code/code/gpuarn.py at master · rmlopes/code · GitHub

Repository navigation

History

100

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

118

119

120

121

122

123

124

125

126

127

128

129

130

131

132

133

134

135

136

137

138

139

140

141

142

143

144

145

146

147

148

149

150

151

152

153

154

155

156

157

158

159

160

161

162

163

164

165

166

167

168

169

170

171

172

173

174

175

176

177

178

179

180

181

182

183

184

185

186

187

188

189

190

191

192

193

194

195

196

197

198

199

200

201

202

203

204

205

206

207

208

209

210

211

212

213

214

215

216

217

218

219

220

221

222

223

224

225

226

227

228

229

230

231

232

233

234

235

236

237

238

239

240

241

242

243

244

245

246

247

248

249

250

251

252

253

254

255

256

257

258

259

260

261

262

263

264

265

266

267

268

269

270

271

272

273

274

275

276

277

278

279

280

281

282

283

284

285

286

287

288

289

290

291

292

293

294

295

296

297

298

299

300

301

302

303

304

305

import ConfigParser, os

import logging

import random

import numpy as np

from time import clock

from numpy import array as nparray

from functools import partial

from bitstring import *

from math import exp

import sys

from sys import stdout

from subprocess import call

from utils import *

from utils.bitstrutils import *

from arn import *

import gpukernel

from pycuda import gpuarray, driver as drv

try:

import matplotlib.pyplot as plt

from matplotlib import collections, legend

except ImportError:

logging.warning('matplotlib not found.')

logging.warning('use of silentmode = 0 will result in an error...')

log = logging.getLogger(__name__)

TFACTORS = 0

STRUCTS = 1

def iterate(arnet,samplerate, simtime, silentmode, simstep,delta,**kwargs):

s = clock()

time = 0

nump = arnet.numtf

numsamples = 1

numrec = arnet.numrec

numeff = arnet.numeff

cuweights_e = arnet.gpu_eweights

cuweights_i = arnet.gpu_iweights

#print kwargs['inputs']

#cuccs = gpuarray.to_gpu(arnet.ccs)

while time < simtime:

#for i in range(numrec):

# arnet.ccs[nump+i] = kwargs['inputs'][i]

_update(arnet.proteins,arnet.ccs,cuweights_e,cuweights_i,delta, **kwargs)

#normalize ccs, ignoring receptors

#print arnet.ccs[:nump]

#cuccs.get(arnet.ccs)

totparcels = (arnet.ccs[:nump].tolist() +

arnet.ccs[nump+numrec:].tolist())

arnet.ccs /= sum(totparcels)

#cuccs.get(arnet.ccs)

#enforce again input/receptor cc

#for i in range(numrec):

# arnet.ccs[nump+i] = kwargs['inputs'][i]

if time % int(simtime*samplerate) == 0:

log.debug('TIME: '+ str(time))

arnet.cchistory = np.column_stack((arnet.cchistory,

arnet.ccs[:nump]))

arnet.receptorhist = np.column_stack((arnet.receptorhist,

arnet.ccs[nump:nump+numrec]))

arnet.effectorhist = np.column_stack((arnet.effectorhist,

arnet.ccs[nump+numrec:]))

time+=simstep

print 'Elapsed time: %f sec.' % (clock()-s,)

return arnet

def _update(proteins, cuccs, exciteweights, inhibitweights,delta,**kwargs):

#print ccs.shape

#print cuccs.shape

esignals = _getSignalArrayCUDA(cuccs[:len(exciteweights)],

exciteweights,**kwargs)

isignals = _getSignalArrayCUDA(cuccs[:len(inhibitweights)],

inhibitweights,**kwargs)

deltas = esignals - isignals

#print deltas.shape

#print cuccs.shape

deltas *= np.float32(delta)

#kwargs['esignals'] *= np.float32(delta)

deltas *= cuccs

#kwargs['esignals'] = kwargs['esignals'] * cuccs

#total = sum(ccs)+sum(deltas)

cuccs += deltas

#ccs/=total

def _getSignalArrayCUDA(ccs, weightstable, **kwargs):

#print ccs.shape

#print weightstable.shape[1]

out = gpuarray.to_gpu(np.zeros(weightstable.shape[1], dtype=np.float32))

cuccs = gpuarray.to_gpu(ccs)

#print signals.shape

#fun = gpukernel.getkernel(16,weightstable.shape[1])

kwargs['gpudot'](out , cuccs, weightstable,

np.int32(weightstable.shape[0]),

np.int32(weightstable.shape[1]))

return np.float32(1.0/len(ccs)) * out.get()

#return np.float32(1.0/len(ccs)) *

def _getprotein(idx, code, bind_size, gene_size, protein_size):

signature = BitStream(bin=

applymajority(code[bind_size*3:bind_size*3+gene_size],

protein_size))

#EXTENDED version - Weak linkage (needs double size gene/proteins)

#p = [code[:self.bind_size],

# code[self.bind_size:self.bind_size*2],

# signature[0:self.bind_size],

# signature[self.bind_size:self.protein_size]]

#ORIGINAL version

p = [idx,

code[:bind_size],

code[bind_size:bind_size*2],

signature,

signature]

log.debug(p)

return p

def _getweights(bindings, bindingsize, beta, **kwargs):

weights = bindings - bindingsize

weights *= beta

return np.exp(weights)

#organized in columns for the target equation

def getbindings(bindtype, proteins, match_threshold,**kwargs):

return nparray([[XORmatching(p[3],otherps[1+bindtype],match_threshold)

for otherps in proteins]

for p in proteins],dtype=np.float32)

class ARNetwork:

def __init__(self, gcode, config, **kwargs):

self.code = gcode

self.simtime = config.getint('default','simtime')

promfun = bindparams(config, buildpromlist)

productsfun = bindparams(config, buildproducts)

self.promlist = promfun(gcode)

self.proteins = productsfun( gcode, self.promlist)

self.effectors=[]

self.effectorproms = promfun(gcode, promoter='00000000')

if self.effectorproms:

#print 'EFFECTORS:', self.effectorproms

self.effectors = productsfun(gcode,self.effectorproms)

self.receptors=[]

self.receptorproms = promfun(gcode, promoter='11111111')

if self.receptorproms:

#print 'RECEPTORS:', self.receptorproms

self.receptors = productsfun(gcode,self.receptorproms)

pbindfun = bindparams(config, getbindings)

weightsfun = bindparams(config, _getweights)

prob = kwargs['problem']

self.numtf = len(self.proteins)

self.numeff = min(len(self.effectors),prob.nout)

self.effectors = self.effectors[:self.numeff]

self.numrec = min(len(self.receptors),prob.ninp)

self.receptors = self.receptors[:self.numrec]

self.ccs = []

if self.promlist:

self.ccs = nparray([1.0/(self.numtf+self.numeff+self.numrec)]*

(self.numtf+self.numeff+self.numrec),

dtype=np.float32)

self._initializehistory()

self._initializebindings(pbindfun)

self._initializeweights(weightsfun)

self.dot_ = gpukernel.getkernel(22,self.eweights.shape[1])

self.esignals = gpuarray.to_gpu(np.zeros(self.eweights.shape[1],

dtype=np.float32))

self.isignals = gpuarray.to_gpu(np.zeros(self.iweights.shape[1],

dtype=np.float32))

for i in range(len(self.proteins)):

self.proteins[i].append(self.ccs[i])

self.simfun = bindparams(config,iterate)

self.delta = config.getfloat('default','delta')

def _initializebindings(self, pbindfun):

self.ebindings = pbindfun(0, self.proteins + self.receptors +

self.effectors)

self.ibindings = pbindfun(1, self.proteins + self.receptors +

self.effectors)

#effectors and dummy receptors do not regulate

if self.effectors or self.receptors:

self.ebindings = self.ebindings[:self.numtf+self.numrec,:]

#print 'ebinds: ', self.ebindings.shape

self.ibindings = self.ibindings[:self.numtf+self.numrec,:]

#print 'ibinds: ', self.ibindings.shape

def _initializeweights(self, weightsfun):

self.eweights = weightsfun(self.ebindings)

#print 'ebinds: ', self.eweights.shape

self.iweights = weightsfun(self.ibindings)

#print 'ebinds: ', self.iweights.shape

self.gpu_eweights = gpuarray.to_gpu(self.eweights)

self.gpu_iweights = gpuarray.to_gpu(self.iweights)

def _initializehistory(self):

self.cchistory=nparray(self.ccs[:self.numtf])

self.receptorhist=nparray(self.ccs[self.numtf:self.numtf+self.numrec])

self.effectorhist=nparray(self.ccs[self.numtf+self.numrec:])

def reset(self):

self._initializehistory()

self.ccs = nparray([1.0/(self.numtf+self.numeff+self.numrec)]*

(self.numtf+self.numeff+self.numrec))

def __str__(self):

return str(self.proteins)

def simulate(self, *inputs):

if not inputs:

inputs = [.0]*self.numrec

#print self.numtf

#print self.numrec

#print str(len(self.ccs) - (self.numtf+self.numrec))

if self.simtime > 0:

self.simfun(self,inputs = inputs,

gpudot = self.dot_,

esignals = self.esignals,

isignals = self.isignals)

for i in range(self.numtf):

self.proteins[i][-1] = self.ccs[i]

#for i in range(len(self.effectors)):

# self.effectors[i][-1] = self.ccs[i+self.numtf+self.numrec]

def stepsimulate(self, proteins, ccs):

_updatenonorm(proteins, ccs, self.eweights, self.iweights, self.delta)

return ccs

def nstepsim(self, n = 1000, *inputs):

self.simfun(self, simtime = n, numeffectors = self.numeff,

inputs = inputs)

for i in range(len(self.proteins)):

self.proteins[i][-1] = self.ccs[i]

################################################################################

### Test ########

################################################################################

if __name__ == '__main__':

_usecuda_ = True

class Problem:

pass

p = Problem()

p.ninp=3

p.nout = 1

arnconfigfile = '../configfiles/arnsimlong.cfg'

saveto = 'genome.save'

log.setLevel(logging.DEBUG)

cfg = ConfigParser.ConfigParser()

cfg.readfp(open(arnconfigfile))

proteins=[]

nump = 0

prob_inp=[.0,.0,.0]

try:

f = open(sys.argv[1], 'r')

genome = BitStream(bin=f.readline())

arnet = ARNetwork(genome, cfg, problem = p)

saveto = sys.argv[1]

except:

while nump < 200 or nump > 512 or numeff == 0 or not arnet.receptors:

genome = BitStream(float=random.random(), length=32)

for i in range(cfg.getint('default','initdm')):

genome = dm_event(genome,

.02)

arnet = ARNetwork(genome, cfg, problem = p)

nump = len(arnet.promlist)

print nump

numeff = len(arnet.effectors)

#f = open(saveto,'w')

#f.write(genome.bin)

#f.close

#exit(0)

#s = clock()

arnet.simulate()

#print clock() - s

if not cfg.getint('default', 'silentmode'):

displayARNresults(arnet.proteins, arnet.cchistory,

cfg.getfloat( 'default','samplerate'), temp=0)

extralabels = ['R']*arnet.numrec + ['E']*arnet.numeff

struct_prots = arnet.receptors + arnet.effectors

hist = np.vstack((arnet.receptorhist,arnet.effectorhist))

displayARNresults(struct_prots, hist,

cfg.getfloat( 'default','samplerate'),

temp = 1, extralabels = extralabels)

for p in arnet.proteins: print p

print 'effectors: ', arnet.effectors

f = open(saveto,'w')

f.write(genome.bin)

f.close

#print genome.bin

code/code/gpuarn.py at master · rmlopes/code · GitHub

Repository navigation

Expand file tree

Breadcrumbs

gpuarn.py

Latest commit

History

Breadcrumbs

gpuarn.py

File metadata and controls