their conditional distributions: the distirubtions of the channel output for given channel inputs. To serve that, we edited a standard 1-dimensional U-Net to be conditional on the channel input. For the channel model, we used a correlated Rayleigh fading model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "300e1153",
   "metadata": {
    "executionInfo": {
     "elapsed": 1182,
     "status": "ok",
     "timestamp": 1669225000074,
     "user": {
      "displayName": "M. Kim",
      "userId": "15172492698916514875"
     },
     "user_tz": -60
    },
    "id": "e2ddcaa7"
   },
   "outputs": [],
   "source": [
    "# Import packages\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "from functools import partial\n",
    "import torch\n",
    "#import tqdm as notebook_tqdm\n",
    "from tqdm import tqdm\n",
    "\n",
    "import sys\n",
    "sys.path.append('src/')\n",
    "\n",
    "# Import modules from own libraries\n",
    "from denoising_diffusion_pytorch_1d_cond import GaussianDiffusion1D, Unet1D, Trainer1D, Dataset1D\n",
    "import channel_models\n",
    "\n",
    "# Set the device\n",
    "use_cuda = torch.cuda.is_available()\n",
    "device = torch.device(\"cuda\" if use_cuda else \"cpu\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d964b7ae",
   "metadata": {},
   "source": [
    "## Choose channel model and diffusion model type"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "71ef4514",
   "metadata": {},
   "outputs": [],
   "source": [
    "channel_model_str = 'corrRayleigh' #available options: AWGN, Rayleigh, SSPA"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bdca15fa",
   "metadata": {},
   "source": [
    "## Channel model settings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "d71ac72d-673e-4318-a3b7-9a572c78b072",
   "metadata": {},
   "outputs": [],
   "source": [
    "def SNR_to_noise(snrdb):\n",
    "    '''Transform EbNo[dB]/snr to noise power'''\n",
    "    snr = 10**(snrdb/10)\n",
    "    noise_std = 1/np.sqrt(2*snr)\n",
    "    return noise_std\n",
    "\n",
    "class Channel_Config:\n",
    "    def __init__(self, channel_model_str, blocklength, training_noise_std, device=\"cpu\", f_dopp_norm=None):\n",
    "        super().__init__()\n",
    "        self.channel_model_str = channel_model_str\n",
    "        self.blocklength = blocklength\n",
    "        self.training_noise_std = training_noise_std\n",
    "        self.device = device\n",
    "        self.f_dopp_norm = f_dopp_norm\n",
    "        self.cov = channel_models.Rayleigh_covariance_generator(n, f_dopp_norm,\n",
    "                                                 dtype_cov=torch.cfloat) if f_dopp_norm is not None else None\n",
    "\n",
    "    def channel_model_fn(self):\n",
    "        channel_model_str = self.channel_model_str\n",
    "        if channel_model_str == 'AWGN':\n",
    "            channel_model = channel_models.ch_AWGN\n",
    "        elif channel_model_str == 'Rayleigh':\n",
    "            channel_model = channel_models.ch_Rayleigh_AWGN_n\n",
    "        elif channel_model_str == 'SSPA':\n",
    "            channel_model = channel_models.ch_SSPA\n",
    "        elif channel_model_str == 'corrRayleigh':\n",
    "            assert f_dopp_norm is not None, 'The normalized doppler frequency, f_dopp_norm, is not given.'\n",
    "            cov = self.cov\n",
    "            channel_model = partial(channel_models.ch_corr_Rayleigh_by_cov, Cov_h=cov)\n",
    "        else:\n",
    "            raise Exception(\"Unrecognized channel model. Possible channel models: AWGN, Rayleigh, SSPA.\")\n",
    "        self.channel_model = channel_model\n",
    "        return channel_model\n",
    "\n",
    "    def show_cov(self):\n",
    "        fig, ax = plt.subplots(figsize=(3, 3))\n",
    "        im = ax.imshow(self.cov.real, cmap='hot', )  # , interpolation='nearest')\n",
    "        #ax.set_title(r'Covariance, $\\bar{f}_D=' + str(self.f_dopp_norm) + '$')\n",
    "        plt.colorbar(im, ax=ax, fraction=0.046)\n",
    "        plt.savefig(f'figures/cov_fD={self.f_dopp_norm:.3f}.pdf', bbox_inches='tight')\n",
    "        plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "7a28507e-c877-4cd3-905b-d913ba39f812",
   "metadata": {},
   "outputs": [],
   "source": [
    "def SNR_to_noise(snrdb):\n",
    "    '''Transform EbNo[dB]/snr to noise power'''\n",
    "    snr = 10**(snrdb/10)\n",
    "    noise_std = 1/np.sqrt(2*snr)\n",
    "    return noise_std"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "7c08fbc5-43ad-4028-849b-db62be2a9724",
   "metadata": {},
   "outputs": [],
   "source": [
    "def generate_training_seq(dataset_size, Config_ch, batch_size = 100):\n",
    "    n = Config_ch.blocklength\n",
    "    training_seq = torch.zeros(dataset_size, 2*2, n)\n",
    "    idx = 0\n",
    "    for i, batch in enumerate(torch.split(training_seq, batch_size)):\n",
    "        bs = batch.shape[0]\n",
    "        \n",
    "        training_x = torch.randn(bs, n, dtype=torch.cfloat, device=Config_ch.device)\n",
    "        channel_model = Config_ch.channel_model_fn()\n",
    "        training_y = channel_model(training_x, Config_ch.training_noise_std, Config_ch.device)\n",
    "        training_x = torch.transpose(torch.view_as_real(training_x), 1, 2)\n",
    "        training_y = torch.transpose(torch.view_as_real(training_y), 1, 2)\n",
    "        training_seq[idx:idx+bs, :, :] = torch.cat((training_y, training_x), dim=1)\n",
    "        idx += bs\n",
    "    return training_seq"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "d9448af7",
   "metadata": {
    "executionInfo": {
     "elapsed": 2418,
     "status": "ok",
     "timestamp": 1669225002490,
     "user": {
      "displayName": "M. Kim",
      "userId": "15172492698916514875"
     },
     "user_tz": -60
    },
    "id": "d3ebbf23"
   },
   "outputs": [],
   "source": [
    "# Channel Model\n",
    "n = 128  # Block length\n",
    "f_dopp_norm = 0.01\n",
    "\n",
    "TRAINING_SNR = 15 # decibel\n",
    "noise_std = SNR_to_noise(TRAINING_SNR)\n",
    "Config_ch = Channel_Config(channel_model_str, n, noise_std, device=device, f_dopp_norm=f_dopp_norm)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "de90cea4-65c4-4ba1-a761-809f257b8253",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 300x300 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "Config_ch.show_cov()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9054f6f0",
   "metadata": {},
   "source": [
    "## Deep Learning Settings - model parameters, training and testing hyperparameters, etc."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "9a44749a-3ce3-457a-8fcd-ddd92b53a1cd",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Diffusion Model \n",
    "\n",
    "model = Unet1D(\n",
    "    dim = 24, \n",
    "    dim_mults = (1, 2, 4, 8),\n",
    "    out_dim = 2, # Due to the assumption of the complex channel\n",
    "    channels = 2 * 2, # Because of feeding the condition sequence having the same size   \n",
    "    channels_out = 2 # Due to the assumption of the complex channel\n",
    ")\n",
    "\n",
    "diffusion = GaussianDiffusion1D(\n",
    "    model,\n",
    "    seq_length = n, # size of the input sequence.\n",
    "    timesteps = 100,\n",
    "    objective = 'pred_v'\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "45cf5564-db4b-4182-ac87-32add3460d62",
   "metadata": {},
   "outputs": [],
   "source": [
    "training_seq = generate_training_seq(10 ** 3, Config_ch)\n",
    "dataset = Dataset1D(training_seq) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "084dfa61-f0f6-44b9-9e60-63de3b02a19c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loading from version 1.9.5\n"
     ]
    }
   ],
   "source": [
    "trainer = Trainer1D(\n",
    "    diffusion,\n",
    "    dataset = dataset,\n",
    "    train_batch_size = 1000,\n",
    "    train_lr = 1e-5,\n",
    "    train_num_steps = 70000,         # total training steps\n",
    "    gradient_accumulate_every = 2,    # gradient accumulation steps\n",
    "    ema_decay = 0.995,                # exponential moving average decay\n",
    "    amp = True,                       # turn on mixed precision\n",
    ")\n",
    "\n",
    "trainer.load(milestone=50)\n",
    "#trainer.train()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "019d16c2-e276-4133-805b-8d160f1f222a",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "62bbe536-c23f-4a3b-b975-69da2fecc735",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "56e62e48-cf1a-49aa-8a9a-db05b65929e0",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "627f20ba-45ca-4319-a2c8-12d82cd06377",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Unet1D(\n",
       "  (init_conv): Conv1d(4, 24, kernel_size=(7,), stride=(1,), padding=(3,))\n",
       "  (time_mlp): Sequential(\n",
       "    (0): SinusoidalPosEmb()\n",
       "    (1): Linear(in_features=24, out_features=96, bias=True)\n",
       "    (2): GELU(approximate='none')\n",
       "    (3): Linear(in_features=96, out_features=96, bias=True)\n",
       "  )\n",
       "  (downs): ModuleList(\n",
       "    (0): ModuleList(\n",
       "      (0): ResnetBlock(\n",
       "        (mlp): Sequential(\n",
       "          (0): SiLU()\n",
       "          (1): Linear(in_features=96, out_features=48, bias=True)\n",
       "        )\n",
       "        (block1): Block(\n",
       "          (proj): Conv1d(24, 24, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 24, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (block2): Block(\n",
       "          (proj): Conv1d(24, 24, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 24, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (res_conv): Identity()\n",
       "      )\n",
       "      (1): ResnetBlock(\n",
       "        (mlp): Sequential(\n",
       "          (0): SiLU()\n",
       "          (1): Linear(in_features=96, out_features=48, bias=True)\n",
       "        )\n",
       "        (block1): Block(\n",
       "          (proj): Conv1d(24, 24, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 24, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (block2): Block(\n",
       "          (proj): Conv1d(24, 24, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 24, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (res_conv): Identity()\n",
       "      )\n",
       "      (2): Residual(\n",
       "        (fn): PreNorm(\n",
       "          (fn): LinearAttention(\n",
       "            (to_qkv): Conv1d(24, 384, kernel_size=(1,), stride=(1,), bias=False)\n",
       "            (to_out): Sequential(\n",
       "              (0): Conv1d(128, 24, kernel_size=(1,), stride=(1,))\n",
       "              (1): RMSNorm()\n",
       "            )\n",
       "          )\n",
       "          (norm): RMSNorm()\n",
       "        )\n",
       "      )\n",
       "      (3): Conv1d(24, 24, kernel_size=(4,), stride=(2,), padding=(1,))\n",
       "    )\n",
       "    (1): ModuleList(\n",
       "      (0): ResnetBlock(\n",
       "        (mlp): Sequential(\n",
       "          (0): SiLU()\n",
       "          (1): Linear(in_features=96, out_features=48, bias=True)\n",
       "        )\n",
       "        (block1): Block(\n",
       "          (proj): Conv1d(24, 24, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 24, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (block2): Block(\n",
       "          (proj): Conv1d(24, 24, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 24, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (res_conv): Identity()\n",
       "      )\n",
       "      (1): ResnetBlock(\n",
       "        (mlp): Sequential(\n",
       "          (0): SiLU()\n",
       "          (1): Linear(in_features=96, out_features=48, bias=True)\n",
       "        )\n",
       "        (block1): Block(\n",
       "          (proj): Conv1d(24, 24, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 24, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (block2): Block(\n",
       "          (proj): Conv1d(24, 24, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 24, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (res_conv): Identity()\n",
       "      )\n",
       "      (2): Residual(\n",
       "        (fn): PreNorm(\n",
       "          (fn): LinearAttention(\n",
       "            (to_qkv): Conv1d(24, 384, kernel_size=(1,), stride=(1,), bias=False)\n",
       "            (to_out): Sequential(\n",
       "              (0): Conv1d(128, 24, kernel_size=(1,), stride=(1,))\n",
       "              (1): RMSNorm()\n",
       "            )\n",
       "          )\n",
       "          (norm): RMSNorm()\n",
       "        )\n",
       "      )\n",
       "      (3): Conv1d(24, 48, kernel_size=(4,), stride=(2,), padding=(1,))\n",
       "    )\n",
       "    (2): ModuleList(\n",
       "      (0): ResnetBlock(\n",
       "        (mlp): Sequential(\n",
       "          (0): SiLU()\n",
       "          (1): Linear(in_features=96, out_features=96, bias=True)\n",
       "        )\n",
       "        (block1): Block(\n",
       "          (proj): Conv1d(48, 48, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 48, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (block2): Block(\n",
       "          (proj): Conv1d(48, 48, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 48, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (res_conv): Identity()\n",
       "      )\n",
       "      (1): ResnetBlock(\n",
       "        (mlp): Sequential(\n",
       "          (0): SiLU()\n",
       "          (1): Linear(in_features=96, out_features=96, bias=True)\n",
       "        )\n",
       "        (block1): Block(\n",
       "          (proj): Conv1d(48, 48, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 48, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (block2): Block(\n",
       "          (proj): Conv1d(48, 48, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 48, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (res_conv): Identity()\n",
       "      )\n",
       "      (2): Residual(\n",
       "        (fn): PreNorm(\n",
       "          (fn): LinearAttention(\n",
       "            (to_qkv): Conv1d(48, 384, kernel_size=(1,), stride=(1,), bias=False)\n",
       "            (to_out): Sequential(\n",
       "              (0): Conv1d(128, 48, kernel_size=(1,), stride=(1,))\n",
       "              (1): RMSNorm()\n",
       "            )\n",
       "          )\n",
       "          (norm): RMSNorm()\n",
       "        )\n",
       "      )\n",
       "      (3): Conv1d(48, 96, kernel_size=(4,), stride=(2,), padding=(1,))\n",
       "    )\n",
       "    (3): ModuleList(\n",
       "      (0): ResnetBlock(\n",
       "        (mlp): Sequential(\n",
       "          (0): SiLU()\n",
       "          (1): Linear(in_features=96, out_features=192, bias=True)\n",
       "        )\n",
       "        (block1): Block(\n",
       "          (proj): Conv1d(96, 96, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 96, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (block2): Block(\n",
       "          (proj): Conv1d(96, 96, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 96, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (res_conv): Identity()\n",
       "      )\n",
       "      (1): ResnetBlock(\n",
       "        (mlp): Sequential(\n",
       "          (0): SiLU()\n",
       "          (1): Linear(in_features=96, out_features=192, bias=True)\n",
       "        )\n",
       "        (block1): Block(\n",
       "          (proj): Conv1d(96, 96, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 96, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (block2): Block(\n",
       "          (proj): Conv1d(96, 96, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 96, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (res_conv): Identity()\n",
       "      )\n",
       "      (2): Residual(\n",
       "        (fn): PreNorm(\n",
       "          (fn): LinearAttention(\n",
       "            (to_qkv): Conv1d(96, 384, kernel_size=(1,), stride=(1,), bias=False)\n",
       "            (to_out): Sequential(\n",
       "              (0): Conv1d(128, 96, kernel_size=(1,), stride=(1,))\n",
       "              (1): RMSNorm()\n",
       "            )\n",
       "          )\n",
       "          (norm): RMSNorm()\n",
       "        )\n",
       "      )\n",
       "      (3): Conv1d(96, 192, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "    )\n",
       "  )\n",
       "  (ups): ModuleList(\n",
       "    (0): ModuleList(\n",
       "      (0): ResnetBlock(\n",
       "        (mlp): Sequential(\n",
       "          (0): SiLU()\n",
       "          (1): Linear(in_features=96, out_features=384, bias=True)\n",
       "        )\n",
       "        (block1): Block(\n",
       "          (proj): Conv1d(288, 192, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 192, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (block2): Block(\n",
       "          (proj): Conv1d(192, 192, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 192, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (res_conv): Conv1d(288, 192, kernel_size=(1,), stride=(1,))\n",
       "      )\n",
       "      (1): ResnetBlock(\n",
       "        (mlp): Sequential(\n",
       "          (0): SiLU()\n",
       "          (1): Linear(in_features=96, out_features=384, bias=True)\n",
       "        )\n",
       "        (block1): Block(\n",
       "          (proj): Conv1d(288, 192, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 192, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (block2): Block(\n",
       "          (proj): Conv1d(192, 192, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 192, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (res_conv): Conv1d(288, 192, kernel_size=(1,), stride=(1,))\n",
       "      )\n",
       "      (2): Residual(\n",
       "        (fn): PreNorm(\n",
       "          (fn): LinearAttention(\n",
       "            (to_qkv): Conv1d(192, 384, kernel_size=(1,), stride=(1,), bias=False)\n",
       "            (to_out): Sequential(\n",
       "              (0): Conv1d(128, 192, kernel_size=(1,), stride=(1,))\n",
       "              (1): RMSNorm()\n",
       "            )\n",
       "          )\n",
       "          (norm): RMSNorm()\n",
       "        )\n",
       "      )\n",
       "      (3): Sequential(\n",
       "        (0): Upsample(scale_factor=2.0, mode=nearest)\n",
       "        (1): Conv1d(192, 96, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "      )\n",
       "    )\n",
       "    (1): ModuleList(\n",
       "      (0): ResnetBlock(\n",
       "        (mlp): Sequential(\n",
       "          (0): SiLU()\n",
       "          (1): Linear(in_features=96, out_features=192, bias=True)\n",
       "        )\n",
       "        (block1): Block(\n",
       "          (proj): Conv1d(144, 96, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 96, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (block2): Block(\n",
       "          (proj): Conv1d(96, 96, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 96, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (res_conv): Conv1d(144, 96, kernel_size=(1,), stride=(1,))\n",
       "      )\n",
       "      (1): ResnetBlock(\n",
       "        (mlp): Sequential(\n",
       "          (0): SiLU()\n",
       "          (1): Linear(in_features=96, out_features=192, bias=True)\n",
       "        )\n",
       "        (block1): Block(\n",
       "          (proj): Conv1d(144, 96, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 96, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (block2): Block(\n",
       "          (proj): Conv1d(96, 96, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 96, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (res_conv): Conv1d(144, 96, kernel_size=(1,), stride=(1,))\n",
       "      )\n",
       "      (2): Residual(\n",
       "        (fn): PreNorm(\n",
       "          (fn): LinearAttention(\n",
       "            (to_qkv): Conv1d(96, 384, kernel_size=(1,), stride=(1,), bias=False)\n",
       "            (to_out): Sequential(\n",
       "              (0): Conv1d(128, 96, kernel_size=(1,), stride=(1,))\n",
       "              (1): RMSNorm()\n",
       "            )\n",
       "          )\n",
       "          (norm): RMSNorm()\n",
       "        )\n",
       "      )\n",
       "      (3): Sequential(\n",
       "        (0): Upsample(scale_factor=2.0, mode=nearest)\n",
       "        (1): Conv1d(96, 48, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "      )\n",
       "    )\n",
       "    (2): ModuleList(\n",
       "      (0): ResnetBlock(\n",
       "        (mlp): Sequential(\n",
       "          (0): SiLU()\n",
       "          (1): Linear(in_features=96, out_features=96, bias=True)\n",
       "        )\n",
       "        (block1): Block(\n",
       "          (proj): Conv1d(72, 48, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 48, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (block2): Block(\n",
       "          (proj): Conv1d(48, 48, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 48, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (res_conv): Conv1d(72, 48, kernel_size=(1,), stride=(1,))\n",
       "      )\n",
       "      (1): ResnetBlock(\n",
       "        (mlp): Sequential(\n",
       "          (0): SiLU()\n",
       "          (1): Linear(in_features=96, out_features=96, bias=True)\n",
       "        )\n",
       "        (block1): Block(\n",
       "          (proj): Conv1d(72, 48, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 48, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (block2): Block(\n",
       "          (proj): Conv1d(48, 48, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 48, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (res_conv): Conv1d(72, 48, kernel_size=(1,), stride=(1,))\n",
       "      )\n",
       "      (2): Residual(\n",
       "        (fn): PreNorm(\n",
       "          (fn): LinearAttention(\n",
       "            (to_qkv): Conv1d(48, 384, kernel_size=(1,), stride=(1,), bias=False)\n",
       "            (to_out): Sequential(\n",
       "              (0): Conv1d(128, 48, kernel_size=(1,), stride=(1,))\n",
       "              (1): RMSNorm()\n",
       "            )\n",
       "          )\n",
       "          (norm): RMSNorm()\n",
       "        )\n",
       "      )\n",
       "      (3): Sequential(\n",
       "        (0): Upsample(scale_factor=2.0, mode=nearest)\n",
       "        (1): Conv1d(48, 24, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "      )\n",
       "    )\n",
       "    (3): ModuleList(\n",
       "      (0): ResnetBlock(\n",
       "        (mlp): Sequential(\n",
       "          (0): SiLU()\n",
       "          (1): Linear(in_features=96, out_features=48, bias=True)\n",
       "        )\n",
       "        (block1): Block(\n",
       "          (proj): Conv1d(48, 24, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 24, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (block2): Block(\n",
       "          (proj): Conv1d(24, 24, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 24, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (res_conv): Conv1d(48, 24, kernel_size=(1,), stride=(1,))\n",
       "      )\n",
       "      (1): ResnetBlock(\n",
       "        (mlp): Sequential(\n",
       "          (0): SiLU()\n",
       "          (1): Linear(in_features=96, out_features=48, bias=True)\n",
       "        )\n",
       "        (block1): Block(\n",
       "          (proj): Conv1d(48, 24, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 24, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (block2): Block(\n",
       "          (proj): Conv1d(24, 24, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "          (norm): GroupNorm(8, 24, eps=1e-05, affine=True)\n",
       "          (act): SiLU()\n",
       "        )\n",
       "        (res_conv): Conv1d(48, 24, kernel_size=(1,), stride=(1,))\n",
       "      )\n",
       "      (2): Residual(\n",
       "        (fn): PreNorm(\n",
       "          (fn): LinearAttention(\n",
       "            (to_qkv): Conv1d(24, 384, kernel_size=(1,), stride=(1,), bias=False)\n",
       "            (to_out): Sequential(\n",
       "              (0): Conv1d(128, 24, kernel_size=(1,), stride=(1,))\n",
       "              (1): RMSNorm()\n",
       "            )\n",
       "          )\n",
       "          (norm): RMSNorm()\n",
       "        )\n",
       "      )\n",
       "      (3): Conv1d(24, 24, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "    )\n",
       "  )\n",
       "  (mid_block1): ResnetBlock(\n",
       "    (mlp): Sequential(\n",
       "      (0): SiLU()\n",
       "      (1): Linear(in_features=96, out_features=384, bias=True)\n",
       "    )\n",
       "    (block1): Block(\n",
       "      (proj): Conv1d(192, 192, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "      (norm): GroupNorm(8, 192, eps=1e-05, affine=True)\n",
       "      (act): SiLU()\n",
       "    )\n",
       "    (block2): Block(\n",
       "      (proj): Conv1d(192, 192, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "      (norm): GroupNorm(8, 192, eps=1e-05, affine=True)\n",
       "      (act): SiLU()\n",
       "    )\n",
       "    (res_conv): Identity()\n",
       "  )\n",
       "  (mid_attn): Residual(\n",
       "    (fn): PreNorm(\n",
       "      (fn): Attention(\n",
       "        (to_qkv): Conv1d(192, 384, kernel_size=(1,), stride=(1,), bias=False)\n",
       "        (to_out): Conv1d(128, 192, kernel_size=(1,), stride=(1,))\n",
       "      )\n",
       "      (norm): RMSNorm()\n",
       "    )\n",
       "  )\n",
       "  (mid_block2): ResnetBlock(\n",
       "    (mlp): Sequential(\n",
       "      (0): SiLU()\n",
       "      (1): Linear(in_features=96, out_features=384, bias=True)\n",
       "    )\n",
       "    (block1): Block(\n",
       "      (proj): Conv1d(192, 192, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "      (norm): GroupNorm(8, 192, eps=1e-05, affine=True)\n",
       "      (act): SiLU()\n",
       "    )\n",
       "    (block2): Block(\n",
       "      (proj): Conv1d(192, 192, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "      (norm): GroupNorm(8, 192, eps=1e-05, affine=True)\n",
       "      (act): SiLU()\n",
       "    )\n",
       "    (res_conv): Identity()\n",
       "  )\n",
       "  (final_res_block): ResnetBlock(\n",
       "    (mlp): Sequential(\n",
       "      (0): SiLU()\n",
       "      (1): Linear(in_features=96, out_features=48, bias=True)\n",
       "    )\n",
       "    (block1): Block(\n",
       "      (proj): Conv1d(48, 24, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "      (norm): GroupNorm(8, 24, eps=1e-05, affine=True)\n",
       "      (act): SiLU()\n",
       "    )\n",
       "    (block2): Block(\n",
       "      (proj): Conv1d(24, 24, kernel_size=(3,), stride=(1,), padding=(1,))\n",
       "      (norm): GroupNorm(8, 24, eps=1e-05, affine=True)\n",
       "      (act): SiLU()\n",
       "    )\n",
       "    (res_conv): Conv1d(48, 24, kernel_size=(1,), stride=(1,))\n",
       "  )\n",
       "  (final_conv): Conv1d(24, 2, kernel_size=(1,), stride=(1,))\n",
       ")"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "621642ac",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "# of parameters (diffusion model): 2338418\n"
     ]
    }
   ],
   "source": [
    "total_params_gen = sum(p.numel() for p in model.parameters() if p.requires_grad)\n",
    "print('# of parameters (diffusion model): ' + str(total_params_gen))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "0e1abffd-daf7-42fe-9fec-cedc95b81c49",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loading from version 1.9.5\n",
      "The trainer is loaded with the model of the 50-th milestone.\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "73be0fa970ef4ed6bd7fdbec6255de1e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "sampling loop time step:   0%|          | 0/100 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d1e8a392d3e14f48b0f40cb9b8a9d6ad",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "sampling loop time step:   0%|          | 0/100 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2c299d603631482a93da886da34557bd",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "sampling loop time step:   0%|          | 0/100 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "437e45b48b3b45fb925ac7969d460020",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "sampling loop time step:   0%|          | 0/100 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "65852ecfb34c4c239803e8eb05e55a00",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "sampling loop time step:   0%|          | 0/100 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3dc5d8f44013404ba6a53a1d2755e8d4",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "sampling loop time step:   0%|          | 0/100 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "acbd58b739fa4f229c15a39c07b7c949",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "sampling loop time step:   0%|          | 0/100 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "66d767b10fa44fb1bc009b1bf78ba381",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "sampling loop time step:   0%|          | 0/100 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f8956596be4648b5814b36bec17da8bf",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "sampling loop time step:   0%|          | 0/100 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "90ed2501d42d4ccd829f2a8a68cf02ac",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "sampling loop time step:   0%|          | 0/100 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "def compare_empirical_cov(trainer, Config_ch, num_samples = 10 **3, bs=100, milestone=None):\n",
    "    # y = Hx = h, when H = diag(h) and x = (1,1,1,...,1,1)^T. \n",
    "    ## For such channel input, generative network can generate the channel state information h.\n",
    "\n",
    "    ## Load the model\n",
    "    if milestone is not None:\n",
    "        trainer.load(milestone=milestone)\n",
    "        print(f\"The trainer is loaded with the model of the {milestone}-th milestone.\")\n",
    "    \n",
    "    ## Sample the channel coefficients by using the identity channel input. \n",
    "    \n",
    "    condition_ones = torch.ones([bs, n], dtype=torch.cfloat, device=device)\n",
    "    condition_ones = torch.transpose(torch.view_as_real(condition_ones), 1,2).contiguous()\n",
    "    h_gen = torch.ones([num_samples // bs * bs, n], dtype=torch.cfloat)\n",
    "    for i in range(num_samples//bs):\n",
    "        generated_h = diffusion.sample(batch_size = bs, condition=condition_ones)\n",
    "        generated_h_t = torch.transpose(generated_h, 1,2)\n",
    "        generated_h_t = generated_h_t.contiguous() # After taking transpose, it's necessary. \n",
    "        \n",
    "        h_gen[i*bs:(i+1)*bs,:] = torch.view_as_complex(generated_h_t)\n",
    "    \n",
    "    cov_emp = torch.cov(h_gen.T).detach().cpu().numpy()\n",
    "    idx = np.arange(n)    \n",
    "    cov_emp[idx, idx] -= 2*noise_std ** 2\n",
    "    \n",
    "    fig_cov_comparison, [ax1, ax2] = plt.subplots(1,2)\n",
    "    im1 = ax1.imshow(Config_ch.cov.real, cmap='hot') # , interpolation='nearest')\n",
    "    ax1.set_title('defined cov.')\n",
    "    plt.colorbar(im1, ax=ax1, fraction=0.046, pad=0.04)\n",
    "    im2 = ax2.imshow(cov_emp.real, cmap='hot')#, interpolation='nearest')\n",
    "    ax2.set_title('emp. cov.')\n",
    "    plt.colorbar(im2, ax=ax2, fraction=0.046, pad=0.04)\n",
    "    fig_cov_comparison.suptitle(\n",
    "        r'$\\bar{f}_{D}$'+f'={Config_ch.f_dopp_norm}, milestone={milestone}, # of Samples={num_samples}', y=0.75)\n",
    "    fig_cov_comparison.tight_layout()\n",
    "    plt.savefig(f'figures/emp_cov_DM_vcos_fD_{f_dopp_norm:.3f}_ns{num_samples}_ms{milestone}.pdf')\n",
    "    plt.show()\n",
    "    return h_gen\n",
    "\n",
    "generated_h = compare_empirical_cov(trainer, Config_ch, num_samples = 10 **5, milestone=50)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "b250d4d6-571d-4d44-ba8d-7aabb5fabdd3",
   "metadata": {},
   "outputs": [],
   "source": [
    "torch.save(generated_h, 'var/generated_h_vcos_ms{milestone}.pt')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "4f39dc86-9064-4265-9545-03eba383ccd5",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 300x300 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "cov_emp = torch.cov(generated_h.T).detach().cpu().numpy()\n",
    "idx = np.arange(n)    \n",
    "cov_emp[idx, idx] -= 2*noise_std ** 2\n",
    "\n",
    "fig_cov_comparison, ax2 = plt.subplots(figsize=(3,3))\n",
    "im2 = ax2.imshow(cov_emp.real, cmap='hot')#, interpolation='nearest')\n",
    "#ax2.set_title('emp. cov.')\n",
    "plt.colorbar(im2, ax=ax2, fraction=0.046, pad=0.04)\n",
    "#fig_cov_comparison.suptitle(\n",
    "#    r'$\\bar{f}_{D}$'+f'={Config_ch.f_dopp_norm}, milestone={milestone}, # of Samples={num_samples}', y=0.75)\n",
    "fig_cov_comparison.tight_layout()\n",
    "plt.savefig(f'figures/emp_cov_DM_vcos_fD_{f_dopp_norm:.3f}_ns{1000}_ms{50}.pdf', bbox_inches='tight')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "36423857-b733-4393-bcb9-10c99580e802",
   "metadata": {},
   "outputs": [],
   "source": [
    "num_samples = 10**5\n",
    "bs = 100\n",
    "\n",
    "# Compute a matrix A such that A*A^H = Cov.\n",
    "        \n",
    "# Compute the Cholesky decomposition.\n",
    "A = torch.linalg.cholesky(Config_ch.cov, upper=False)  # torch.cholesky(Cov, upper=False)\n",
    "# print(torch.mm(A, A.conj().T))\n",
    "# print(Cov_h)\n",
    "A = A[None, :, :].repeat(bs, 1, 1).to(device)\n",
    "h_t = torch.zeros(num_samples//bs * bs, 2*n)\n",
    "\n",
    "for i in range(num_samples // bs):\n",
    "    # generate normally distributed matrices\n",
    "    h = torch.normal(mean=0., std=1., size=(bs, n), device=device, dtype=torch.cfloat)\n",
    "    h = h[:, :, None]\n",
    "    # Correlated channel matrices h = Ah,  A[n,n] h[n, 1]\n",
    "    h = torch.bmm(A, h)\n",
    "    h = h.view(bs, n)\n",
    "    h_t[bs*i:bs*(i+1),:] = torch.view_as_real(h).view(-1,2*n).contiguous()\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "0b0317c0-aec6-4dca-9b11-9b1be3d92cbc",
   "metadata": {},
   "outputs": [],
   "source": [
    "def _sort_rows_tf(matrix, num_rows):\n",
    "  \"\"\"Sort matrix rows by the last column.\n",
    "\n",
    "  Args:\n",
    "      matrix: a matrix of values (row,col).\n",
    "      num_rows: (int) number of sorted rows to return from the matrix.\n",
    "\n",
    "  Returns:\n",
    "      Tensor (num_rows, col) of the sorted matrix top K rows.\n",
    "  \"\"\"\n",
    "  tmatrix = tf.transpose(a=matrix, perm=[1, 0])\n",
    "  sorted_tmatrix = tf.nn.top_k(tmatrix, num_rows)[0]\n",
    "  return tf.transpose(a=sorted_tmatrix, perm=[1, 0])\n",
    "\n",
    "def _sliced_wasserstein_tf(a, b, random_sampling_count, random_projection_dim):\n",
    "  \"\"\"Compute the approximate sliced Wasserstein distance.\n",
    "\n",
    "  Args:\n",
    "      a: (matrix) Distribution \"a\" of samples (row, col).\n",
    "      b: (matrix) Distribution \"b\" of samples (row, col).\n",
    "      random_sampling_count: (int) Number of random projections to average.\n",
    "      random_projection_dim: (int) Dimension of the random projection space.\n",
    "\n",
    "  Returns:\n",
    "      Float containing the approximate distance between \"a\" and \"b\".\n",
    "  \"\"\"\n",
    "  s = tf.shape(input=a)\n",
    "  means = []\n",
    "  for _ in range(random_sampling_count):\n",
    "    # Random projection matrix.\n",
    "    proj = tf.random.normal([tf.shape(input=a)[1], random_projection_dim])\n",
    "    proj *= tf.math.rsqrt(\n",
    "        tf.reduce_sum(input_tensor=tf.square(proj), axis=0, keepdims=True))\n",
    "    # Project both distributions and sort them.\n",
    "    proj_a = tf.matmul(a, proj)\n",
    "    proj_b = tf.matmul(b, proj)\n",
    "    proj_a = _sort_rows(proj_a, s[0])\n",
    "    proj_b = _sort_rows(proj_b, s[0])\n",
    "    # Pairwise Wasserstein distance.\n",
    "    wdist = tf.reduce_mean(input_tensor=tf.abs(proj_a - proj_b))\n",
    "    means.append(wdist)\n",
    "  return tf.reduce_mean(input_tensor=means)\n",
    "\n",
    "\n",
    "\n",
    "def _sort_rows(matrix, num_rows):\n",
    "  \"\"\"Sort matrix rows by the last column.\n",
    "\n",
    "  Args:\n",
    "      matrix: a matrix of values (row,col).\n",
    "      num_rows: (int) number of sorted rows to return from the matrix.\n",
    "\n",
    "  Returns:\n",
    "      Tensor (num_rows, col) of the sorted matrix top K rows.\n",
    "  \"\"\"\n",
    "  tmatrix = torch.transpose(matrix, 1, 0) \n",
    "  sorted_matrix = torch.topk(tmatrix, k=num_rows)[0] #tf.nn.top_k(tmatrix, num_rows)[0]\n",
    "  return torch.transpose(sorted_matrix, 1, 0) \n",
    "\n",
    "def _sliced_wasserstein(a, b, random_sampling_count, random_projection_dim=1):\n",
    "  \"\"\"Compute the approximate sliced Wasserstein distance.\n",
    "\n",
    "  Args:\n",
    "      a: (matrix) Distribution \"a\" of samples (row, col).\n",
    "      b: (matrix) Distribution \"b\" of samples (row, col).\n",
    "      random_sampling_count: (int) Number of random projections to average.\n",
    "      random_projection_dim: (int) Dimension of the random projection space.\n",
    "\n",
    "  Returns:\n",
    "      Float containing the approximate distance between \"a\" and \"b\".\n",
    "  \"\"\"\n",
    "  s = a.shape #tf.shape(input=a)\n",
    "  means = []\n",
    "  for _ in range(random_sampling_count):\n",
    "    # Random projection matrix.\n",
    "    proj = torch.randn(s[1], random_projection_dim) \n",
    "    proj *= (torch.sum(proj ** 2, dim=0)) ** (-0.5) # normalizing the projection \n",
    "    # Project both distributions and sort them.\n",
    "    proj_a = torch.matmul(a, proj) \n",
    "    proj_b = torch.matmul(b, proj) \n",
    "    proj_a = _sort_rows(proj_a, s[0])\n",
    "    proj_b = _sort_rows(proj_b, s[0])\n",
    "    # Pairwise Wasserstein distance.\n",
    "    wdist = torch.mean( torch.abs(proj_a - proj_b) ) \n",
    "    means.append(wdist)\n",
    "  return np.mean(means) \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "ca6d911c-8765-41e8-92d3-20e312c091bd",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.24970737\n"
     ]
    }
   ],
   "source": [
    "generated_h_t = torch.view_as_real(generated_h).view(-1,2*n)\n",
    "wasserstein_distance = _sliced_wasserstein(generated_h_t, h_t, 128)\n",
    "print(wasserstein_distance)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "c65956d9-5663-45de-a79f-b8d7edb4cde0",
   "metadata": {},
   "outputs": [],
   "source": [
    "torch.save(generated_h_t, 'samples/DM_vcos_generated_h.pt')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "f333800a-df3f-4823-9b5b-a67d956eff2d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([100000, 256])"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "h_t.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4ed14231-a8f6-4ce1-bcc3-0262b5a03547",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "provenance": [
    {
     "file_id": "https://github.com/muahkim/ChGen-CondDiffusion/blob/main/E2E_framework_Rayleigh_AWGN-M16N7.ipynb",
     "timestamp": 1666621806973
    }
   ]
  },
  "gpuClass": "standard",
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}