Training-BinsFromMoments.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "a316912f-5b10-4dfa-83db-2d9e8197c718",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib\n",
    "from matplotlib import cm\n",
    "\n",
    "import zarr\n",
    "import xarray as xr\n",
    "import os\n",
    "from sklearn import preprocessing\n",
    "\n",
    "import torch\n",
    "from torch.utils.data import Dataset, DataLoader\n",
    "from torch.autograd import Variable\n",
    "import torch.utils.data as data_utils\n",
    "from torch.nn import Linear, Conv1d, ConvTranspose1d\n",
    "\n",
    "from torch.nn import Sequential as Seq, Linear as Lin, ReLU, Sigmoid, ConstantPad1d\n",
    "from torch.functional import F\n",
    "\n",
    "from torch.optim.lr_scheduler import OneCycleLR, ReduceLROnPlateau\n",
    "from tqdm import tqdm\n",
    "\n",
    "from tqdm.notebook import tqdm\n",
    "import json"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "a8e64ee0-abb0-49ee-8a4d-82eced7b78e2",
   "metadata": {},
   "outputs": [],
   "source": [
    "from models import DriverDatasets, MicroAutoEncoder, CNNEncoderVAE,CNNDecoder"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "47dea496-06fe-4b16-bcc5-72bf84cd515a",
   "metadata": {},
   "source": [
    "## Load the Datasets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "3c575fb5-46d9-4515-ab5e-5b0c7b93b680",
   "metadata": {},
   "outputs": [],
   "source": [
    "trainloader=torch.load('/home/jovyan/neural-dsds/neural-dsds/data/trainloader_onecat.pth')\n",
    "valloader=torch.load('/home/jovyan/neural-dsds/neural-dsds/data/valloader_onecat.pth')\n",
    "testloader=torch.load('/home/jovyan/neural-dsds/neural-dsds/data/testloader_onecat.pth')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "e9564a92-eeaf-4acc-a2a4-71412d65a7d0",
   "metadata": {},
   "outputs": [],
   "source": [
    "class DecoderBinFromMoments(torch.nn.Module):\n",
    "    def __init__(self,n_bins=35,n_moms=4,n_hidden=50):\n",
    "        super(DecoderBinFromMoments, self).__init__()\n",
    "\n",
    "        self.decoder = CNNDecoder(n_bins=n_bins,n_latent=n_moms)\n",
    "    \n",
    "    def forward(self,momt0):\n",
    "        \n",
    "        bs = momt0.shape[0]\n",
    "\n",
    "        Rt0 = self.decoder(momt0[:,:,1])\n",
    "        \n",
    "        return Rt0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "8e86ce49-c4a1-415d-bd27-22ab5826cfd9",
   "metadata": {},
   "outputs": [],
   "source": [
    "def recon_loss(recon_x,x):\n",
    "    mseloss = torch.nn.MSELoss()\n",
    "    mse = mseloss(recon_x,x)\n",
    "\n",
    "    return mse"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "780336db-8333-4bcb-811c-485b0672bec7",
   "metadata": {},
   "outputs": [],
   "source": [
    "a0 = 1 #e-3\n",
    "b = 0\n",
    "\n",
    "def train(model,dataloader,criterion,device,epoch):\n",
    "    model.train()\n",
    "    \n",
    "    for data in dataloader:\n",
    "        M,dMdt,bin0,bin1coal,bin1condevap,binmag,M1,dM1dt,M2,dM2dt=data\n",
    "        bin1 = bin1coal #bin1condevap #\n",
    "\n",
    "        M = M.to(device)\n",
    "        bin0 = bin0.to(device)\n",
    "        binmag = binmag.to(device)\n",
    "        bs = bin1.size(0)\n",
    "        #with torch.autograd.set_detect_anomaly(True):\n",
    "        optimizer.zero_grad()\n",
    "\n",
    "        normRt0 = model(torch.Tensor(M).float())\n",
    "\n",
    "        normbin0 = bin0.float()*torch.broadcast_to(binmag[:,0:2].float().unsqueeze(dim=2),(bs,2,35))\n",
    "            \n",
    "            \n",
    "        Rloss0 = recon_loss(normRt0,bin0.float()) \n",
    "\n",
    "        loss = Rloss0\n",
    "\n",
    "        loss.backward()\n",
    "\n",
    "        optimizer.step()\n",
    "        \n",
    "    return model\n",
    "def test(model,dataloader,criterion,device,val=False):\n",
    "    model.eval()\n",
    "    correct = 0\n",
    "    R0all = 0\n",
    "    R1all = 0\n",
    "    Lall = 0\n",
    "    losstot = 0\n",
    "    recon = 0\n",
    "\n",
    "    with torch.no_grad():\n",
    "        for data in dataloader:\n",
    "            M,dMdt,bin0,bin1coal,bin1condevap,binmag,M1,dM1dt,M2,dM2dt=data\n",
    "             \n",
    "            M = M.to(device)\n",
    "            bin0 = bin0.to(device)\n",
    "            binmag = binmag.to(device)\n",
    "            bs = bin0.size(0)\n",
    "            #with torch.autograd.set_detect_anomaly(True):\n",
    "            optimizer.zero_grad()\n",
    "\n",
    "            normRt0 = model(torch.Tensor(M).float())\n",
    "\n",
    "            normbin0 = bin0.float()*torch.broadcast_to(binmag[:,0:2].float().unsqueeze(dim=2),(bs,2,35))\n",
    "            \n",
    "            Rloss0 = recon_loss(normRt0,bin0.float()) \n",
    "\n",
    "            lossall = Rloss0\n",
    "            \n",
    "            nn = len(dataloader.dataset)\n",
    "\n",
    "            losstot = losstot+lossall\n",
    "        if val==True:\n",
    "            sched.step(losstot/nn)\n",
    "\n",
    "        return losstot/nn\n",
    "\n",
    "def inference(model,dataloader,device,nlatent,singlecat=True):\n",
    "    model.eval()\n",
    "    dataset = dataloader.dataset\n",
    "\n",
    "    ntargets = nlatent+2\n",
    "    nmoms = 4\n",
    "    nrates = 3\n",
    "    bs = 500\n",
    "    print(len(dataset))\n",
    "    \n",
    "    pred = np.zeros((len(dataset),ntargets,nrates))\n",
    "    real = np.zeros((len(dataset),ntargets,nrates))\n",
    "    \n",
    "    #[500, 3, 3, 2]) torch.Size([500, 3, 3, 2] \n",
    "    if singlecat:\n",
    "        realM = np.zeros((len(dataset),nmoms,nrates))\n",
    "        realdMdt = np.zeros((len(dataset),nmoms,nrates))\n",
    "    else:\n",
    "        realM = np.zeros((len(dataset),3,3,2))\n",
    "        realdMdt = np.zeros((len(dataset),3,3,2))    \n",
    "    \n",
    "\n",
    "    realbin0 = np.zeros((len(dataset),2,35))\n",
    "    predbin0 = np.zeros((len(dataset),2,35))\n",
    "\n",
    "    jj = 0\n",
    "    i = 0\n",
    "\n",
    "    for data in dataloader:\n",
    "        M,dMdt,bin0,bin1coal,bin1condevap,binmag,M1,dM1dt,M2,dM2dt=data\n",
    "        M = M.to(device)\n",
    "        normRt0 = model(torch.Tensor(M).float())        \n",
    "        \n",
    "        bs = normRt0.size(0)\n",
    "        normbin0 = bin0.float()*torch.broadcast_to(binmag[:,0:2].float().unsqueeze(dim=2),(bs,2,35))\n",
    "        \n",
    "        \n",
    "        if normRt0 is not(None):\n",
    "            if singlecat:\n",
    "                realM[jj:jj+bs,:,:] = M1.cpu().detach().numpy().reshape(bs,nmoms,nrates)\n",
    "                realdMdt[jj:jj+bs,:,:] = dM1dt.cpu().detach().numpy().reshape(bs,nmoms,nrates)\n",
    "            else:\n",
    "                realM[jj:jj+bs,:,:] = M2.cpu().detach().numpy().reshape(bs,3,3,2)\n",
    "                realdMdt[jj:jj+bs,:,:] = dM2dt.cpu().detach().numpy().reshape(bs,3,3,2)\n",
    "\n",
    "            realbin0[jj:jj+bs,:] = normbin0.cpu().detach().numpy().reshape(bs,2,35)\n",
    "            predbin0[jj:jj+bs,:] = normRt0.cpu().detach().numpy().reshape(bs,2,35)\n",
    "\n",
    "            \n",
    "            jj=jj+bs\n",
    "            i = i+1\n",
    "        else:\n",
    "            break\n",
    "\n",
    "    return realM,realdMdt,realbin0,predbin0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "d84c3232-4a86-49bd-9f59-bba011a5193d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training model n =  0\n",
      "Training model n =  1\n",
      "Training model n =  2\n",
      "Training model n =  3\n",
      "Training model n =  4\n",
      "Training model n =  5\n",
      "Training model n =  6\n",
      "Training model n =  7\n",
      "Training model n =  8\n",
      "Training model n =  9\n"
     ]
    }
   ],
   "source": [
    "n_rep = 10\n",
    "\n",
    "for j in range(0,n_rep):\n",
    "    print(\"Training model n = \",str(j))\n",
    "    init_lr = 1e-3\n",
    "    weight_decay = 1e-3\n",
    "    total_epochs = 1000\n",
    "\n",
    "    # Learning rate strings so its in the correct format for the filenames\n",
    "    lrval='{:3.0e}'.format(init_lr)[0]\n",
    "    lrexp = '{:3.0e}'.format(init_lr)[4]\n",
    "\n",
    "    model = DecoderBinFromMoments(n_bins=35,n_moms=4)\n",
    "\n",
    "\n",
    "    # Check if there is a gpu\n",
    "    device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\n",
    "    #print(device)\n",
    "    #print(model)\n",
    "    model = model.to(device)\n",
    "\n",
    "    optimizer = torch.optim.AdamW(model.parameters(), lr=init_lr, weight_decay=weight_decay)\n",
    "\n",
    "    sched =  ReduceLROnPlateau(optimizer, 'min')\n",
    "    criterion = torch.nn.MSELoss()\n",
    "    \n",
    "    val_loss_all = []\n",
    "    train_loss_all = []\n",
    "    \n",
    "    e0 = 0\n",
    "    epochs = 500\n",
    "    best_epoch = 0\n",
    "    \n",
    "    min_loss = 1e10\n",
    "\n",
    "    for epoch in tqdm(range(e0,epochs)):\n",
    "        mod = train(model,trainloader,criterion,device,epoch)\n",
    "        train_loss = test(model,trainloader,criterion,device)\n",
    "        val_loss = test(model,valloader,criterion,device,val=True)\n",
    "\n",
    "        train_loss_all.append(train_loss.cpu())\n",
    "        val_loss_all.append(val_loss.cpu())\n",
    "\n",
    "        if epoch%50==0:\n",
    "            print(f'{epoch:03d}, Train: {train_loss:.8e}, Val.: {val_loss:.8e}')\n",
    "            \n",
    "        if val_loss<min_loss:\n",
    "            bestmodel = model\n",
    "            min_loss = val_loss\n",
    "            best_epoch = epoch\n",
    "    \n",
    "    tmp_trainloss = \"MomentDistsModels/OneCatMomentDecoder_CNN_n{}_mse_{}epochs_losses_sc.npz\".format(j,epochs)\n",
    "    tmp_valloss = \"MomentDistsModels/OneCatMomentDecoder_CNN_n{}_mse_{}epochs_valloss_sc.npz\".format(j,epochs)\n",
    "    mod_weights = \"MomentDistsModels/OneCatMomentDecoder_CNN_n{}_mse_{}epochs_sc\".format(j,epochs)\n",
    "    \n",
    "    print(tmp_trainloss)\n",
    "    print(tmp_valloss)\n",
    "    print(mod_weights)\n",
    "    print(best_epoch)\n",
    "    \n",
    "    torch.save(bestmodel.state_dict(),mod_weights)\n",
    "    \n",
    "    train_loss_np = np.array(train_loss_all)\n",
    "    val_loss_np = np.array(val_loss_all)\n",
    "\n",
    "    np.savez(tmp_trainloss,trainl=train_loss_np,vall=val_loss_np)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "3bf99d24-0baf-4910-817f-be0f74325e27",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 800x650 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 640x480 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "trainloss = []\n",
    "valloss = []\n",
    "n_rep = 10\n",
    "\n",
    "fig = plt.figure(figsize=(8,6.5))\n",
    "norm = matplotlib.colors.Normalize(vmin=2.0, vmax=20.0)\n",
    "cols = cm.jet(norm(np.arange(0,n_rep)),bytes=True)/256\n",
    "\n",
    "for j in range(0,n_rep):\n",
    "    epochs = 500\n",
    "    fnloss = 'MomentDistsModels/OneCatMomentDecoder_CNN_n{}_mse_{}epochs_losses_sc.npz'.format(j,epochs)\n",
    "    \n",
    "    loss = np.load(fnloss)\n",
    "    trainloss.append(loss['trainl'])\n",
    "    valloss.append(loss['vall'])\n",
    "    \n",
    "    plt.plot(trainloss[j],label=r\"Train, ($N$ = \"+str(j)+\")\",c = tuple(cols[j]))\n",
    "    plt.plot(valloss[j],label=r\"Val., ($N$ = \"+str(j)+\")\", linestyle='dashed',c = tuple(cols[j]))\n",
    "\n",
    "plt.legend(loc='upper right',ncol=2,frameon=False)\n",
    "plt.yscale('log')\n",
    "plt.ylabel(\"MSE\")\n",
    "plt.xlabel(\"Epochs\")\n",
    "plt.show()\n",
    "plt.tight_layout()\n",
    "#plt.savefig(\"MSELoss_OneMoments_Decoder.png\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "0c853e29-a966-4b93-93e7-37c1d364bed6",
   "metadata": {},
   "outputs": [],
   "source": [
    "from scipy.stats import wasserstein_distance\n",
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "ec53594f-4622-4c2d-8e42-f869b4711cdd",
   "metadata": {},
   "outputs": [],
   "source": [
    "def kl_divergence(p, q):\n",
    "    q = q+1e-14\n",
    "    p = p+1e-14\n",
    "    return np.sum(np.where(p != 0, p * np.log(p / q), 0))\n",
    "def mse(p, q):\n",
    "\n",
    "    return np.sum((p-q)**2)/len(p)\n",
    "\n",
    "def wasserstein(a,b):\n",
    "    a = a/(np.sum(a,keepdims=True)+1e-14)\n",
    "    b = b/(np.sum(b,keepdims=True)+1e-14)\n",
    "    \n",
    "    cdf_a = np.cumsum(a)\n",
    "    cdf_b = np.cumsum(b)\n",
    "\n",
    "    return wasserstein_distance(cdf_a,cdf_b)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "38665f6b-59ce-493b-b9bb-e199ef228d8c",
   "metadata": {},
   "outputs": [],
   "source": [
    "def calcwasserstein(bina,binb):\n",
    "    bina = bina/(np.sum(bina,axis=1,keepdims=True)+1e-14)\n",
    "    binb = binb/(np.sum(binb,axis=1,keepdims=True)+1e-14)\n",
    "    \n",
    "    cdf_a = np.cumsum(bina,axis=1)\n",
    "    cdf_b = np.cumsum(binb,axis=1)\n",
    "    \n",
    "    dist = np.zeros(cdf_a.shape[0])\n",
    "    distkl = np.zeros(cdf_a.shape[0])\n",
    "    distmse = np.zeros(cdf_a.shape[0]) \n",
    "    for i in range(0,cdf_a.shape[0]):\n",
    "        dist[i]=wasserstein_distance(cdf_a[i,:],cdf_b[i,:])\n",
    "        distkl[i]=kl_divergence(bina[i,:],binb[i,:])\n",
    "        distmse[i]=mse(bina[i,:],binb[i,:])\n",
    "    return dist,distkl,distmse"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "8f934f71-804b-4a00-81b5-d33956021445",
   "metadata": {},
   "outputs": [],
   "source": [
    "epochs=500"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "7f704de1-8687-4bc1-bba0-2d5675ced386",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training model n =  0\n",
      "26214\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/ipykernel_6238/1545530616.py:46: RuntimeWarning: invalid value encountered in log\n",
      "  d = {r\"Wasserstein Dist. ($l_{1}$)\":np.log(wdistances[j,:,0]),'KL-Div.':np.log(kldistances[j,:,0]),\"MSE\": np.log(msedistances[j,:,0]),\"Realization\":realizations[j,:].astype(int)}\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training model n =  1\n",
      "26214\n",
      "Training model n =  2\n",
      "26214\n",
      "Training model n =  3\n",
      "26214\n",
      "Training model n =  4\n",
      "26214\n",
      "Training model n =  5\n",
      "26214\n",
      "Training model n =  6\n",
      "26214\n",
      "Training model n =  7\n",
      "26214\n",
      "Training model n =  8\n",
      "26214\n",
      "Training model n =  9\n",
      "26214\n",
      "MSE:  0.0007239543024094051 0.0010079735091241355\n",
      "KL:  0.22046197529943262 0.22319099938418066\n",
      "Wasserstein:  0.012811399428271376 0.013114996302987281\n"
     ]
    }
   ],
   "source": [
    "n_rep = 10\n",
    "n_test = len(testloader.dataset)\n",
    "\n",
    "wdistances = np.zeros((n_rep,n_test,2))\n",
    "kldistances = np.zeros((n_rep,n_test,2))\n",
    "msedistances = np.zeros((n_rep,n_test,2))\n",
    "\n",
    "wdistrep = np.zeros((n_rep,2))\n",
    "kldistrep = np.zeros((n_rep,2))\n",
    "msedistrep = np.zeros((n_rep,2))\n",
    "\n",
    "realizations = np.ones((n_rep,n_test))\n",
    "\n",
    "for j in range(0,n_rep):\n",
    "    print(\"Training model n = \",str(j))\n",
    "\n",
    "    nlatent = 3\n",
    "    sc = True\n",
    "    \n",
    "    model = DecoderBinFromMoments(n_bins=35,n_moms=4)\n",
    "\n",
    "\n",
    "    # Check if there is a gpu\n",
    "    device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\n",
    "    #print(device)\n",
    "    #print(model)\n",
    "    model = model.to(device)\n",
    "\n",
    "    mod_weights = \"MomentDistsModels/OneCatMomentDecoder_CNN_n{}_mse_{}epochs_sc\".format(j,epochs)\n",
    "    model.load_state_dict(torch.load(mod_weights))\n",
    "\n",
    "    testM,testdMdt,testrealbin0,testpredbin0=inference(model,testloader,device,nlatent,singlecat=sc)\n",
    "    # ss = calcwasserstein(testrealbin0[:,0,:],testpredbin0[:,0,:])\n",
    "    # print(ss.shape)\n",
    "    wdistances[j,:,0],kldistances[j,:,0],msedistances[j,:,0]=calcwasserstein(testrealbin0[:,0,:],testpredbin0[:,0,:])\n",
    "    wdistances[j,:,1],kldistances[j,:,1],msedistances[j,:,1]=calcwasserstein(testrealbin0[:,1,:],testpredbin0[:,1,:])\n",
    "    \n",
    "    wdistrep[j,0] = np.mean(wdistances[j,:,0])\n",
    "    wdistrep[j,1] = np.std(wdistances[j,:,1])\n",
    "    kldistrep[j,0] = np.mean(kldistances[j,:,0])\n",
    "    kldistrep[j,1] = np.mean(kldistances[j,:,1])\n",
    "    msedistrep[j,0] = np.mean(msedistances[j,:,0])\n",
    "    msedistrep[j,1] = np.mean(msedistances[j,:,1])\n",
    "    realizations[j,:]=np.ones(n_test)*j\n",
    "    \n",
    "    d = {r\"Wasserstein Dist. ($l_{1}$)\":np.log(wdistances[j,:,0]),'KL-Div.':np.log(kldistances[j,:,0]),\"MSE\": np.log(msedistances[j,:,0]),\"Realization\":realizations[j,:].astype(int)}\n",
    "    df = pd.DataFrame(data=d)\n",
    "    \n",
    "    if j==0:\n",
    "        onecatdf = df\n",
    "    else:\n",
    "        onecatdf = pd.concat([onecatdf,df])\n",
    "    \n",
    "print(\"MSE: \", np.mean(msedistances[j,:,0].flatten()),np.std(msedistances[j,:,0].flatten()))\n",
    "print(\"KL: \", np.mean(kldistances[j,:,0].flatten()),np.std(kldistances[j,:,0].flatten()))\n",
    "print(\"Wasserstein: \", np.mean(wdistances[j,:,0].flatten()),np.std(wdistances[j,:,0].flatten()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "d192a76a-ae90-4d1c-9941-cee584613296",
   "metadata": {},
   "outputs": [],
   "source": [
    "np.save(\"recmetrics/onecatwasserstein\",wdistrep)\n",
    "np.save(\"recmetrics/onecatkl\",kldistrep)\n",
    "np.save(\"recmetrics/onecatmse\",msedistrep)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "ef188932-85bc-4a8c-b8fa-59f103442eaf",
   "metadata": {},
   "outputs": [],
   "source": [
    "onecatdf.to_csv(\"recmetrics/onecatdf.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fed4c0ce-80c1-4ac6-9853-884c461ee0f3",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "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": "3.11.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}