result of erf is never more precise than double

3 results were exact to double the precision

20 largest ulp errors (stored in a double)
   1.98487 ulp for erf(0.1112354518409497) = 0.12499999999999974)
   1.50491 ulp for erf(0.58157230595001552) = 0.58918909104436068)
   1.50203 ulp for erf(-0.25361458087937655) = -0.28015442007041214)
   1.50007 ulp for erf(-0.22061468666489414) = -0.24495665101412453)
   1.50000 ulp for erf(-0.11950033357811427) = -0.13420256609422709)
   1.49999 ulp for erf(-0.25361458087806538) = -0.28015442006902463)
   1.49996 ulp for erf(0.068886556292221221) = 0.07760737738994096)
   1.49757 ulp for erf(-0.11950033357842287) = -0.13420256609457029)
   1.49524 ulp for erf(-0.23798923314695503) = -0.26355711689075012)
   1.09535 ulp for erf(0.11123545184094996) = 0.12499999999999997)
   1.07619 ulp for erf(0.11123545184094995) = 0.12499999999999997)
   1.00574 ulp for erf(0.47693627620446882) = 0.49999999999999911)
   1.00000 ulp for erf(0.11123545184094509) = 0.12499999999999456)
   0.99964 ulp for erf(0.47693627620446388) = 0.49999999999999467)
   0.82720 ulp for erf(0.47693627620446982) = 0.5)
   0.59433 ulp for erf(0.64444816208681766) = 0.63790867356268988)
   0.56675 ulp for erf(1.0747008388524679) = 0.87145295513009235)
   0.56316 ulp for erf(0.58157230595001896) = 0.58918909104436334)
   0.54795 ulp for erf(-1.2011468739873308) = -0.91062016678759206)
   0.54448 ulp for erf(0.53220088553867761) = 0.54833715828603613)

<tr><td align=left><a href=erf.res>erf</a></td><td align=right>1277</td><td align=right>1.98487</td><td>0.1112354518409497</td><td>0.12499999999999974</td></tr>