Plotting: adding first Fig4 panels.

1f5331db · Lars Hubatsch · fce1407c · 1f5331db
Commit 1f5331db authored 4 years ago by Lars Hubatsch
--- a/Plots_Droplet_FRAP.ipynb
+++ b/Plots_Droplet_FRAP.ipynb
@@ -24,7 +24,8 @@
    "sns.set_style(\"ticks\")\n",
    "rcParams['axes.linewidth'] = 0.75\n",
    "rcParams['xtick.major.width'] = 0.75\n",
-    "rcParams['ytick.major.width'] = 0.75"
+    "rcParams['ytick.major.width'] = 0.75\n",
+    "rcParams['text.usetex']=True"
   ]
  },
  {
@@ -39,7 +40,8 @@
    "pl.rcParams.update(params)\n",
    "\n",
    "def nice_fig(xla, yla, xli, yli, size, fs=12): \n",
-    "    rc('font',**{'family':'sans-serif','sans-serif':['Helvetica']})\n",
+    "#     rc('font',**{'family':'sans-serif','sans-serif':['Helvetica']})\n",
+    "    rc('font',**{'family':'serif','serif':['Palatino']})\n",
    "    plt.gcf().set_size_inches(size[0], size[1])\n",
    "    plt.xlabel(xla,fontsize=fs)  \n",
    "    plt.ylabel(yla,fontsize=fs)\n",
@@ -296,7 +298,7 @@
   "cell_type": "markdown",
   "metadata": {},
   "source": [
-    "## Figure 1:"
+    "### Figure 1: Fitting $D_{in}$ and data analysis."
   ]
  },
  {
@@ -448,6 +450,71 @@
    "save_nice_fig(fol+'Fig1/tot_recov.pdf')"
   ]
  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Figure 4: Obtaining info about outside."
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "**Panel: Partitioning vs. $D_{out}$, showcasing four different simulation start cases.**"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "P_Do = np.loadtxt(fol+'/Fig4/Part_vs_Do.csv', delimiter=',')\n",
+    "P = [5, 150, 5, 150]\n",
+    "D_o = [0.1, 0.1, 1, 1]\n",
+    "plt.gca().set_prop_cycle(None)\n",
+    "nice_fig('Partitioning P', '$D_{out}$ [$\\mu m^2/s$]', [0.9,320], [0.000001,340], [2.3,2])\n",
+    "lines = plt.loglog(P_Do[0, :], P_Do[1:, :].transpose())\n",
+    "plt.plot(P_Do[0, :], P_Do[0, :], '--', c='grey')\n",
+    "plt.legend([lines[2], lines[0], lines[3], lines[1]],\n",
+    "           ['0.2', '0.02', '0.0067', '0.00067'], ncol=2, frameon=False,\n",
+    "               title=r'\\underline{$D_{out}$/P [$\\mu m^2/s$]:}', columnspacing=0.5, labelspacing=0.3,\n",
+    "           loc=(0.4, 0), handletextpad=0.4, handlelength=0.5)\n",
+    "plt.gca().set_prop_cycle(None)\n",
+    "plt.plot(P[0], D_o[0], 'd')\n",
+    "plt.plot(P[1], D_o[1], 'd')\n",
+    "plt.plot(P[2], D_o[2], 'd')\n",
+    "plt.plot(P[3], D_o[3], 'd')\n",
+    "plt.annotate('$D_{out}$/P = 1 $\\mu m^2/s$', [1,40], c='grey')\n",
+    "plt.xticks([1, 10, 100]);\n",
+    "save_nice_fig(fol+'Fig4/D_vs_P.pdf')"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "**Panel: Cost function**"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "P_Cost = np.loadtxt(fol+'/Fig4/Part_vs_Cost.csv', delimiter=',')\n",
+    "nice_fig('Partitioning P', 'Cost function [a.u.]', [0.9,320], [0.000000001,0.01], [2.3,2])\n",
+    "lines = plt.loglog(P_Cost[0, :], P_Cost[1:, :].transpose())\n",
+    "plt.legend([lines[2], lines[0], lines[3], lines[1]],\n",
+    "           ['0.2', '0.02', '0.0067', '0.00067'], ncol=2, frameon=False,\n",
+    "           title=r'\\underline{$D_{out}$/P set to:}', columnspacing=0.5, labelspacing=0.3,\n",
+    "           loc=(0.081, 0), handletextpad=0.4, handlelength=0.5)\n",
+    "plt.xticks([1, 10, 100]);\n",
+    "save_nice_fig(fol+'Fig4/D_vs_Cost.pdf')"
+   ]
+  },
  {
   "cell_type": "code",
   "execution_count": null,

 %% Cell type:markdown id: tags:
 ### FRAP geometries
 %% Cell type:code id: tags:
 ``` python
 from fem_sol import frap_solver
 from matplotlib import rc, rcParams
 import fem_utils
 import matplotlib.pyplot as plt
 import numpy as np
 import pandas as pd
 import seaborn as sns
 fol = '/Users/hubatsch/Desktop/DropletFRAP/Latex/Figures/'
 sns.set_style("ticks")
 rcParams['axes.linewidth'] = 0.75
 rcParams['xtick.major.width'] = 0.75
 rcParams['ytick.major.width'] = 0.75
+rcParams['text.usetex']=True
 ```
 %% Cell type:code id: tags:
 ``` python
 import pylab as pl
 params = {'legend.fontsize': 9,
          'legend.handlelength': 1}
 pl.rcParams.update(params)
 def nice_fig(xla, yla, xli, yli, size, fs=12):
-    rc('font',**{'family':'sans-serif','sans-serif':['Helvetica']})
+#     rc('font',**{'family':'sans-serif','sans-serif':['Helvetica']})
+    rc('font',**{'family':'serif','serif':['Palatino']})
    plt.gcf().set_size_inches(size[0], size[1])
    plt.xlabel(xla,fontsize=fs)
    plt.ylabel(yla,fontsize=fs)
    plt.xlim(xli)
    plt.ylim(yli)
    plt.tick_params(axis='both', which='major', labelsize=fs)
 def save_nice_fig(name):
    plt.savefig(name, format='pdf', dpi=300, bbox_inches='tight',
                transparent=True)
 ```
 %% Cell type:code id: tags:
 ``` python
 me = ['Meshes/multi_drop_gauss.xml', 'Meshes/multi_drop_gauss_med.xml',
     'Meshes/multi_drop_gauss_far.xml', 'Meshes/multi_drop_gauss.xml',
     'Meshes/multi_drop_gauss_med.xml', 'Meshes/multi_drop_gauss_far.xml']
 point_lists = [[[4, 4.5, 0.5], [4, 3.5, 0.5], [3.5, 4, 0.5], [4.5, 4, 0.5]],
               [[4, 5, 0.5], [4, 3, 0.5], [3, 4, 0.5], [5, 4, 0.5]],
               [[4, 5.5, 0.5], [4, 2.5, 0.5], [2.5, 4, 0.5], [5.5, 4, 0.5]],
               [[4, 4.5, 0.5], [4, 3.5, 0.5], [3.5, 4, 0.5], [4.5, 4, 0.5]],
               [[4, 5, 0.5], [4, 3, 0.5], [3, 4, 0.5], [5, 4, 0.5]],
               [[4, 5.5, 0.5], [4, 2.5, 0.5], [2.5, 4, 0.5], [5.5, 4, 0.5]]]
 phi_tot_int = [.99, .99, .99, .9, .9, .9]
 phi_tot_ext = [.01, .01, .01, .1, .1, .1]
 G_in = [1, 1, 1, .1, .1, .1]
 G_out = [1, 1, 1, 0.99/0.9, 0.99/0.9, 0.99/0.9]
 f_i = []
 for p, m, p_i, p_e, G_i, G_o in zip(point_lists, me, phi_tot_int,
                                    phi_tot_ext, G_in, G_out):
    f = frap_solver([4, 4, 0.5], m, name='FRAP_multi_'+m[:-4]+str(G_i), point_list=p,
                    T=50, phi_tot_int=p_i, phi_tot_ext=p_e, G_in=G_i, G_out=G_o)
    f.solve_frap()
    f_i.append(f)
 ```
 %% Cell type:code id: tags:
 ``` python
 alphas = np.linspace(0,2*np.pi, 20)
 ns = np.c_[np.cos(alphas), np.sin(alphas), np.zeros(len(alphas))]
 eps = np.linspace(0, 0.23, 100)
 profs = []
 for i in range(len(f_i)):
 #     if i>2:
        profs.append([])
        for j in range(50):
            values=[]
            fs = fem_utils.load_time_point(f_i[i].name+'t_p_'+str(j)+'.h5',
                                           f_i[i].mesh)
            for n in ns:
                values.append([fs([4, 4, 0.5]+e*n) for e in eps])
            profs[i].append(np.mean(np.transpose(values), 1))
 ```
 %% Cell type:code id: tags:
 ``` python
 np.savetxt('t_p.csv', profs, delimiter=',')
 ```
 %% Cell type:code id: tags:
 ``` python
 ft = f_i[1]
 meta_data = np.r_[ft.dt, ft.T, eps]
 np.savetxt('meta_data.csv', meta_data, delimiter=',')
 ```
 %% Cell type:code id: tags:
 ``` python
 plt.plot(eps,np.transpose(profs)[:,:])
 ```
 %% Cell type:code id: tags:
 ``` python
 nice_fig('t [s]', 'intensity (a.u)', [0,50], [0,1.1], [1.5,2])
 plt.plot([np.mean(x)/f_i[0].phi_tot_int for x in profs[0]],
         lw=2, label='d=0.5', ls='-')
 plt.plot([np.mean(x)/f_i[1].phi_tot_int for x in profs[1]],
         lw=2, label='d=1', ls='--')
 plt.plot([np.mean(x)/f_i[2].phi_tot_int for x in profs[2]],
         lw=2, label='d=1.5', ls=':')
 plt.plot(range(0, 100), np.ones(100), linestyle='--', color='k')
 plt.title('$\Phi_{out}=0.01}$', size=12)
 plt.gca().get_yaxis().set_visible(False)
 save_nice_fig(fol+'Fig3/tot_recov_neighbours_bad.pdf')
 ```
 %% Cell type:code id: tags:
 ``` python
 nice_fig('t [s]', 'intensity (a.u)', [0,30], [0,1.1], [1.5,2])
 plt.plot([np.mean(x)/f_i[3].phi_tot_int for x in profs[3]],
         lw=2, label='d=0.5', ls='-')
 plt.plot([np.mean(x)/f_i[4].phi_tot_int for x in profs[4]],
         lw=2, label='d=1', ls='--')
 plt.plot([np.mean(x)/f_i[5].phi_tot_int for x in profs[5]],
         lw=2, label='d=1.5', ls=':')
 plt.plot(range(0, 100), np.ones(100), linestyle='--', color='k')
 plt.title('$\Phi_{out}=0.1}$', size=12)
 plt.legend(prop={'size': 9}, frameon=False)
 save_nice_fig(fol+'Fig3/tot_recov_neighbours_good.pdf')
 ```
 %% Cell type:code id: tags:
 ``` python
 nice_fig('x [$\mu m$]', 'intensity (a.u)', [0,0.25], [0,1.1], [3.8,2])
 l_sim = plt.plot(eps, np.transpose(profs[0])[:,::8]/f_i[0].phi_tot_int, '#1f77b4', lw=2.5)
 plt.plot(range(0, 100), np.ones(100), linestyle='--', color='k')
 l_fit = plt.plot(np.linspace(0, 0.23, 100), np.transpose(ml)[:,::8],
                 ls='--', c='orange', lw=1.5)
 plt.legend([l_sim[0], l_fit[0]], ['Simulation', 'Fit'], prop={'size': 9}, frameon=False)
 save_nice_fig(fol+'Fig3/spat_recov_neighbours.pdf')
 ```
 %% Cell type:code id: tags:
 ``` python
 # Define parameters for all simulations
 point_list = [[4, 4, 0.5], [4, 4, 1.5], [4, 4, 4],
              [4, 4, 0.5], [4, 4, 1.5], [4, 4, 4]]
 me = ['coverslip.xml', '1_5.xml', 'symmetric.xml',
      'coverslip.xml', '1_5.xml', 'symmetric.xml']
 phi_tot_int = [.99, .99, .99, .9, .9, .9]
 phi_tot_ext = [.01, .01, .01, .1, .1, .1]
 G_in = [1, 1, 1, .1, .1, .1]
 G_out = [1, 1, 1, 0.99/0.9, 0.99/0.9, 0.99/0.9]
 f_cs = []
 # Zip all parameters, iterate
 for p, m, p_i, p_e, G_i, G_o in zip(point_list, me, phi_tot_int,
                                   phi_tot_ext, G_in, G_out):
    f_cs.append(frap_solver(p, 'Meshes/single_drop_'+m,
                name='FRAP_'+m[:-4]+str(G_i), T=60, phi_tot_int=p_i,
                phi_tot_ext=p_e, G_in=G_i, G_out=G_o))
    f_cs[-1].solve_frap()
 ```
 %% Cell type:code id: tags:
 ``` python
 z = [0.5, 1.5, 4, 0.5, 1.5, 4]
 profs_cs = []
 for i, z_i in enumerate(z):
    profs_cs.append([])
    for j in range(50):
        values=[]
        fs = fem_utils.load_time_point(f_cs[i].name+'t_p_'+str(j)+'.h5',
                                       f_cs[i].mesh)
        for n in ns:
            values.append([fs([4, 4, z_i]+e*n) for e in eps])
        profs_cs[i].append(np.mean(np.transpose(values), 1))
 ```
 %% Cell type:code id: tags:
 ``` python
 np.savetxt('t_p_neighbours.csv', profs_cs[0], delimiter=',')
 ```
 %% Cell type:code id: tags:
 ``` python
 nice_fig('t [s]', '', [0,50], [0,1.1], [1.5,2])
 ls = ['-', '--', ':']
 for i, f in enumerate(f_cs[0:3]):
    plt.plot([np.mean(x)/f.phi_tot_int for x in profs_cs[i]],
             label='d='+str(z[i]), ls=ls[i], lw=2)
 plt.plot(range(0, 100), np.ones(100), linestyle='--', color='k')
 plt.title('$\Phi_{out}=0.01}$', size=12)
 plt.gca().get_yaxis().set_visible(False)
 save_nice_fig(fol+'Fig3/tot_recov_cs_bad.pdf')
 ```
 %% Cell type:code id: tags:
 ``` python
 nice_fig('t [s]', 'intensity (a.u)', [0,30], [0,1.1], [1.5,2])
 ls = ['-', '--', ':']
 for i, f in enumerate(f_cs[3:]):
    plt.plot([np.mean(x)/f.phi_tot_int for x in profs_cs[i+3]],
             label='h='+str(z[i]), lw=2, ls=ls[i])
 plt.plot(range(0, 100), np.ones(100), linestyle='--', color='k')
 plt.title('$\Phi_{out}=0.1}$', size=12)
 plt.legend(prop={'size': 9}, frameon=False)
 save_nice_fig(fol+'Fig3/tot_recov_cs_good.pdf')
 ```
 %% Cell type:code id: tags:
 ``` python
 ml_neigh = np.loadtxt('/Users/hubatsch/Desktop/DropletFRAP/matlab_fit_neigh.csv',
                      delimiter=',')
 nice_fig('x [$\mu m$]', 'intensity (a.u)', [0,0.25], [0,1.1], [3.8,2])
 l_sim = plt.plot(eps, np.transpose(profs_cs[0])[:,::8]/f_cs[0].phi_tot_int, '#1f77b4',
                 lw=2.5, label='Simulation')
 plt.plot(range(0, 100), np.ones(100), linestyle='--', color='k')
 l_fit = plt.plot(np.linspace(0, 0.23, 100), np.transpose(ml_neigh)[:,::8],
         ls='--', c='orange', lw=1.5)
 plt.legend([l_sim[0], l_fit[0]], ['Simulation', 'Fit'], frameon=False)
 save_nice_fig(fol+'Fig3/spat_recov_coverslip.pdf')
 ```
 %% Cell type:markdown id: tags:
-## Figure 1:
+### Figure 1: Fitting $D_{in}$ and data analysis.
 %% Cell type:markdown id: tags:
 **Panel: comparison PGL-3 diffusivity with Louise's viscosity**
 %% Cell type:code id: tags:
 ``` python
 louise = pd.read_csv('/Users/hubatsch/Desktop/DropletFRAP/Latex/Figures/Fig1/Louise.csv')
 lars = pd.read_csv('/Users/hubatsch/Desktop/DropletFRAP/Latex/Figures/Fig1/Lars.csv')
 ```
 %% Cell type:code id: tags:
 ``` python
 fig, ax1 = plt.subplots()
 ax2 = ax1.twinx()
 plt.sca(ax1)
 sns.lineplot(x="conc", y="D", data=lars, color=sns.color_palette()[1])
 sns.scatterplot(x="conc", y="D", data=lars, color=sns.color_palette()[1], alpha=0.7)
 plt.xlabel('$c_{salt}\; [mM]$')
 plt.ylabel('$D_{in} \;[\mu m^2\cdot s^{-1}]$', color=sns.color_palette()[1])
 plt.yticks([0, 0.05, 0.1], rotation=90, color = sns.color_palette()[1])
 plt.ylim(0, 0.1)
 ax1.set_zorder(1)
 ax1.patch.set_visible(False)
 plt.sca(ax2)
 sns.lineplot(x="conc", y="vis", data=louise, color=sns.color_palette()[0], label='data from ref[xxx]')
 nice_fig('c_{salt} [mM]', '$\eta^{-1} \;[Pa\cdot s]^{-1}$', [40,190], [0,7.24], [2.3,2])
 plt.yticks(color = sns.color_palette()[0])
 plt.ylabel('$\eta^{-1} \;[Pa\cdot s]^{-1}$ ', color = sns.color_palette()[0])
 plt.legend(frameon=False, fontsize=9)
 save_nice_fig(fol+'Fig1/Lars_vs_Louise.pdf')
 ```
 %% Cell type:markdown id: tags:
 **Panel:coacervates PLYS/ATP, CMD/PLYS**m
 %% Cell type:code id: tags:
 ``` python
 coacervates = pd.read_csv('/Users/hubatsch/Desktop/DropletFRAP/Latex/Figures/Fig1/Coacervates.csv')
 sns.stripplot(data=coacervates, jitter=0.35, alpha=0.8,**{'marker': '.', 'size': 8, 'edgecolor': 'black', 'color': 'k'})
 ax = sns.barplot(data=coacervates, facecolor=(1, 1, 1, 0), edgecolor=(0.6, 0.6, 0.6), errcolor=(0.6, 0.6, 0.6), capsize=.2, ci='sd', errwidth=1.5)
 plt.setp(ax.lines, zorder=100)
 nice_fig(None, '$D_{in} \;[\mu m^2\cdot s^{-1}]$', [None, None], [0,6], [2.3,2])
 plt.xticks([0,1], ('CMD/PLYS', 'PLYS/ATP'), rotation=20)
 plt.xlim(-0.7, 1.7)
 save_nice_fig(fol+'Fig1/Coacervates.pdf')
 ```
 %% Cell type:code id: tags:
 ``` python
 ax.patches
 ```
 %% Cell type:markdown id: tags:
 **Panel: time course CMD**
 %% Cell type:code id: tags:
 ``` python
 CMD = np.loadtxt(fol+'/Fig1/CMD_timecourse.csv', delimiter=',')
 CMD_fit = np.loadtxt(fol+'/Fig1/CMD_fit_timecourse.csv', delimiter=',')
 l_sim = plt.plot(CMD[:, 0], CMD[:, 1:], '#1f77b4', lw=3, label='Simulation')
 l_fit = plt.plot(CMD_fit[:, 0], CMD_fit[:, 1:], '--', lw=2, c=sns.color_palette()[1], label='Simulation')
 plt.plot(range(0, 10), np.ones(10)*np.min(CMD_fit[:, 1]), linestyle='--', color='k', lw=2)
 nice_fig('x [$\mu m$]', 'intensity (a.u)', [0,np.max(CMD_fit[:, 0])], [0,0.6], [2.3,2])
 save_nice_fig(fol+'Fig1/CMD_spat_recov.pdf')
 ```
 %% Cell type:markdown id: tags:
 **Panel: time course PGL-3**
 %% Cell type:code id: tags:
 ``` python
 PGL = np.loadtxt(fol+'/Fig1/PGL_timecourse.csv', delimiter=',')
 PGL_fit = np.loadtxt(fol+'/Fig1/PGL_fit_timecourse.csv', delimiter=',')
 l_sim = plt.plot(PGL[:, 0], PGL[:, 1:], '#1f77b4', lw=3, label='Simulation')
 l_fit = plt.plot(PGL_fit[:, 0], PGL_fit[:, 1:], '--', lw=2, c=sns.color_palette()[1], label='Simulation')
 plt.plot(range(0, 10), np.ones(10)*np.min(PGL_fit[:, 1]), linestyle='--', color='k', lw=2)
 nice_fig('x [$\mu m$]', 'intensity (a.u)', [0,np.max(PGL_fit[:, 0])], [0,None], [2.3,2])
 save_nice_fig(fol+'Fig1/PGL_spat_recov.pdf')
 ```
 %% Cell type:markdown id: tags:
 **Panel: time course total intensity**
 %% Cell type:code id: tags:
 ``` python
 PGL = np.loadtxt(fol+'/Fig1/PGL_tot.csv', delimiter=',')
 ATP = np.loadtxt(fol+'/Fig1/ATP_tot.csv', delimiter=',')
 CMD = np.loadtxt(fol+'/Fig1/CMD_tot.csv', delimiter=',')
 # fig, ax1 = plt.subplots()
 # ax2 = ax1.twiny()
 # plt.sca(ax1)
 nice_fig('$t/T_{max}$', 'intensity (a.u)', [0,200], [0,0.62], [2.3,2])
 # plt.sca(ax2)
 # ax2.tick_params(axis="x",direction="in")
 plt.plot(PGL[::10, 0]/np.max(PGL[1:-1:2, 0]), PGL[::10,1], label='PGL-3', c='#CC406E', markersize=3, alpha=0.7, lw=2)
 plt.plot(ATP[::1, 0]/np.max(ATP[:, 0]), ATP[::1,1], label='PLYS/ATP', c='#FF508A', markersize=3, alpha=0.7, lw=2)
 plt.plot(CMD[::5, 0]/np.max(CMD[:, 0]), CMD[::5,1], label='CMD/PLYS', c='#7F2845', markersize=3, alpha=0.7, lw=2)
 plt.legend(frameon=False, fontsize=9)
 plt.xlim(0, 1)
 save_nice_fig(fol+'Fig1/tot_recov.pdf')
 ```
+%% Cell type:markdown id: tags:
+### Figure 4: Obtaining info about outside.
+%% Cell type:markdown id: tags:
+**Panel: Partitioning vs. $D_{out}$, showcasing four different simulation start cases.**
+%% Cell type:code id: tags:
+``` python
+P_Do = np.loadtxt(fol+'/Fig4/Part_vs_Do.csv', delimiter=',')
+P = [5, 150, 5, 150]
+D_o = [0.1, 0.1, 1, 1]
+plt.gca().set_prop_cycle(None)
+nice_fig('Partitioning P', '$D_{out}$ [$\mu m^2/s$]', [0.9,320], [0.000001,340], [2.3,2])
+lines = plt.loglog(P_Do[0, :], P_Do[1:, :].transpose())
+plt.plot(P_Do[0, :], P_Do[0, :], '--', c='grey')
+plt.legend([lines[2], lines[0], lines[3], lines[1]],
+           ['0.2', '0.02', '0.0067', '0.00067'], ncol=2, frameon=False,
+               title=r'\underline{$D_{out}$/P [$\mu m^2/s$]:}', columnspacing=0.5, labelspacing=0.3,
+           loc=(0.4, 0), handletextpad=0.4, handlelength=0.5)
+plt.gca().set_prop_cycle(None)
+plt.plot(P[0], D_o[0], 'd')
+plt.plot(P[1], D_o[1], 'd')
+plt.plot(P[2], D_o[2], 'd')
+plt.plot(P[3], D_o[3], 'd')
+plt.annotate('$D_{out}$/P = 1 $\mu m^2/s$', [1,40], c='grey')
+plt.xticks([1, 10, 100]);
+save_nice_fig(fol+'Fig4/D_vs_P.pdf')
+```
+%% Cell type:markdown id: tags:
+**Panel: Cost function**
+%% Cell type:code id: tags:
+``` python
+P_Cost = np.loadtxt(fol+'/Fig4/Part_vs_Cost.csv', delimiter=',')
+nice_fig('Partitioning P', 'Cost function [a.u.]', [0.9,320], [0.000000001,0.01], [2.3,2])
+lines = plt.loglog(P_Cost[0, :], P_Cost[1:, :].transpose())
+plt.legend([lines[2], lines[0], lines[3], lines[1]],
+           ['0.2', '0.02', '0.0067', '0.00067'], ncol=2, frameon=False,
+           title=r'\underline{$D_{out}$/P set to:}', columnspacing=0.5, labelspacing=0.3,
+           loc=(0.081, 0), handletextpad=0.4, handlelength=0.5)
+plt.xticks([1, 10, 100]);
+save_nice_fig(fol+'Fig4/D_vs_Cost.pdf')
+```
 %% Cell type:code id: tags:
 ``` python
 ```